Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Virtually human? Negotiation of (non)humanness and agency in the sociotechnical assemblage of virtual influencers
(USC Thesis Other)
Virtually human? Negotiation of (non)humanness and agency in the sociotechnical assemblage of virtual influencers
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
VIRTUALLY HUMAN?
NEGOTIATION OF (NON)HUMANNESS AND AGENCY IN THE SOCIOTECHNICAL
ASSEMBLAGE OF VIRTUAL INFLUENCERS
by
Do Own (Donna) Kim
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(COMMUNICATION)
August 2022
Copyright 2022 Do Own (Donna) Kim
ii
Acknowledgements
My favorite Korean poem is titled flower, kkot [Flower], by Kim Chunsu. The part that
has stayed with me over the years goes:
“Before I called their name, they were nothing but a gesture.
When I called their name, they came to me, and became a flower.”
I believe what my mentors, my colleagues, my friends, and my beloved family have done
for me throughout my doctoral studies is help me find my name and call me by that name. In
everything I write, I say, I think, there is a piece of you. You believed in what could have been a
passing dandelion seed —perhaps some potentials, but not seen. You helped me find my colors,
my scent. And intellectually bloom. USC Annenberg has been the ground that helped me feel
rooted in. Los Angeles has been my home.
I thank my mentors for their wisdom, support, and generosity. You believed in me and
helped me believe in myself. You let me grow. I thank my co-advisors Professor Henry Jenkins
and Professor Dmitri Williams and my dissertation committee member Professor Mike Ananny
for being great role models. You taught me who I want to be as an academic as well as as a
person. The commitment to embody my studies has been your gift that will stay with me. You
keep me vigilant and empathetic. I thank Professor Dong Hoon Ma. You were the first person
who saw something in me before I could ever see myself as an academic. You generously
encouraged and supported my journey into this path. For that, I am forever indebted to your
mentorship. I also thank Professor Hye Won Shin, Professor Margaret McLaughlin, Professor
Carmen Lee, Professor Holly Willis, and many other faculty and mentors for their mentorship
iii
and support along the way. You inspire me to give back the lessons, kindness, and understanding
that you have so generously and patiently given me.
I thank USC Annenberg, Korea Foundation for Advanced Studies (KFAS), USC Korea
Studies Institute (KSI), and USC Graduate School. I feel so lucky and grateful to have had so
many people and organizations that advocated for me, allowing me access to the blissful
privilege of being able to prioritize concentrating on my studies. Sarah Holterman and Anne-
Marie Campian, you are what make USC Annenberg a proud and loving home for so many of us.
Your genuine care for us, your patience, and your unmatched expertise were truly magical. Staff
and committee members at KFAS, especially Yihyun Kim sunsengnim, thank you for catching
and trusting the gleam in me and helping me grow from a wide-eyed undergraduate student to a
Doctor of Philosophy. PhD is by no means an easy journey for many, but your support made it
easier for me. Sarah Shear, Linda Kim, Gloria Koo, and many others who made being part of
KSI a wonderful experience, thank you for welcoming me. In addition, I am grateful for the
institutional financial support that I received, including KFAS doctoral grant, USC Annenberg
Graduate Fellowship & Assistantship, USC Annenberg Summer Research Grant, USC KSI
Summer Fieldwork Grant, USC KSI Graduate Student Affiliate grant, and USC Annenberg
Graduate Fellowship Research and Creative Project Awards.
I thank my Annenberg friends and colleagues. PhDivas, accountability group members,
tree planters, K-town Café friends, and the Annenberg community, my PhD journey would not
have been the same without you. Thanks to you all, I did not merely survive but lived and
thrived. You helped make my todays more bearable, yesterdays more memorable, and
tomorrows more imaginable. I was able to finish my dissertation and grow as a scholar. My days
in LA were rich and beautiful because of you. Thank you, PhDivas: Jeeyun Sophia Baik, Hye
iv
Min Kim, Franny Corry, Soledad Altrudi, Andrea Alarcon, Nathaniel Ming Curran, Sulafa
Zidani, Lauren Sowa, Yiqi Li, Briana Ellerbe, Jingyi Sun, Anna Loup, Liyuan (Leah) Wang, and
Brooklyne Gipson. Thank you, Annenberg friends and colleagues: Lichen Zhen, Sukyoung Choi,
Steffie Kim, Eugene Jang, Eugene Lee, Ho-chun Herbert Chang, Jack Lipei Tang, Yuanfeixue
Nan, Thi Ngoc Bich (Becky) Pham, So Yun Ahn, Junyi Lv, Ed Kang, Rachel Moran, TJ Billard,
Ruthie Kelly, Sonia Shaikh, Jillian Kwong, Ignacio Cruz, Yunwen (Kathy) Wang, Aveva Yusi
Xu, Tyler Quick, Paulina Lanz, Steven Proudfoot, Chris Persaud, Kyooeun Jang, Mingxuan
(Elaine) Liu, Natalie Jonckheere, Calvin Liu, Hyun Tae (Calvin) Kim, Joo-hwa Hong, Mina
Park, Sierra Bay, Maximilian Brichta, and many more. Space limits me from mentioning
everyone here and expressing how special each of you are to me but I thank you for all the
wonderful memories. I am also grateful to Sangita Shresthova, Fred Morstatter, and Amanda
Ford, as well as my research groups Civic Paths, Media as Sociotechnical Systems (MASTS),
and That Game Group.
I thank my research participants for their generous time, interest, and enthusiasm. My
dissertation research would not have been possible without you. Your experiences, insights, and
knowledge were invaluable to this research. Moreover, your passion inspired and motivated me.
You helped me think and write. Thank you for so generously sharing your thoughts, feelings, and
pieces of yourself with me.
I thank my friends and all of those who cheered me on across the globe. You were there
with me and for me. You made me know that my writing and research are worthy and important.
You made me know that I am loved and cared about. Thank you, Eunjung (Kristin) Jung, Silvia
Kim, Julia Park, Emily Yoon, Michelle Sham, Younkyu Lee, and many other unnie, oppa,
sunbae, dongsaeng, and chingus who made LA extra sunny for me. Thank you, Ginam Kim,
v
Edward Sun, Jakob Sprenger, Chaeyoon Yoo, Hayoung Lim, Jaekyung (Chloe) Ahn, Taeyoung
Kim-Lee, and many more good friends that I am too shy to name (you know who you are!) who
continued to remind me who Donna is/can be and continued to invite me to their world,
regardless of where we were. Thank you, Regina Kim for being the best apartment manager
anyone can ever imagine having; you made LA not home away from home but home. I will
remember your yummy homecooked meals (which you coincidentally brought over again as I
was writing this). In the same spirit, I want to thank K-town coffee shops, especially Sharp
Specialty Coffee and Mak, for being not my office away from office but office. Your coffee and
work space kept me happy and productive.
I thank my family. Thank you, Jiah Lim, Eunbyul Lee, Dajung Lee, Sowoon Son, and
Jaeyeon Lee for loving me and letting me love you. My dearest friends, you are my sisters. I do
not doubt my presence because I have you. You anchor me. I know I am extremely lucky
because my life has gifted me with your friendship—a family that is connected by heart. You
keep me strong. You motivate me to give. You make me want to share for the better for all. My
parents Kyung Sook Kang and Hye Joon Kim, and my sister Do Dom Kim, thank you is not
enough for me to express it all. More than anything, it is your unconditional love and support
that kept me going. I love myself and the world because of you. I am who I am because of you.
Thank you. Eat well, be healthy.
Lastly, I thank myself. Thank you, Donna. I am very proud of you.
Perhaps we are all but gestures before being called by our names. Call me Do Own
(Donna) Kim, Ph.D. Whenever you need me, whenever you miss me, I will call you by your
name, too. Thank you.
vi
Table of Contents
Acknowledgements…………………………………………………………………………… ii
List of Tables………………………………………………………………………………… ix
List of Figures………………………………………………………………………………… x
Abstract……………………………………………………………………………………… xii
Introduction…………………………………………………………………………………… 1
The Case: Virtual Influencers………………………………………………………… 3
Guiding Questions and Contributions………………………………………………… 5
Methodology…………………………………………………………………………… 8
Chapter Overview…………………………………………………………………… 16
Chapter 1: “I am the First Virtual”: Narratives of Novelty and Technocapitalistic
Imaginaries of Virtuality……………………………………………………………… 20
Chapter Summary…………………………………………………………………… 20
Narratives of Novelty………………………………………………………………… 21
The Problem of Novelty……………………………………………………… 22
Virtual Influencer History…………………………………………………………… 28
1990s Japan and Korea: Virtual Idols and Cyber Singers…………………… 30
Virtual Bands: Pre-90s Fictional Characters to Gorillaz…………………… 46
2000s: The Rise of Hatsune Miku…………………………………………… 52
2010s to Now: Virtual Influencers and V-Tubers…………………………… 57
Variegated Definitions……………………………………………………………… 61
Traits of virtual influencers…………………………………………………… 63
Virtual “Influencer” and Six Traits………………………………………… 69
Technological Approximations……………………………………………… 71
Sibling V-terms……………………………………………………………… 77
At the Boundary……………………………………………………………… 83
vii
Conclusion…………………………………………………………………………… 84
Chapter 2: “Virtual Woman of Color, 20 Years Old”: Politics of Representation and
Self-presentation……………………………………………………………………… 87
Chapter Summary…………………………………………………………………… 88
(Re-)Presenting Who? ……………………………………………………………… 89
Front/Backstage……………………………………………………………… 89
Presentation as Performance………………………………………………… 95
Cases: Virtual Woman of Color, 19 Years Old……………………………………… 104
Physical versus Virtual……………………………………………………… 104
My Friend Robot…………………………………………………………… 106
Anonymous, Authentic, Appropriating……………………………………… 114
Physical Bodies of a Virtual Body…………………………………………… 119
Hypersexualized……………………………………………………………… 125
Sex/Gender-swapped………………………………………………………… 128
Abandoning the Virtual Body…………………………………………………136
Accountability………………………………………………………………. 141
Woman of Legal Age………………………………………………………… 143
Race/ethnicity and Intersections……………………………………………. 156
Conclusion…………………………………………………………………………… 164
Chapter 3: Behind the Scenes In Real Life: Humans, Things, and Places…………………... 167
Chapter Summary…………………………………………………………………… 167
Networked Humans and Nonhumans……………………………………………… 169
Behind the Scenes…………………………………………………………………… 175
Work Process and Standards………………………………………………… 177
Hidden Humans……………………………………………………… 177
As Told by Tools? …………………………………………………… 188
Silent Standards……………………………………………………… 194
Body Copy Rights…………………………………………………… 203
viii
Platform Politics…………………………………………………………… 211
Crossroads………………………………………………………………… 217
Local Cultures……………………………………………………… 217
Intersecting Industries……………………………………………… 221
Money, Media, and Intermediaries………………………………… 228
Conclusion…………………………………………………………………………… 234
Chapter 4: Too Real but Too Fake: Meta-authentic Media on Demand…………………… 237
Chapter Summary…………………………………………………………………… 237
Meta-authentic Media……………………………………………………………… 237
Meta-authenticity…………………………………………………………… 237
In Theory and Practice……………………………………………………… 243
Authenticity on Demand…………………………………………………… 253
Conclusion………………………………………………………………………… 257
Conclusion…………………………………………………………………………………… 258
References…………………………………………………………………………………… 261
Appendix A: List of Virtual Influencers…………………………………………………… 313
ix
List of Tables
Table 1. List of Interviewees………………………………………………………………… 13
Table 2. Timeline of Virtual Celebrities……………………………………………………… 29
Table 3. Traits of Virtual Influencers………………………………………………………… 68
Table 4. Technological Approximations of Virtual Influencers……………………………… 76
Table 5. List of Virtual Influencers (accessed on March 31, 2022) ………………………… 313
x
List of Figures
Figure 1. Research Account…………………………………………………………………… 11
Figure 2. Sibling Terms……………………………………………………………………… 79
Figure 3. Representation-Represented Relationship………………………………………… 81
Figure 4. Top 18 Virtual Influencers………………………………………………………… 87
Figure 5. Meme’s Women’s Body Hair Campaign………………………………………… 155
Figure 6. Google Search Term Trend for “Metaverse” …………………………………… 167
Figure 7. Virtual Influencer and Original Photo…………………………………………… 178
Figure 8. My First Original Virtual Human Created with Daz 3D………………………… 195
Figure 9. Image Created by Following Daz 3D Tutorial…………………………………… 196
Figure 10. Sex/gender-coded Base Poses in Daz 3D……………………………………… 200
Figure 11. 3d-sk’s Human Model-based Skin Texture Product on RenderHub 3D (2018) … 204
Figure 12. A Sample Photo Set (RenderHub 3D, 2018) …………………………………… 204
Figure 13. Edited Skin Texture Using the Sample Photo Set (RenderHub 3D, 2018) ……… 205
Figure 14. Two Virtual Influencers and a Kpop Star………………………………………… 208
Figure 15. Example Network of a Company-run Virtual Influencer………………………… 234
Figure 16. A Simplified Model of Virtual Influencer’s Meta-authenticity………………… 243
Figure 17. Virtual Influencer Serah Reikka’s Artist Mia Locklace’s Self-introduction…… 246
xi
Figure 18. Imma Gram’s Three-day Home Broadcasting IKEA Campaign………………… 250
xii
Abstract
This dissertation investigated how humanness and agency are being negotiated by various
human and nonhuman actors in the sociotechnical phenomenon of virtual influencers, defined as
photorealistic CGI “humans” who are, or aspire to be, social media influencers. “Virtually
human” technological nonhumans and technology-augmented humans are increasingly co-
inhabiting social environments. This undermines existing research, industry, and regulatory
practices that stem from anthropocentric assumptions, calling for renewed attention on the
supposed ontological divide between humans and nonhumans. Interlaced with multiple
contextual and performative ambiguities, virtual influencers are compelling cases of this
contextual shift. The dissertation drew on findings from phenomenological ethnographic
research that incorporated observant participation of various digital sites, semi-structured
interviews with global human participants, and archived digital data, such as social media posts,
news, podcasts, and derivative creative contents. Despite commonly being narrated as novel,
analysis of virtual influencers’ (pre)history and their variegated definitions pointed at many
continuing patterns of power, intermingled with discursive practices around the future. Multi-
directional approach to networks of virtual influencers demonstrated their socially meaningful,
“real” relational effects. It also showed how their virtual bodies can strategically connect and
disconnect from directly or discursively connected physical bodies to escape accountability.
Inclusion of nonhuman actors in networks suggested that despite the phenomenon’s supposed
liberating immateriality, virtual influencers come into being through networks of bodies, things,
and places, some of which were made hidden through historical structures of oppressive
exploitation. I propose meta-authenticity, defined as the desire or achievement of authenticity in
practices of extreme inauthenticity, as an accountability-motivated concept for critically
xiii
assessing boundary-crossing media’s “real” consequences without losing experiential nuances. I
argue for a shared sense of responsibility that fully recognizes each actor’s capacity to make
difference, including our own. Guidelines for ethical engagement and implications on the future
of human communication culture are also discussed.
1
Introduction
Everyday social spaces are increasingly being co-inhabited by humans and “virtually
human” technological entities that are not only becoming more sophisticated, but also
perceptually indistinguishable from humans. Nowadays, people can get news from bot
journalists, gossip with artificial intelligence (AI) conversational agents, and share intimacy with
personalized sex robots. Often, the goal in designing these technological entities is convincing
human-likeness. For instance, in 2020’s Consumer Electronics Show, Samsung showcased
“Neon,” an AI digital agent project that intended to create “independent but virtual living
being[s]” (para. 8) who look, feel, and act like humans (Smith, 2020). At the same time, humans
too are being introduced with newer ways to technologically present themselves, particularly in
mediated spaces. For instance, looks are being selectively altered with photo filters (Hong et al.,
2020), letters are being co-written with AI writing enhancers (Hancock et al., 2020), and video
evidences are facing legislative conundrums with the availability of deep fake technologies that
can convincingly switch out human faces from the original material (Maras & Alexandrou,
2019). Simply put, nonhuman technological entities have enhanced capacities to look and act
like humans, and humans have enhanced capacities to look and act like technological entities.
More importantly, these “virtually human” and “virtually nonhuman” cyborgs (D. J. Haraway,
1991) can interact with each other.
Then, what does it mean to be (non)human? How is the line being drawn, and does it
matter? The increasing co-inhabitancies are outdating existing anthropocentric frameworks in
communication research and are demanding a renewed attention to the seeming ontological
dichotomy between human and nonhuman (Guzman & Lewis, 2020). In fact, Xu and Liao
(2018) suggested that the distinction between computer-mediated communication (CMC) and
2
human-machine communication (HMC) has become blurred with the advent of technologies that
enable interactions with both human and nonhuman actors. Simply put, communication research
that binarily privileges human communicators is becoming incompatible with contemporary
social environment. There has been an emerging number of efforts to address this ontological
challenge among communication scholars, albeit more concentrated around the cases of AI (e.g.
Gambino et al., 2020; Guzman & Lewis, 2020; Hancock et al., 2020; Neff & Nagy, 2018; Xu &
Liao, 2018). For instance, Hancok and colleagues (2020) proposed the framework of AI-
Mediated Communication, in which they focus on how AIs can operate on behalf of humans in
message constructions. Guzman and Lewis (2020) reviewed the functional, relational, and
metaphysical dynamics of AIs to urge HMC scholars to reimagine how technology can be
approached as a communicator. Similarly, Gambino and colleagues (2020) argued that the
widely applied computer as social actors (CASA) framework misses how humans can now draw
on human-media social scripts in their interactions with technological actors, instead of relying
on human-human social scripts.
However, the ontological rupture is yet to be resolved. Guzman and Lewis (2020) located
the gap in the lack of thorough empirical examinations of what being human entails in the
current context of confluences. Ambiguities and variations in the interpretations of what it means
to be human are largely unaccounted for, circularly resulting in explorations on the rupture that
remain bound to certain ontological assumptions. For instance, there is an implicit
anthropocentrism in the idea of “smartness” in machines, which is based on the notion that
artificial performers are “(poor) imitations” of humans (Kuijer & Giaccardi, 2018, p. 10). This
human definition of intelligence can result in evaluations of machine capacities that may not be
realistic for nonhuman entities. Another example would be a reliance on a universalized imagery
3
of humans that presumes certain conditions and capabilities as the standard status of being, as
well as certain types of relationships with technologies. Posthumanist thinkers have repeatedly
questioned assumptions underlying the universal figure of “the Man,” such as its gender, racial,
and cultural background (e.g., Braidotti, 2013; Haraway, 2016), and feminist science and
technology studies (STS) scholars have shown empirical evidences of the generalization’s
consequences. For instance, Chasin’s study on the historical changes in machine and human
labor (1995) illustrated how the schematic opposition between human subjects and other-than-
human objects reproduces not only machine other-than-human objects but also human other-
than-human objects. To sum up, an unexamined reliance on the binary ontology leads to not only
incomplete theorizations, but also can reproduce structural marginalization in technological
practices.
The Case: Virtual Influencers
This dissertation investigated the phenomenon of virtual influencers to study the
negotiations around the notion of humanness and agency in this context of ontological challenge.
Virtual influencers are defined as photorealistic computer-generated imagery (CGI) “humans”
who are, or aspire to be, social media influencers. Like any other conventional social media
influencers, they strategically communicate and interact with their audiences to foster a
relationship that seems more authentic than what may be expected from traditional celebrities. In
fact, according to a report by HypeAuditor (Baklanov, 2019), virtual influencers have nearly
triple engagement rate from their followers compared to “real” influencers. Indeed, they can be
very popular. A self-proclaimed “robot queen” and a “change-seeking robot with the drip”
Miquela has garnered over three million followers on Instagram (@lilmiquela, accessed on June
4
16, 2022
1
; quotes accessed on March 15, 2020 and March 27, 2020, respectively). “The world’s
first digital supermodel” Shudu has over 233,000 followers (@shudu.gram, accessed on June 16,
2022) and Tokyo-based “virtual girl” Imma has 402,000 followers (@imma.gram, accessed on
June 16, 2022). In this respect, virtual influencers are functionally no different from “human”
influencers. In fact, a casually browsing user may not notice their technological origin due to
their highly sophisticated photorealism and lively personality.
The phenomenon of virtual influencers is a rich boundary-crossing case that is ample
with cyborgian ambiguities on agency, body and place, and authenticity among many things. As
in-between beings, virtual influencers help interrogate common binaries from diverse angles
such as that between human and nonhuman, as well as that between the virtual and “the real.” In
spite of their richness, research on virtual influencers is still limited (Oliveira & Chimenti, 2021),
with most of them focusing on marketing or interactive effects (e.g., Choudhry et al., 2022;
Kádeková & Holienčinová, 2018; Sands et al., 2022; Thomas & Fowler, 2020) or remaining at
an exploratory descriptive level. Although there is a handful of exceptions (e.g., Black, 2019;
Robinson, 2020; Sobande, 2021), empirical sociotechnical investigation on their cultural
implications is still lacking despite the phenomenon’s ongoing rapid expansion.
Meanwhile, in practice, virtual influencers’ supposed novelty and innovativeness are
being uncritically celebrated. Virtual influencers self-label them as new and the first, with many
interested parties such as investors, marketers, and media cheering them on. It is true that this
particular manifestation is novel in the sense that the phenomenon brings together various recent
developments in technologies and social media cultures, but social and cultural patterns that it
1
All access dates included in this dissertation indicate the last accessed dates.
5
embodies is not completely unforeseen. Lost in the celebrations are their unmonitored,
unregulated, and at times spectacularized emulations of humans.
This dissertation approaches virtual influencers as a case of meta-authentic media
phenomenon, with “meta-authenticity” loosely defined as the desire or achievement of
authenticity in practices of extreme inauthenticity. Meta-authenticity is a dynamic outcome of
ongoing collaborative performance by historically and culturally entangled human and
nonhuman actors. This means that meta-authentic virtual influencers can have “real”
consequences, and each of the actors can have “real” influence on the phenomenon’s shape. This
urges virtual influencers’ relational value to be taken seriously. At the same time, it demands
accountability from those that have been selectively claiming virtual influencers’ humanness,
such as those that have been avoiding questions of ethics by arguing that virtual influencers are
“not real” despite interacting with them as “real” beings.
How we narrate virtual influencers and other meta-authentic media matters. As an actor
involved in this phenomenon, this dissertation narrates virtual influencers as what are not
distanced from the past nor the real. Instead, I approach them as what are firmly grounded in
historical and material relations, as well as human bodies that dwell across various spatial
contexts.
Guiding Questions and Contributions
My motivation in studying virtual influencers has not been to arrive at a definitive
ontology of human and nonhuman, but rather to understand from the ground-up the negotiation
process around it to derive applicable theoretical and practical implications that can be
implemented in meta-authentic media practices. Big questions around the blurring ontology that
6
guided my investigation were: What does it mean to be (non)human?; Which actors are involved
in the negotiation of (non)humanness, and how?; What does nonhuman agency look like (does it
exist)? These questions were then specified into the following research questions to study the
case:
RQ1. What is the (pre)history of virtual influencers?
RQ2. Which actors are involved in the negotiation process of (non)humanness in the case
of virtual influencers?
RQ3. How is (non)humanness of virtual influencers being negotiated by these actors?
RQ4. How can virtual influencers’ agency be conceptualized and what are the
implications?
Virtual influencers were approached as cyborgs (Haraway, 1991), or an ongoing,
networked collection of multiple parts. With this assumption, sociotechnical assemblage of
virtual influencers was traced from an actor-network theory (ANT)-inspired approach to benefit
from its systematic open-ended thinking that includes nonhuman entities and discourses in the
scope of “the social.” However, this was in reflection of critical and political consciousness from
posthumanistic approaches and cultural studies. This was to analyze the reproductive and
creative meaning-makings in the negotiations rather than stopping at simply describing them,
and to maintain a reflexive awareness of how my mode of thinking may affect the research
process. I also drew on CMC and HMC theories on mediated interaction, such as self-
presentation, presence, and the CASA paradigm, to locate applicable places for theoretical
updates and to reflect on the ways the negotiations can be interdisciplinarily addressed.
7
This dissertation’s contributions are theoretical and practical. It addressed six non-
mutually exclusive issues. 1) What does this phenomenon mean for research on interactions and
social effects? 2) What does this for the contested conceptualizations of nonhuman agency? 3)
What does this mean for power and structure? 4) What does this mean for future design and use?
5) What does this mean for accountability and ethics? And combined, 6) what does this mean for
studying communication, particularly those concerning digital cultures?
This dissertation’s original contribution, meta-authenticity, is a concept that can be
applied to other boundary-crossing sociotechnical phenomenon, such as AR/VR/XR
technologies, AI-aided systems of automation, and blockchain-based cryptocurrencies that many
expect to propel the Metaverse. This dissertation also extended the multi-directional model of
selves (Banks, 2017) to nonhuman (i.e., virtual influencer) self-presentations while
complementing it with a layered approach to physical and virtual places to discuss the persisting
importance of the physical body in human experiences (D. Williams & Kim, 2019). This
combined approach allowed flexibility for embracing diverse forms of engagement with virtual
influencers while demanding accountability from each responsible actor. Implications of the
phenomenon on existing communication theories were also discussed, such as how the
anthropocentric assumption of human-nonhuman dichotomy in parasocial relationship and
CASA can result in inaccurate empiricism.
Contributions also include more case-specific explorations and analyses, although
research insights need not be fixed to the case. For instance, readers may find the following
useful for understanding this phenomenon and other thematically or conceptually connected
phenomena: chronology of virtual celebrities—which uniquely brought together detailed cases
from three language groups (Japanese, Korean, English), discussion of traits and typologies of
8
virtual influencers, case analyses of their self-/re-presentation, and discussion of human and
nonhuman entanglement in the phenomenon, accompanied by a sample network model. Those
interested in the issue of accountability around boundary-crossing technologies may also find the
suggested guidelines on responsible engagement useful, as well as discussions of legal and
ethical issues.
Methodology
This dissertation is based on the author’s phenomenological digital ethnography that
incorporated fieldwork data from observant participation, transcripts from semi-structured
interviews, and archived digital content, such as social media posts, news, podcasts, and
derivative creative contents. Concentrated fieldwork continued for roughly one year (Jan 11,
2021 - Jan 5, 2022), following several years of informal familiarization. For instance, on March
2, 2018, I created a private Facebook post with quick musings on the phenomenon, accompanied
by four news article links. On August 20, 2020, I set up a research account on Instagram,
although it was used only after acquiring research approval from my university’s Institutional
Review Board (IRB), granted on Dec 3, 2020. Similarly, less formalized research has been
continuing after the concentrated period. For instance, occasional deep field engagement, such as
live event attendance and dedicated short-term immersion, took place until April 2022 during
analysis and writing stages. Casual engagement with virtual influencers’ social media posts and
related online communities has been continuing, as well as daily skimming of email digests from
Google Alert and online newsletter subscriptions. Therefore, while core arguments reflect data
analysis from the concentrated period, specific cases discussed in this dissertation may
incorporate insights from meaningful updates since then.
9
Inspired by actor-network theory or ANT (Latour, 1996), I traced networked relations in
the phenomenon by following the actions of both human and nonhuman actors that emerged
during the course of research. That is, I strived to methodologically and theoretically privilege
actors’ perspectives, although with an understanding that this dedication still rested on my
situation as a researcher (see Casper, 1994). I began my fieldwork by focusing on a number of
pre-selected actor groups that were identified from the informal familiarization period, such as
popular virtual influencers, fans, creators, and news sources. However, I did not limit my
investigation to these groups and naturally expanded the scope to other involved actors, such as
different types of virtual celebrities, virtual human design software, and crypto asset promoters.
In fact, with virtual influencers, such a multi-sited, multi-actor angle was essential because the
phenomenon is not limited to a single physical nor digital locale (Hine, 2007; Marcus, 1995).
Therefore, rather than confining my research to a predetermined site such as a certain virtual
influencer’s social media feed, I aimed to flexibly “follow” the flow of communication and
interaction around the phenomenon to collect thick data that reflect its fluidity and dynamism
(Caliandro & Gandini, 2016).
Phenomenological ethnography is both compatible with ANT’s openness and can help
compensate for its shortcomings around researcher’s positionality. It is an approach that fully
acknowledges and confronts situated performativity of not only on-site participants but also in
the research process itself. Phenomenology assumes that each actor has their own “life-world”—
the totality of the world that is subjectively experienced—that they live by reflecting on their
open-ended stocks of knowledge (Goulding, 2005; Honer & Hitzler, 2015). Individual life-
worlds are not only difficult to access but are heterogeneous. This means that people must
continuously reorient themselves in their everyday life because other peoples’ life-worlds coexist
10
in a mutually constructive fashion with their own (Honer & Hitzler, 2015; vom Lehn & Hitzler,
2015). Phenomenological ethnographers conduct “observant participation” to understand
meaning-makings that occur in this process. “Participant observation” may be the more
commonly known term, but the order of the words is intentional. Observant participation’s
lexical emphasis on participation is in active recognition of the researcher’s situatedness and
subjectivity and how these factors can influence the phenomenon (Katz & Csordas, 2003;
Pfadenhauer & Grenz, 2015). Thus, in addition to detailed phenomenological observations,
researcher’s subjective reconstruction of their internal viewpoint around their field participation
is acknowledged as an important data point. In other words, researchers are to observe the
phenomenon as it unfolds while noting how they are co-constructing it during the research
process. To express its emphasis on capturing living forms in real-time social interactions,
phenomenological ethnography has been described with the term “anti-culture” (Katz & Csordas,
2003) and as a middle position between naturalism and constructionism (Maso, 2007). These
descriptions highlight another important benefit of the method for the current project: it
challenges a presumed, closed conceptualization of humans and nonhumans (e.g., Neff & Nagy,
2016; Pfadenhauer & Grenz, 2015).
Phenomenological ethnography encourages reaching triangulation by circularly
abstaining from theoretical and common sense knowledge, second hand knowledge (e.g.
interviews, observations), and subjective experience, as well as by alternating between data
collection and analysis (Honer & Hitzler, 2015; Pfadenhauer & Grenz, 2015). I utilized three
data collection phases and interim data organizational techniques to achieve this effect. Activity
of focus for each of the three phases were deep observant participation, in-depth interview, and
news article archiving, respectively. General fieldwork, such as social media monitoring and
11
screen capturing, continued throughout the entire period. Collected data, which included notes in
addition to archived images, videos, and texts, were periodically reviewed and organized. Breaks
between the phases were particularly useful for this task, serving as interim pause points for
basic analysis. Example themes for charts that I managed include a compilatory list of Instagram
virtual influencers, a summary chart of field observation highlights, an in-depth interview
tracker, and lists of news articles.
Figure 1
Research Account
For field observation and to interact with potential research participants, I used a research
Instagram account that was dedicated to the topic of virtual influencers (Figure 1). The profile
page indicated my primary relationship to the phenomenon with the phrase “Digital Culture
Researcher” in three languages with a link to my academic website. The account handle,
“@phdigitaldonna” was also chosen to signal my positionality. Whenever it was necessary for
me to set up an account on a different platform, I created a similar profile unless their
requirements or norms demanded that I use my personal account with further identifying
12
information, such as on LinkedIn and Facebook. Regardless, I always clearly stated my
affiliation and overview of the research when I personally reached out to participants.
One of the routine observant participation activities was monitoring virtual influencers’
Instagram feeds and their 24-hour only Story posts. This often led me to other sites within
Instagram and beyond, from their profiles on a range of other social media platforms to fan
Discord channels, creator-focused Facebook groups, news sources, derivative YouTube videos,
live events, virtual conventions, podcasts, and software programs to name a few. I mainly
focused on exploring those in languages that I have proficiency in—English, Korean, and
Japanese, but used online translation services when those in different languages were part of the
larger conversation. I took fieldnotes and archived noteworthy digital content through screen
captures, screen recordings, and transcription. Similarly with Cotter (2019), I determined these
data’s verbatim inclusion and anonymization in the dissertation by gauging participants’
expectations. For instance, although some of the sites required membership, they were sizable
groups that were listed or promoted publicly with rather simplistic membership approval process,
such as answering questions like why I am interested in the group or posting a non-identifiable
bio. Therefore, I collected data from these groups with the assumption that members can
“reasonably expect to be observed by strangers” but paraphrased quotes and anonymized user
names in the dissertation to ensure protection of possibly vulnerable population (Townsend &
Wallace, 2016, p. 8). Comments and posts on openly accessible sites were quoted verbatim but
anonymized, with some exceptions that have been additionally paraphrased to prevent
identification. Virtual influencers that presumably have “publicity expectations” (Richterich,
2020) were not anonymized, determined by whether they have been verified as a public figure by
the platform or are run by a media-recognized company. Archived materials were re-accessed
13
during the manuscript writing and revision stages. All access dates included in this dissertation
indicate the last accessed dates.
In addition to observant participation, twenty semi-structured in-depth interviews were
conducted with adult participants from diverse backgrounds (Table 1). One interview was over
email, two were over live chat messengers, and seventeen were over the video call platform
Zoom. Participants were recruited through targeted outreach, online fliers, and snowballing
technique. Targeted outreach combined inquiring through company websites, cold messaging
virtual influencers and creators on social media or via email and reaching out to members in
virtual influencer-related groups after a sufficient time of immersion. All outreach and
scheduling messages included information about me, the project, and the interview process. A
full text version of the informed consent was also made accessible via a link, and I sent sample
questions in advance when requested. Synchronous interviews lasted between 50 to 110 minutes,
excluding the time spent on the informed consent procedure. All the interviews were conducted
in a private, enclosed physical place and were recorded and transcribed with consent. The
interviewees had the option to request their interview files, in which case I emailed them as soon
as requested file(s) were ready. Automated basic transcriptions were generated from Zoom and
Clova transcription services but were personally triple-checked and corrected for increased
accuracy and data immersion. Only the transcriptions were used for analysis after this process.
Table 1
List of Interviewees
# Type Area* Name** Company Instagram Handle*** Mode Language
1 Creator Europe Annika Kessel CosmiQ Universe
leyalovenature
aya.stellar
Video English
2 Fan Charlie Chat English
3 Marketer Asia Christopher Video English
4 Media
North
America
Christopher Travers VirtualHumans.org Audio English
14
5 Creator Asia
Daichi Kambayashi,
Sonya
Atali meme.konichiwa Video
Japanese,
English
6 Creator
North
America
Dulce Baerga dulce303 Video English
7 Creator Asia Jae Wook Oh DOB Studio ruuui_li Video Korean
8 Creator Europe Jawad El Houssine serahreikka Video English
9 Creator Oceania Jeasy Sehgal Graphic Monk Video English
10 Creator Africa Jessica James digi.akuchi Audio English
11 Creator
North
America
Mikirah Muse mikirahmuse Video English
12
Virtual
Influencer
Monna Haddid
monna_haddid
monna_haddid_official
Chat English
13 Creator
North
America
Nina Hawkins Lilium Labs ninocence Video English
14 Creator Africa Reggie Goff abawils Video English
15 Creator Asia Seihee Kim Klleon woo.ju.like Video Korean
16 Creator Asia SIA Bangkok ai_ailynn Email English
17 Creator
South
America
Tatyana Maison Amany ivaany.h
Video,
Audio
English
18 Creator Europe Ted Werdolf ted_werdolfs_cgirls Video English
19 Creator Europe Tom Pastor lisa_gen_lg Video English
20 Creator
North
America
Xander Smith Video English
Note. Interviews were conducted during September – December 2021. The chart reflects
information collected on the day of the interview and is alphabetically ordered by first name
(company name if not available). Some interviewee name(s) and their virtual influencer(s), both
past and ongoing projects, are omitted from the list and/or changed in the dissertation.
*Location-based. Some reflect the country the interviewee identified with.
**Includes pseudonyms.
*** Associated (previous) virtual influencer or digital twin account if present.
Separately from those collected during the general fieldwork process, more
comprehensive datasets of news articles and other digital native sources were collected through
two methods. Firstly, using MediaCloud’s Global English Language Sources and US Top
Sources 2018 database, I scraped a list of English articles with the keywords “virtual influencer”
and “CGI influencer” that were published during the five-year period between September 11,
2016 and September 10, 2021. The initial list had 177 articles. Duplicate, irrelevant, or
inaccessible articles were removed. Some invalid links were updated rather than omitted when
the correct version was available upon searching the article title on Google. The cleaned list had
127 articles. I created a PDF file of each article by full screen capture or page printout, choosing
15
whichever method that best preserved the original formatting. PDF files were then converted to
detect text.
Secondly, I used Google Alert’s keyword-based daily emails to complement this dataset.
I manually created a list of English sources with the keywords “virtual influencer” and “CGI
Influencer” and Korean sources with the keywords “virtual influencer,” “virtual influencer
[Korean],” and “gasang influencer [Korean]” that were published during the roughly six-month
period between March 16, 2021 and September 10, 2021. Originally, I had also subscribed to the
keyword “digital influencer” from Google Alert as it was one of the terms that I saw during my
observations but deleted it after receiving too many irrelevant results. From 359 archived emails
containing one or several links each, I created a list of 355 articles with 258 Korean sources and
97 English sources. This was the result of manually visiting each link and discarding not only
duplicate, irrelevant, or inaccessible sources but also pages with one of the keywords but in the
form of complied search results, non-original scraped text, different language-based post, or
extremely short comment without further context. In addition, links overlapping with
MediaCloud results were deleted, too. Faulty links were replaced with the correct address when
available. Links that directed its visitor to a different, richer content page, such as to a video
material or a podcast, were replaced accordingly, too. Similarly with the MediaCloud list, after
arriving at a cleaned list, I created a PDF file of each source. However, when the link led to an
audio or a video file, I additionally created a transcript. PDF files were then converted to detect
text, although due to formatting most Korean sources were undetectable.
Collected data were analyzed with a grounded theory approach (Glaser & Strauss, 1967).
Following the ethos of phenomenological ethnographic research (Honer & Hitzler, 2015;
Pfadenhauer & Grenz, 2015), I periodically revisited and organized collected data, which served
16
as a means to routinely engage in informal coding and constant comparison. After completion of
the concentrated data collection period, I analyzed the datasets through close reading and open
coding. Outputs from interim data analysis such as summary charts and written reflections aided
this process, particularly subjective experience-based data like fieldnotes and selectively
archived materials during fieldwork. Text-heavy secondary data such as transcripts and news
articles were additionally thematically analyzed (Braun & Clarke, 2006) using qualitative data
analysis (QDA) software NVivo 12 for better triangulation. Initial codebook was drafted based
on interpretive themes from prior rounds of analysis. These codes were refined by using every
fifth transcript in the order interviewed as samples. In the case of MediaCloud and Google Alert
datasets, multiples of tens in the order published were selected as samples, additionally using the
interview codebook as a reference point. Resulting codebooks were used to analyze respective
datasets with the aim of refining interpretations to reach theoretical saturation. Some categories,
such as traits of virtual influencers, were more granularly thematically analyzed to complement
the overall analysis.
Chapter Overview
Chapter 1 delves into how narratives of novelty are constructed around the phenomenon
of virtual influencers. Common patterns of power can be traced by connecting contemporary
cases with those that are narratively placed outside or at the edge of the novelty of the
phenomenon. Therefore, in this chapter, I discuss the history of virtual influencers and diverse
definitional efforts around it. I argue that definitions of virtual influencers are inconsistent not
necessarily because the phenomenon is new and thus not yet orderly, but because its novelty is
intermingled with discursive practices around the future. The history section introduces the
context of this phenomenon and relevant themes by discussing the predecessors of virtual
17
influencers from around the globe, particularly focusing on Japanese, Korean, and American
cases. Readers who are not familiar with the phenomenon would find the section useful. The
following section investigates how virtual influencers are being defined and described, as well as
what assumptions underlie these discursive practices, by drawing on news, interviews, and social
media posts. Commonly summoned traits of virtual influencers, technologies that tend to be
(wrongfully) attributed to them, and their sibling terms are discussed. I argue that it is important
to pay close attention to the narratives as it affects our future path, not simply about virtual
influencers but also about other “new” technologies and sociocultural patterns that they are
nested in.
Chapter 2 interrogates the politics of virtual influencers’ presentations and the “real”
effects that they can have. I propose approaching virtual influencers as beings that are composed
of multiple self-networks that are co-created by human and nonhuman collaborators. This
chapter particularly zooms in on more salient actors, such as virtual influencers, their creators,
and fans, and how these actors’ desires intersect. I argue that we must take into consideration
how their virtual bodies can affect physical bodies, whether they are directly or discursively
connected with them. This includes bodies that share identity categories that virtual influencers
emulate or can represent. To investigate the relations, I apply the multi-directional framework on
various cases of sexually and/or racially/ethnically objectified virtual influencers and discuss
how their virtual bodies strategically connect and disconnect from various physical bodies that
they can be grouped together with. Some guidelines for ethical practice are suggested. While this
chapter urges the virtual to be taken as “real,” it does so not to advocate for the binary hierarchy
between the virtual and the physical, but rather for a hybrid perspective that continues to actively
18
recognize how human experiences across different spatial layers are yet still inevitably
interconnected with our physical existence.
Chapter 3 broadens the analytic scope to the agency of less salient human and nonhuman
actors, taking inspiration from STS, posthumanist approaches, and cultural studies. Despite the
utopian imaginaries around supposedly immaterial, liberating Metaverse, virtual influencers
come into being through networks of bodies, things, and places. Actors identified in this chapter
includes hidden human laborers, creative tools, social media platforms, localities, adjacent
industries, marketers, and media. The benefit of including these human and nonhuman actors in a
multi-directional, networked approach is that historical patterns and material relations that
bolster structural patterns can be more accurately traced and responsibly responded to. I center
the discussion on analyzing those related to human oppressions, particularly regarding exploitive
labor relations. Reflecting on both the pragmatic and ethical urgency, I argue that what is
desirable in this cyborg phenomenon is a shared sense of accountability that fully recognizes
each actor’s capacity to make a difference, including our own.
Chapter 4 is about why and how we must ask the question of “real” in virtual influencers
and what this inquiry means to the intersection between the social notion of being human,
communication technologies, and contemporary media cultures. It brings together the findings
through the original concept of “meta-authenticity,” loosely defined as the desire or achievement
of authenticity in practices of extreme inauthenticity. It is a dynamic result of collaborative
performance by historically and culturally entangled human and nonhuman actors. Theoretical
and practical contributions are discussed by demonstrating how the concept may be applied to a
case. I also reflect on its implications on the future of human communication culture, such as
regarding meta-authentic technological actors’ capacities to provide authentic interaction
19
constantly and convincingly on demand. Above all, I argue that the concept of meta-authenticity
encourages critical assessments of boundary-crossing media phenomena’s “real” effects and
consequences without losing their experiential nuances, and vice versa.
20
Chapter 1. “I am the First Virtual”: Narratives of Novelty and
Technocapitalistic Imaginaries of Virtuality
Chapter Summary
Virtual influencers are interesting not because they are new but are being positioned as
new. The goal of this chapter is to disentangle virtual influencers from the myth of being the first
and new; namely, the narrative of novelty. To do this, the chapter presents a sketch of what
virtual influencers are. It starts with an overview of how the field of communication and its
adjacent fields have approached narratives of novelty around media and communication
technologies. It discusses how these narratives espouse visions of the future and expectations
about the social and cultural patterns implied within them. This section is followed by a brief
history of virtual influencers, from their predecessors to contemporary iterations. While many
virtual influencers claim to be “the first,” their history suggests that their freshness may come
from being positioned as so rather than actually being novel. This section also introduces some
of the themes that will be delved into in the subsequent chapters. Lastly, the chapter will discuss
variegated definitions of virtual influencers to show how novelty has been constructed in the
phenomenon. By discussing various descriptive efforts that attempt to bound the phenomenon, it
is possible to access desires embedded in their respective narrative, as well as join the social
process of re-telling their course of action in the future. I draw on news, interviews, and social
media posts to discuss commonly summoned traits of virtual influencers, technologies that tend
to be (wrongfully) attributed to them, and their sibling terms. Discussions of history and
terminologies in this chapter also serve as useful resources for analyses in the subsequent
chapters.
21
Narratives of Novelty
This dissertation defines virtual influencers as photorealistic CGI virtual humans who are,
or aspire to be, social media influencers. Many virtual influencers—or however they are
labeled—adorn the titles first and new, both self-christened and given. On their respective
Instagram profile, Shudu (@shudu.gram
2
; for more information about mentioned virtual
influencers, see Appendix A) calls herself “The World’s First Digital Supermodel,” Rozy Oh
(@rozy.gram) “KOREA’S FIRST VIRTUAL INFLUENCER,” and Aba Wils (@abawils)
“Africa’s First Virtual Influencer” (all accessed February 4, 2022). A Forbes article described
Noonoouri’s (@noonoouri) endeavors as something that can “tap into consumers in a completely
new way” (Coleman, 2018), and a Mashable article suggested that we should think of our “new
CGI friends as the pixelated pioneers of a new, formulated frontier” (Hill, 2019). Marketers have
been touting them as “an innovative new part of the influencer landscape” and “cutting edge”
(Stanley, 2020), or as “the New Trend in Social Media Marketing” (Tayenaka, 2020) that brands
must explore to “have that next iteration of what a brand’s voice is to inhabit those virtual
spaces” (Hackl, 2021). These labels also adorn the fine-grained activities that the virtual
influencers engage in. For instance, for Korea’s virtual influencer Rozy, this included activities
such as being the first virtual influencer to appear on a Korean TV commercial for financial
services (J. Yoo, 2021) and being the first virtual influencer to showcase a solo fashion spread in
a specific domestic magazine (Bae, 2021).
However, as the second section of this chapter will demonstrate, history of the virtual
influencers’ predecessors leaves what is new or first about them not so fresh, if not somewhat
2
Account handles are Instagram handles unless otherwise indicated.
22
underwhelming. Of course, this is not to deny that there have been some changes from the past
media environment, as the contemporary concept of “influencer” suggests in the term virtual
influencer. It is also possible that people may not have been too aware of the predecessors or that
they simply wished to stand out among the sea of human and virtual human influencers. Still, it
is important to approach the arbitrary discursive practices around virtual influencers with an
understanding of their mythical aura of being new. This is because it helps us distill some of the
intentions and consequences implied in their usages. In fact, in the history of media and
communication technologies, even the claims of newness are not new, as well as related
discussions on the interests we need to locate within the claims.
The Problem of Novelty
Now possibly perceived as no longer so new, the spread of the term “new media” had
invited many debates around the notion of novelty. In the founding issue of the reputed
communication journal New Media & Society, Roger Silverstone (1999) argued in reflection of
its name that novelty implied in new media is not what should be “assumed nor taken for
granted,” as thus is “the problem” (p. 12). Some of the earlier efforts at tackling the problem of
“new” media included works that aimed to define it from more formal and technical grounds. For
instance, in The Language of New Media, Lev Manovich (2001) proposed five technical
characteristics of new media that he believed to differentiate them from traditional media:
numerical representation, modularity, automation, variability, and transcoding. However, these
attempts were accompanied by the understanding that their newness was not simply a
technological matter, but also that of content and organization within and beyond the media of
concern (see Silverstone, 1999). In fact, Manovich also argued that there are computer and
cultural layers to new media, and the former’s logic, such as how computers understand, process,
23
and represent data, can significantly affect the cultural logic of media (pp. 63-
64).Communication scholars like Sonia Livingstone (1999) envisioned this process to be further
collaborative and participatory than what had been possible before with traditional media; new
media’s capacity for interactivity was expected to bring changes in audience roles and
engagement. In short, while new media’s novelty was traced in technical and material shifts,
these shifts were understood to be intertwined with the social and the cultural.
The problem of novelty was also approached by questioning not what is new, but how
and why certain media became perceived as new. These works focused on both how novelty is
constructed in technology narratives and its dependence on the binary notion of the old and new.
Media historian Carolyn Marvin (1988) studied the ways which media and communication
technologies that were once considered new become considered as old, emphasizing the social
and cultural processes rather than technical: “They are constructed complexes of habits, beliefs,
and procedures embedded in elaborate cultural codes of communication” (p. 8). Lisa Gitelman’s
work (2006) also famously warned against simplistically treating new media as a self-acting
force that brought forth a new epoch or a rupture. Instead, she suggested our keen awareness on
how new media worked together with existing social, economic, human, and material forces—
including the inequities within them—to refashion itself as new and others as old. The two
scholars were not alone in terms of having located (some of) the roots of technological
development within social and cultural contexts (e.g., Marx, 2010; Mumford, 1962; Pinch &
Bijker, 2012; Postman Neil, 1998; Rogers, 2003; R. Williams, 2014; Winner, 1988), rather than
deterministically viewing technologies as independent drivers of history (e.g., Baudrillard, 1994;
McLuhan, 1964; for detailed analysis, see Carey, 1967). Still, their work is important to the task
of understanding virtual influencers’ novelty because they focused not only on the relational
24
process of development, but how technologies have gained the labels of being old and new
during it.
Indeed, “new” is a socially agreed upon label, and one that is not static nor once-only.
Processes like wide diffusion and standardization may seem to permanently rid a media
phenomenon of the label, but it is possible for it to re-“appear as new, if not newer, than the first”
in different communication forms (Peters, 2009, p. 22). The continuous revival of the notion of
fictional celebrity in the timeline in the next section will demonstrates this. Such resurgences are
possible because newness is not a “self-evident social category,” but what results from people’s
mutual interpretation (Gershon & Bell, 2013, p. 259). In the same vein, media historian and
technology scholar Simone Natale (2016a) argued for viewing “new” and “old” as relational
terms, rather than as absolute statuses that are disparate from how people associate with media.
The ways which people can relate to media and communication technologies are not limited to
their technical features but encompass semiotic and experiential factors such as people’s
perceptions, imaginations, habits, and experiences (Natale, 2016a). Therefore, virtual
influencers’ newness is not simply a technical matter but can be interpreted as results of, and at
the same time tactics to influence, the discourses and practices around them. As the crux of the
virtual influencers’ narrative (Bory, 2019; Natale, 2016b), story (Nagy & Turner, 2019), or
imaginaire (Flichy, 2007), it conjures the magic (Murray, 2020) or sublime (Mosco, 2004) that
pardons virtual influencers from scrutiny.
Various actors are involved in narrating such tales of a “new” media phenomenon. In
their study of how virtual reality’s (VR) newness is constructed, Jeff Nagy and Fred Turner
(2019) focused on developers and marketers. They observed how the speakers at developers’
conferences had rerouted the genealogy of VR in a way that could sidestep unsuccessful past
25
cases of VR while repositioning it as the enlightened, “new” course of the future that their fellow
developers should also place their faith in. Marketers of Dos Equis beer on the other hand were
observed to use its newness simply as the shiny wrapper for making the familiar, conservative
marketing message seem “new and improved” through the “magic of the legitimacy exchange”
(p. 545). Scientific research can contribute to this process as well. Janet H. Murray’s critical
study (2020) on “magical thinking” (p. 13) around VR and the life lived in the virtual world—
i.e., the Metaverse—discussed the presumed inevitability of the VR-immersed world and
technology determinism in related research. Emphasizing how construction of such a magical
narrative can accrue its own power that is independent of the technology’s actual capacities,
Murray pointed out that such unchecked beliefs embedded in scientific research can lend
misleading authority to the supposedly new, futuristic medium while overlooking the willing
participation from the users that they require, as well as the techniques like having a “threshold
object”—e.g., the act of putting on a pair of goggles to enter the VR world—to help with the
process. In fact, regardless of whether the tale is optimistic or pessimistic, foregrounding
newness can flatten or distort the types and intensities of audience engagements that help settle
the technology (or the social perception of it) within the context of everyday (Carpentier, 2011).
For instance, the popularly repeated stories of inexperienced audiences’ panic to the moving
image of a train in the earlier days of cinema—i.e., “the train effect”—and that to the “War of
the Worlds” announcement on fake Martian invasion at the beginning of radio broadcast era have
been helping establish the symbolic transition from the old to the new communicative era of
powerful media effects, despite both being exaggerations of what in reality were small, mild
reactions (Natale, 2016b; Pooley & Socolow, 2013). Furthermore, it is also possible for
technologies themselves to contribute to its own narrative and technical trajectory when imbued
26
with magic or “voodoo,” especially in the absence of experts (Whitson, 2018). Just like humans,
things—from seemingly trivial things like rental keys that silently nag people to return them
through its bulkiness to seemingly immaterial technologies for virtual influencers—constitute
social relations and engage in meaning negotiations within them (Latour, 1994; Natale, 2016b).
In other words, narratives of novelty stem from and are intertwined with existing
patterns, beliefs, and agendas in history that result from negotiations among various actors. For
instance, regarding the narratives surrounding the promises of the digital, the “Internet
imaginarie” has been argued to be comprised of mostly utopian visions of the impending dawn
of a new era where democratic freedom around corporeality, politics, and economy would be
realized (Flichy, 2007). Such myths on the sublime prowess of digital technologies offer the
society and its members an entrance into a different reality that transcends the banality of the
everyday, however in effect by distinguishing itself from, and thus denying, history (Mosco,
2004). While comfortable this denial may be to many folks, it is not so innocent because it also
means overlooking the existing patterns of power in naive hopes of a transcendent future. As
communication scholar Vincent Mosco (2004) said, “to deny history is to remove from
discussion active human agency, the constraints of social structure, and the real world politics”
(p. 35). That is, whether they are true or not, narratives of novelty are capable of affecting the
negotiations on what the “new” media phenomenon’s impact can be by influencing social actors’
beliefs and behaviors around them, however in a way that can hide persisting patterns of power
that may continue to structure the supposedly novel phase of the world (Nagy & Turner, 2019;
Natale, 2016b). This is why we must not get lost in the claims of “new” and “the first” in looking
at virtual influencers.
27
Regardless of their allure, we must not forget that stories of newness are indeed stories
that various storytellers come together to circulate, some with less mutually beneficial desires
despite the awe and hope they may put forth as gloss. This is why we must look beyond the
novelty that may seem to justify and hold together the variegated definitions, and into the
historical continuances. Virtual influencers’ seeming proximity to the supposed ontological
uniqueness of human adds further weight to the importance of this task since their narratives
therefore can more directly affect visions around human lives. For instance, Paolo Bory’s (2019)
study on the narratives around highly publicized competitive gaming AIs explained the
differences between the “too humanlike” chess AI Deep Blue and the “un-humanlike” go AI
AlphaGo (p. 636). “Too humanlike” Deep Blue was narrated as a material machine that could
imitate or even surpass humans, a tale beneficial for promoting the capacities of the computer
device manufacturer IBM that sponsored its match with Gary Kasparov. On the other hand, “un-
humanlike” AlphaGo was narrated as an immaterial system that could creatively and even
beautifully contribute to interactions between human intelligence and artificial intelligence, a
story that helped put Google’s DeepMind, the base project that initiated AlphaGo’s match with
Lee Saedol, in a different, positive light. Depicted as beautiful and even god-like, AlphaGo was
effectively positioned as a novel being that could communicate and learn together with, or
perhaps even teach, humans. While both Deep Blue and AlphaGo fall under the same
technology, AI, their stories encouraged a different reading of what it could mean to the
humanity.
In the following sections, I will discuss the timeline of virtual influencers’ (pre)history,
followed by an analysis of contemporary discursive practices around them. I will interrogate the
ways in which they are being defined behind the sublime of “new” and “the first” to understand
28
how they are being articulated into contemporary society. As discussed in this section, these
negotiations can push towards a certain consensus on their design and uses, possibly even acting
as a “conservative force” that can reinforce “already existing and often unspoken agreements
about the ways that culture should be organized” (Nagy & Turner, 2019, p. 548). Some of the
direct consequences that are particularly relevant to their definitions would be policy and legal
decisions, as what virtual influencers mean determines the pertinent stakeholders, arguments,
applications, and desired outcomes (for an example around the definition of the “Internet,” see
Crawford, 2007). It can also affect how and by whom identities and categories are represented,
discussed, and challenged (Gershon & Bell, 2013), potentially with powerful effects on humans
as many contemporary virtual influencers boast their convincing visual and behavioral human-
likeness. In short, what we can understand from scrutinizing the past and present of virtual
influencers is the direction the phenomenon and its surrounding narratives are nudging us
towards, which, when the shine fades, may in fact not be so novel.
Virtual Influencer History
While virtual influencers may seem new, they have predecessors: virtual celebrities. This
section presents a brief timeline of their history to resist ascribing rash newness to the
phenomenon, as well as to introduce some of the themes that will be visited in the subsequent
chapters. The history is recounted with a generous flexibility to their evolutionary tree. This is
because the malleability of the term makes it difficult to clearly bound what could count as their
predecessors, as we will see in the succeeding section. Therefore, the following chronology is
meant to serve as a compilation of viable comparison points that can help draw out historical
overlaps and shifts in the course of virtual influencers’ development. Although not exhaustive, it
includes both popular and relatively lesser known CGI predecessors of virtual influencers around
29
the globe (see Table 2). These examples hint at the twists and turns in the history that contributed
to their contemporary messy definition. While CGI-made photorealistic humanoid virtual
celebrities were prioritized in organizing the timeline, some important physical and non-
humanoid figures were included to enrich the discussion.
Table 2
Timeline of Virtual Celebrities
Year Name
-1980s
Fictional band
1958 • Alvin and the Chipmunks
1966 • The Monkees
1969 • The Archies
1970 • Josie and the Pussycats
1985 • Jem and the Holograms
1990s
Virtual idol
Cyber singer
1990 • Haga Yui
1991 • Sailor Moon
1992 • Dokyusei
1994 • Macross Plus
• Tokimeki Memorial
1996 • Date Kyoko DK-96
• Idoru
• Lara Croft (Tomb Raider)
1998 • Terai Yuki
• Adam
• Lusia
• CYDA
• Gorillaz
1999 • Diki
2000s
Virtual band
Vocaloid
2000 • Hana
• Intz
2001 • Aki Ross (Final Fantasy: Spirits
Within)
2002 • Limit
2006 • Dethklok
2007 • Hatsune Miku
2010s-
Virtual influencer
V-tuber
2011 • Ami Yamato
• Your Favorite Martian
2012 • Aimi Eguchi
• Barbie
• “Pac-O-Gram”
2016 • Miquela
• Kizuna AI
2017 • Shudu
• Code Miko
2018 • K/DA
2019 • Data Ayano
30
• AI Angel
Note. Types beneath each decade do not represent all existing types but the main examples
discussed in this chapter.
1990s Japan and Korea: Virtual Idols and Cyber Singers
Music was the choice of path to fame for the earliest virtual celebrities. In Japan, the first
“virtual idols” are known to have appeared in the 1990s, such as Haga Yui [debut: 1990], Date
Kyoko [1996], and Terai Yuki [1998] (Galbraith, 2009; Sone, 2017; “Terai Yuki,” 2021). In
Korea, the “cyber singer” Adam [1998] struck the country as a cultural sensation, followed by
his lesser known cyber contemporaries Lusia [1998] and CYDA [1998] (Y. Jung, 2019; D. Kang,
1999; S. Park, 2016). UK’s virtual band Gorillaz [1998] also debuted in the late 90s, which
arguably is one of the longest surviving and one of the most critically well-received virtual
celebrities, having achieved musical feats such as (holographically) headlining the 2010
Glastonbury music festival and having received one award and eleven nominations for Grammy
Awards.
To begin with Japan, Haga Yui
3
has been described as “the first virtual idol” (Sone,
2017). Sounding similar to the Japanese word for frustrating (hagayui), Haga Yui debuted in
February 1990 as a faceless teen idol. Sporting a ponytail, this supposedly 5’2’’ (158 cm) 97 lbs
(44kg) 15-year old girl appeared in interviews, music performances, and fan meetings with her
eyes physically or graphically covered. This included in-your-face techniques such as a curtain
that Haga Yui hid behind during fan handshake meetings, a hand-held prop to manually cover
her eyes, camera angles that would cut off the upper part of her face, and various rudimentary
computer graphics like blurs, mosaics, colored boxes, and shaped cut-outs. The promotional
3
Japanese names start with the last name.
31
image of her debut single Starlight Passport (hoshizora no pasupōto) featured a side angle
picture of a ponytailed girl laying on her stomach, clad in a school uniform. The girl’s eyes were
covered with an orange colored word block that read “HAGA YUI” in black letters. While Haga
Yui was by no means a photorealistic full-CGI human, nor was likely to be mistaken as one,
there are several factors that still make her an important figure in the history of virtual
influencers.
Haga Yui was not simply a “non-existent” celebrity that relied on the existing media
culture for her tongue-in-cheek cultural references, but first and foremost a being that was born
out of a participatory project that involved multiple individuals’ knowing and unknowing
collective contributions. She was produced as a parody project about Japan’s 1980’s idol culture
where young, innocent, and perfect girl who would “not even go to the toilet” (“Haga Yui,”
2022) was envisioned as the ideal imagery of an idol star. It was in November of 1989 during
Nippon Broadcasting System’s (nipponhōsō) radio show Ijuuin Hikaru’s All Night Japan (ijūin
hikaru no ōrunaito nippon) where the idea for Haga Yui began to ignite attention. After hearing
the host Ijuuin’s playful jibe at the typical ring that idol star names have, listeners began to send
in postcards that chimed in on his poke at Japanese idol culture. This subsequently grew into a
listener participatory project where Ijuuin and his listeners collectively drew on idol star
conventions to construct a profile for the unidentifiable yet crush-worthy girl idol Haga Yui. She
was convincing as a teen idol but also satirical. For instance, on the surface, her debut single
Starlight Passport can be taken as a bubbly, cheery pop song about teenage crush. However, the
same cutesy high-pitched voice that sings “love with cheer” and “fly, two people’s universe”
also innocently chirps “you are used to my mischievous side profile, but it’s a lie” and “if you
32
are an adult, please notice, I’m frustrated [hagayuino; alternatively, this can be translated as
“Haga Yui’s”] my boy.”
Beyond ideation, Haga Yui herself was also a collection of individuals. Ijuuin mentioned
having hired fifty seven individuals to play her part, including for different public facing roles
such as for personality appeal, singing voice, and photo albums. In fact, while Ijuuin led the
project, his listeners were core parts of Haga Yui’s idol career. For instance, many of the
contestants at the second Miss Ponytail Contest (the first of which Haga Yui claimed to have
won) were female listeners of Ijuuin, although models who were unaware of the Haga Yui
project also participated, including the winner of the event. Haga Yui’s early retirement decision
to study abroad in Taiwan was also the result of collective feedback from listeners, concluding
with a small number of fans gathered at the Haneda Airport for her unofficial, unaired send off in
October, 1990 (“Haga Yui,” 2022; Kracker, 2014; pureyādo, 2011).
The case of Haga Yui raises important points to consider when hacking what may seem
like a strange phenomenon of virtual stardom. Of course, being a dedicated collective parody
project, Haga Yui’s mission differs from many contemporary virtual influencers who instead
tend to strive for a standalone name for themselves. Moreover, her CGI was at best sufficient
compared to contemporary cases. Unlike many of them who attempt to remain secretive of their
origin, she was also not shy to hide that she relied on physical bodies that played her—for
instance, at one of the fan photo events, three Haga Yuis appeared as a ruse and a self-satire: a
“cute” girl, a “normal-looking” girl, and a non-Japanese girl (pureyādo, 2011). Regardless of
these differences, she was one of the earliest cases that proved that an identifiable human may
not be necessary in the path to fame, or perhaps that the lack of it can be the part of the appeal.
Haga Yui’s open secret about having multiple body models also suggests that there is no need for
33
a one-to-one match between the on-stage celebrity and the behind-the-mask person. Indeed, at
the aforementioned photo event, the fans were to pick one out of the three girls who may or may
not be Haga Yui to take a picture with. It could be said that what enlivened Haga Yui was not the
mystery of the mask but the playfulness in the myth of it. While the mystery invited attention,
the highlight of the show was not meant to be, nor was expected to be, the moment of grand
reveal of the hidden talent, but the masked performance itself (c.f., The Masked Singer, Fox;
originally adapted from The King of Mask Singer, MBC). And as it was the mask that mattered,
her mask was readily lent to whoever could best help create her image, whether they were a
voice actor with the cutest voice or a model who could most jubilantly sport a ponytail.
Another key aspect that marks Haga Yui as an important early example is that it was not
a creator-driven, unidirectional project. While Ijuuin took charge, his radio listeners were core
members of the project from the ideation stage to the project’s closure. They were not simply
fans who had the extra privilege of being privy to Haga Yui’s secrets, but also co-producers of
those secrets. An essay by a heavy listener who was involved in the project (pureyādo, 2011)
recalled that as the project began to accrue attention, the fans were starting to “realize that the
character born by coincidence was increasingly becoming bigger,” and how “this was becoming
a movement while they were feeling like an accomplice to it.” The writer stressed that “the
biggest weapon” of the project was “the imagination of listeners and fans,” arguing that the
process of creating their own version of Haga Yui by picturing her future steps had touched on
the joyful essence of idols’ existence.
Indeed, Haga Yui would not have been able to be conceived without pre-existing
Japanese idol culture. She was born from Ijuuin and his listeners’ mutual understanding of 80’s
Japanese idol culture and was posed as a being that could convincingly integrate into the existing
34
mix of equally youthful, bubbly teen girls. The project was bold but therefore not entirely new in
terms of who Haga Yui were to be in the context of Japanese entertainment media. She even had
an (fictional) exclusive scandal photo feature in a gossip magazine, following the footsteps of
fellow idols. Said to be born on April 15, 1974, Haga Yui is now supposed to be nearing 50,
whether she decided to stay in Taiwan or relocate elsewhere. However, much like how many of
the cheery, innocent teen idols are pressured to be, her early retirement means that Haga Yui the
idol can forever remain young even without the help of elaborate CGI techniques that
characterize contemporary virtual influencers.
Taking inspiration from Haga Yui, major Japanese talent agency HoriPro introduced the
first fully 3D CGI virtual idol Date
4
Kyoko DK-96 in 1996 as part of their Digital Kids (“DK”)
project. The DK project was their attempt to expand into the untapped virtual celebrity market
with CGI humans that they believed would benefit from “the age of “multimedia”” (HoriPro
Inc., 1997; Nikkan Saizō, 2008). HoriPro gathered together not only their talent production
experts who were overseeing tasks such as artist management, promotion, and TV production,
but also computer programmers from their Visual Science Laboratory to actualize their sketch of
a 162 cm (5’4’’; “still growing!”), 43kg (95 lbs) sweet-toothed 16-year old girl with an 82-56-82
cm (32-22-32 inch) bust-waist-hip ratio (HoriPro Inc., 1997; macbain, 1999). Date Kyoko drew
on various state-of-art graphical technologies at the time, including motion capture techniques
for both facial and bodily movement so that she could eventually appear live on TV. In fact, she
was a shocking technological feat at that time: a video production director at HoriPro recalled
4
Pronounced Da-te.
35
that foreign technicians believed HoriPro to be a computer-related venture company instead of a
talent agency due to her innovativeness (Nikkan Saizō, 2008).
Date Kyoko traversed across different plains in virtual and physical worlds. In the
promotional music video of her debut single LOVE COMMUNICATION, Date Kyoko was
shown to sing in a physical studio. She also walked among pedestrians and idly hung around at
what seems to be an American city (e.g., 1:48 Trump Tower, 3:12 American flags on a building;
Frank, 2007). Intermittently, the video had her dance actively with her arms and legs
energetically flinging against the backdrop ranging from various footages of cityscape at day and
night to a white, empty computer graphical space. These scenes boasted both her technical
capacities as a CGI-created human and related market appeals, namely her adaptability to diverse
spatial, linguistic, and temporal contexts. In fact, in 1999, she re-debuted in Korea with the new
name “DiKi.” This time, she sang in Korean in her music video Albatross but showed off the
exact same dance moves from her previous video, although this time in a number of different
digital spaces and by a beach pier. Diki was not successful in Korea. However, HoriPro’s attempt
at applying photorealistic graphics to their talent pipeline still marks Date Kyoko as an important
predecessor to virtual influencers, many of whom also stress similar comparative benefits to
human influencers, namely regarding their adaptability. In this respect, as HoriPro claimed, she
was “the beginning or “sample test” of the new genre of media personalities, “Virtual Idol””
(HoriPro Inc., 1997).
Despite the interest she accrued as the first 3D virtual idol in Japan, Date Kyoko was not
a commercial success, leaving a trail of other not so successful Digital Kids after the initial DK-
96 version. There were several difficulties. To begin with, her appearance was prohibitively
estimated at several million yen (1 million yen = ~9,000 dollars) per appearance on a show,
36
which was not an acceptable amount at the time especially for a fresh star-to-be (Nikkan Saizō,
2008). Moreover, unlike Ijuuin’s Haga Yui project, she was perceived as too uncanny-looking
and company-driven without much room for fans to participate (Sone, 2017). While she had
gained some fans (Considine, 1996), these reasons contributed to her ultimate failure in reaching
the relational intimacy and interactivity that her official interview proclaimed to ground her
existence in (HoriPro Inc., 1997):
“Anyone who thinks that I am “non-existent” because I am a virtual idol may believe that
I exist only in the personal computer or in the world of computer graphics. But those who
think that Kyoko Date is someone they can relate to may believe that I am like a pen pal
people used to have when it was cool to have pen pals. Kind of like a creature living in
people’s hearts.”
While Date Kyoko never personally achieved the level of fame some of the contemporary
virtual celebrities are known for, she has been a figure that continues to live along the timeline of
virtual celebrities’ development. Date Kyoko’s DK-96 version went through a series of revamps
over the decades, such as DK-97 where she had a new voice actor and DK-2001 where she
adorned newly designed blond space buns, well-ahead of the currently most famous virtual
influencer Lil Miquela who is characterized by her space buns. DK-2001 also appeared in the
virtual world Second Life in 2007 as a non-playable character (NPC) for an event (“Date
Kyōko,” 2022; Nikkan Saizō, 2008). Most recently, her name resurfaced around 2019 with the
introduction of Date Ayano, a self-proclaimed daughter of now 40-year old Date Kyoko and an
aspiring v-tuber—a shorthand for virtual YouTuber which refers to a strand of virtual streamers
that are amassing great traction especially in some East Asian regions, typically stylized after 2D
Japanese anime aesthetics (Nakabayashi, 2019).
37
Terai Yuki, originally a character from Kutsugi Kenichi’s manga Libido before she was
redesigned in 3D independently from the manga by the artist in 1998, is another noteworthy case
from 1990s Japan. Virtual idol Terai Yuki is a 166cm (5’5’’) tall 17-year old girl with a bust-
waist-hip ratio of 86-59-85 cm (34-23-34 inch) who supposedly time-traveled from 2017
(Galbraith, 2009; “Terai Yuki,” 2021). While Terai Yuki shared similar characteristics with
preceding virtual and non-virtual idols, such as her cute smile, young age, chirpy voice, and the
superfluous details on her body size, there are four factors that distinguishes her in the timeline.
First, she was a transmedia crossover case. She was not first created to be in the idol
entertainment business like Haga Yui or Date Kyoko, although Terai Yuki the virtual idol was
not an exact copy of Libido’s Terai Yuki who appeared as an older sex worker in the manga.
This leads to the second factor: she was independently produced as a 3D art hobby project,
instead of starting out as a collective project, either playful (Haga Yui) or commercially-driven
(Date Kyoko). Third, these factors resulted in her unique 3D design that did not strive for
complete photorealism but rather one with a hint of manga-style look with big eyes and sharp
nose that modeled after Libido’s Terai Yuki (for image comparisons, see Gallery in Kutsugi,
n.d.). This design was considered as a unique feat that was made possible not only via Kutsugi’s
technical skills but also his artistic eyes as a manga artist, something that was perceived to set her
apart from other 3D CGI female characters that emerged around this time (“Terai Yuki,” 2021).
Fourth, therefore, her fame was tied to user interests in 3D CGI home software. Her first big
break came after being well-received in the design software Shade’s online community, which
led to her being hired as the model for the software’s commercial. Her idol endeavors between
late 90s to early 2000s included more conventional media activities such as photo books, TV,
music, and game, but she was also sold as a 3D model data package for the software Shade in
38
1998 and later for another design software called Poser in 2005. There have been additional
efforts by independent 3D artists to import her to a different software, such as Daz 3D, one of the
current most popular 3D design software (e.g., will2power, 2018). Each of these factors—
transmedia crossover, independent artist, cartoon-like aesthetic, and at-home 3D software—hint
at the diverse paths many contemporary virtual stars-to-be have been taking, away from the
conventional path paved by top-down human talent production companies.
In the late 90s, Japan’s neighboring country South Korea also began to test the waters
with their first “cyber singer” Adam who debuted in 1998 with his music video Love that Does
Not Exist in the World (sesange eomneun sarang) from his album Genesis, becoming a cultural
sensation (Y. Jung, 2019; D. Kang, 1999; S. Park, 2016). Unlike the Japanese virtual idols, he
was developed by a venture software company called Adam Soft and was set as a 20-year old
adult male virtual human. His lore was that he traveled from a cyber world called EDEN
(Eastern Digital Entertainment Network) to be together with a human woman he loves.
Accordingly, he is seen on his first music video tearfully dreaming of how they can be together,
unfortunately only able to see her from afar: “Although we can’t be together, from where I can
see you I will protect you from the world / I can only love you from my heart but I will, love that
does not exist in the world.” He additionally poured his heart out in the form of poems and diary
entries in the cyber world, on his official website (Adamsoft, n.d.). However, it was not only his
carefully crafted heartsick sensibility nor his “ideal” look that combined “Eastern” and
“Western” facial features which supposedly appealed to Korean teens (Hong, 1998) that left an
impression on Korean people. It was also the interest in unresolved mysteries around him that
matched his fame as an appealing singer.
39
Adam’s quick climb to the stardom befit his title as a cultural sensation, but his equally
sudden disappearance was what made him further memorable. Since his debut, he accrued great
interest from various media sectors as the perfect star for “one source multi use” as an adaptable
virtual human (Jeong, 2006). This included attention and requests from diverse sectors, such as
reports and interviews on various mass media, a TV commercial casting for LG’s soft drink
Lemonia, a sponsorship from a Japanese fashion brand NICE CLAUP, and even an honorary
admission as a computer science student at KAIST, a highly prestigious science and technology-
focused university in Korea (Jeong, 2006; Jo, 2016). However, Adam soon began to disappear
from the screen after a lukewarm response to his second album EXODUS, only leaving a trail of
teasing rumors that ranged from having to do mandatory service in the Korean army to
unexpectedly dying from a computer virus (Jeong, 2006; Jo, 2016). According to Jung Duk-
Hyun
5
, a popular culture critic and the former public relations director of Adam who additionally
oversaw his story and character development, the core reason for his disappearance was Adam’s
production labor and cost, similar to Date Kyoko (Jeong, 2006): “In order to move Adam’s
mouth a few times, we needed several days of CG work. And of course that much money, too.
TV appearance requests flowed in, but Adam had to gradually shut his mouth.”
However, there was another mystery around Adam that makes him an even more special
mirror to reflect on: Park Sung Chul, the hidden singer behind Adam. After many years of being
faceless, Park finally received the long overdue public attention when he appeared as a surprise
on a Korean TV program called Two Yoo Project Sugarman Season 2 (2018), a show that invites
somewhat forgotten singers of popular Korean songs from the past. He finally revealed on TV
5
Korean names start with the last name. First names typically consist of two syllables which when written in
English may be hyphenated, separated with a space, or connected without a space.
40
that he was the voice of Adam, as well as the person behind Adam’s motion-captured
movements (Jeon, 2018). Park had been an aspiring artist even before he got the role of Adam.
He was a member of the band called Issue before he went to serve in the army, which was where
he auditioned for Adam. He recalled Adam Soft rushing to establish Adam as the first cyber
singer before other venture firms beat them to it (adam&zero TV, 2021). While Adam Soft
prepared the grounds for Adam’s big break, Jung, the former public relations director,
commented that he firmly believes that it was Park’s singing skills that brought fame to Adam
despite the company’s lack in technology and experience (JTBC Entertainment, 2018). However,
due to his Non-Disclosure Agreement (NDA), Park was not able to reveal his identity until the
Adam project ended (Jeong, 2006; adam&zero TV, 2021). In fact, he was actively discouraged
from pursuing a musical career for his own name. The then-manager of Adam even dissuaded
him by saying that if Park comes out, it will be “the death of both Adam and him” because Park
did not have the looks to succeed in the industry (Jeong, 2006). Perhaps a trick of his fate, Park
was later able to break into the entertainment industry as himself but with the pseudonym Zero, a
singer who sang the soundtracks for a K-drama called Beautiful Days (2001) where a faceless
online singer of the same name appeared (S. Kang, 2001). Yet, since this drama was a hit in
Japan rather than in Korea, it was in Japan that he established his singing career, although with
the name Zero not Park Sung Chul.
Park’s career on one hand reflects the face-centric culture of Korean music industry at the
time where talented voices were to stay behind the curtain for beautiful faces to rake in the
money (Jeong, 2006; D.-W. Kim, 2015), but on the other evokes the important question of
“who” the virtual human is and where the credit should lie. To add to the puzzle, while Park was
the singing voice, Jung, as the public relations director who oversaw Adam’s persona and story
41
development, was the writing voice (JTBC Entertainment, 2018). Then, there have been guesses
that the real model of Adam must have been the Korean actor Won Bin due to their visual
likeness (dongadatkeom yeonyenyuseutim, 2013). This unreferenced fragmentation of labor in
the cyber singer Adam’s stardom can be described as “virtual labor,” which Stahl (2011)
described as “performative labor that appears [sic] to be performed by an individual but that is
actually the result of division of labor incorporating creative and technical workers, intellectual
property, and high-tech equipment.” This is what marks many of the virtual celebrities different
from self-masked artists like Daft Punk (1993-2021)—the helmet-headed house music duo Guy-
Manuel de Homen-Christo and Thomas Bangalter—or Marshmello (2015-present)—a giant
marshmallow-shaped headgear-wearing DJ Christopher Comstock, whose identities and related
labor tend to be more transparent and accredited. Perhaps not every collaborator behind a virtual
star project wishes to be identified, and for some the merit may be in the relatively anonymous
collective process. However, to others, especially those who have been striving towards their
own fame like Park, the question of “to disclose or not to disclose” may be synonymous to “to
die or not to die” in terms of establishing their career. While Jung added that Park actually had a
very appealing, handsome face (Jeong, 2006), the industry’s attention was on the carefully
crafted, most “ideal” face that could best appeal to teens—the face of the “cyber star who is free
from concerns on scandals or missing shows” (Hong, 1998).
While Adam is the most well-remembered virtual celebrity of the time in Korea, there
were other cyber singers and virtual human projects that followed Adam’s footsteps, both in
terms of appearance and disappearance. Lusia [pronounced Ryu-Sia] is known as the female
version of Adam in Korea. She was created by Hyundae Informatics in 1998 “as a pure-looking
girl in the cyberspace” (H-information, n.d.) and debuted the same year with her album The
42
Reason I Came to This World (naega sesange on iyu). Unlike other virtual celebrities, she
emphasized her digital origin rather than trying to completely assimilate: on her official profile
(H-information, n.d.), her height is stated as 18,527 polygon, weight as 1.63mb (“doesn’t fit a
single floppy disk”), and favorite food as stable 0.2Kwh electricity (c.f., Adam’s profile stated
that his blood type is O, Adamsoft, n.d.). While Lusia attracted some attention from sponsors and
media (Maeilsinmun, 1998), she ended up switching to books rather than music due to low sales
record (S. Kim, 2000). Another notable point is that when she reappeared with her second album
AD2015, she completely transformed. When she sang The Reason I Came to This World in 1998,
she was a cozy sweater-wearing “pure-looking girl” who sang “like coincidence like fate we’ve
never met / but I love you more than anyone / I can’t touch you I can’t hold you / but I’m around
you.” But in 1999, she became a nipple-revealing bikini-clad sexy space warrior who sang “I
might not have the courage yet to think of you as a man / please wait for me you are just a friend
to me” (Love Feeling) against the backdrop where robot laser gun shooters and space explosions
busily cross-fade—a video adaptation of Lusia’s novel CYPIA (H-information, n.d.; D. Kang,
1999). It could be said that she was one of the first virtual celebrities that showed that
hypersexualized transition is one of the viable uses of these new “one source multi use” cyber
singers, regardless of who they may have first come across as. After all, as a 1998 column (E.
Lee) on the phenomenon said:
“Adam and Lusia are artificial beings that were crafted meticulously for commercial
purposes….They will attempt simultaneous public appearances and never be let down by
physical exhaustion like humans nor ever complain about their manager’s (owner’s)
control….the best “commodities”.”
43
The idea of commodity can be also applied to the Korean-adapted version of the Japanese
virtual idol Date Kyoko DK-96, DiKi, who debuted in Korea with her song Albatross in 1999.
DiKi’s attempt at the Korean market is meaningful as it was the first case that enacted on the
cross-cultural potential of virtual stars. Still, it also can be interpreted as an extremely
investment-centric move that was somewhat disconnected from audiences. A fan-made web
archive of Date Kyoko mused: “What does someone do with a failed virtual performer? they get
sold to anyone how [sic] want her” (TU Kyoko Date DK96 Archive, n.d.). While language was
not lost in translation, some translation between cyber singer culture and existing entertainment
fan culture was possibly lost. Despite jovially lip-syncing in Korean about how “this is a practice
to fly high” and that she “can be confident with you” while showing off the dance moves that
had previously shocked Japan, Diki did not fly in Korea. DiKi’s official website shared her fan
club president’s view on the failure. They believed that cyber singers generally attracted
attention from computer graphic-related professionals instead of their intended market—i.e.,
teens, also commenting that in DiKi’s case there was an additional hurdle because she had been
turned down from major Korean broadcasters because of her Japanese origin (HoriPro &
indecom, 2003).
Perhaps it was mostly the former, i.e., failure to maintain fans. Korea’s third domestic
cyber singer CYDA is also barely remembered, despite having strategically targeted girl fans of
female artists with her self-conscious yet confident debut song about her looks and popularity
that was titled I Don’t Like the Truth (1998), also interpretable as “I don’t like Jin-sil,” a popular
female actor at the time (D.-W. Kim, 2015). In fact, cyber singers’ underwhelming connection
with fans was recognized even by Adam. After they folded the project, Adam Soft’s CEO said
that among many reasons including production costs, human labor, and limited capacity for
44
offline promotions, “the drawback of cyber characters is that, unlike traditional singers, it is
difficult for them to maintain continuous relationships with their fans” (B. Lee, 2002).
Despite not being successful, what was notable about these cyber singers who appeared at
the cusp of the new millennium in comparison to those from early to mid-90s is their attempts at
fan-star and fan-fan interactions, possibly to combat the relational limitation that Adam Soft’s
CEO mentioned (B. Lee, 2002). For instance, unlike Date Kyoko’s 1996 Japanese website for
her DK-96 version (HoriPro Inc., 1997), Diki’s revamped website from 2001 (HoriPro &
indecom, 2003) had a fan board and a chat room menu. Similarly, Lusia’s website had a
WebBoard menu where she personally responded to some of the fans’ messages (H-information,
n.d.) and CYDA’s had a chat room with comic-strip images (Yesnet Co.,LTD., 2002). An even
more intuitive example would be Adam Soft’s virtual baby project Hana, who was to be born at
00:00 am on January 1st, 2000. Unlike Adam’s website where there was no place for interaction
(Adamsoft, n.d.), the idea of Hana was that as more people expressed their wish for peace,
environment, and Korea’s reunification by clicking on the main page of his website—with the
limits of one click per person per day and up to 10,000 participants, he would grow up faster to
start working as an ambassador representing those values as soon as the clicks hit the number
30,000 (Adamsoft, 2000; H. Song, 1999). His website visitors could also leave encouraging
messages on his message board, as well as participate in a donation drive event by leaving
messages on bulletin boards dedicated to missions for social good. Unfortunately, despite this
effort to encourage participation, Hana never got to fully grow into “a kid full of star potential
and artistic potential” (Adamsoft, 2000). Among many reasons, it was perhaps the rather top-
down way the company expected Hana that led to the indefinite delay, which contrasts with the
way Ijuuin’s listeners gathered together for Haga Yui.
45
Ironically starting off with a parody project of calculated entertainment culture, 1990s
Japanese virtual idols’ and Korean cyber singers’ births tended to be loudly celebrated with
technocapitalistic dreams of controlled perfection, some of the narratives of which are repeated
in those of contemporary virtual influencers. Regardless of their grand appearances, however,
their death was a silent disappearance, buried amid the chasms in technological realities and lost
fan cultural translations that flashy promises glossed over. As the Korean indie artist Nahzam
Sue poignantly reminisced in his 2016 retro-funk song titled Cyberstar Adam, the birth and
demise of 1990s virtual celebrities were entangled with the tales of cutting edge future, the same
fantasies of beauty and automation that have been told and retold over the decades:
“Born knowing when he will die / Beautiful cyber singer Adam / Born from the hands of
irresponsible / I was born then will die
In the sloppy shading / that compose me / I, like crazy, / cry and laugh / regardless
tomorrow / I will disappear
Somehow pushed to stand at the cutting edge / marvelous cyber singer Adam / when a
new sharp edge appears / will get buried before taking a step
In the handful of polygons / that compose me / I, like crazy, / cry and laugh / regardless
tomorrow / I will disappear
[English] Just type “Hello world” / Just type “Hello world” / Just type “Hello world” /
Please type into main server
[English] Take an automatic response / Take an automatic response / I got a word to say
[English] I don't think / It's not a sarcasm / It's a karma of / Cyberstar Adam
46
Drawing the Idea called beauty / moved forward unfashionable technology /do not be sad
your ancestors / disappeared from fake gimmick
In the fantasy about the future / that compose them / all, like crazy, / cry and laugh /
entangled so no more / can be identified
[English] I'm a Cyberstar / Like so many others / I'm a Cyberstar / Like so many others /
I'm a Cyberstar / There's no way to revive / I'm a Cyberstar / No reason to survive”
Virtual Bands: Pre-90s Fictional Characters to Gorillaz
While the above recounts the timeline of earliest human look alike predecessors to virtual
influencers, there were more viscerally “non-human” celebrities before and during their
existence: fictional bands and characters. Less constrained by production costs and the uncanny
feeling from near-but-not-quite human looks (Mori, 1970), there have been several animation
and story characters that demonstrated the potentials of the concept well before 3D CGI entered
the scene. At first, this was mostly in the form of being heavily combined with or branching out
from a fictional story world, rather than being just focused on establishing a musical career. That
is, unlike the Japanese and Korean cases in the 1990s, they were not produced to be substitutable
equivalents of human artists in the star production pipeline per se, but rather as those reliant on
or belonging to a fictional content world.
One of the earliest and most well-known cartoon musicians in North America is Alvin
and the Chipmunks. They were first created in 1958 by Ross Bagdasarian under the pen name
David Seville, who after his commercial success with a sped-up, high-pitched song introduced
three chipmunk characters Alvin, Simon, and Theodore to represent the altered voices (Conner,
47
2016). Alvin and the Chipmunks began to formally adopt their characteristic 2D cartoon faces
and human accessories after their 1961 TV show The Alvin Show (CBS) (Conner, 2016).
The Archies, a fictional band belonging in the universe of the US comic series about high
school students’ lives called Archie Comics (1939-), is another well-known fictional musical
star. The idea was driven by Don Kirshner, who had worked as the music supervisor on an NBC
TV comedy series called The Monkees (1966-1968) which is about an aspiring rock band of the
same name. Despite the show’s search for men who could “authentically” be members of the
band, Kirshner’s approach was rooted in a business mogul-like mindset where he viewed the
process of creating a hit song as akin to building and then leasing a real estate (Stahl, 2011). The
band was a hit under Kirshner’s direction but The Monkees, which included professional actors
rather than only having “authentic” musicians and musicians-to-be, were increasingly frustrated
at the irony and the “dishonesty” of pretending to perform the songs recorded by skilled
musicians behind the scenes (Stahl, 2011). The band did not survive long after attempting to
redeem their authenticity by firing Kirshner, who on the other hand found success after applying
the same model and storyline to the animated adaptation of the Archie Comics The Archie Show
(CBS, 1968-1969). This was when the series first introduced the cartoon band The Archies.
Their first song Sugar, Sugar (1969) was a hit with a sales record of ten million. However, much
like the virtual labor involved in Korea’s cyber singer Adam, The Archie’s behind-the-scenes
singer Ron Dante never got to perform as himself as per the rights, leaving the teen-aged 2D
cartoon characters as the faces of the band. Later, Archie Comics applied the idea to their
accompanying series on an all-girl rock band Josie and the Pussycats (1963-), which debuted as
a fictional cartoon band of the same name via animation (1970) and then subsequently as a real-
life band with human musician counterparts (1970).
48
In fact, the chronology of virtual celebrities can stretch extremely long and broad if all
stardom resulting from fictional media were to be included, because humans have been capable
of building a one-sided sense of relationship with media figures. Known as “parasocial
relationship” in social science research (Horton & Wohl, 1956), people have been able to find
deep, meaningful connections with fictional characters regardless of their form and existential
state. Therefore, if the criterion is solely based on their fictional character and their fans’
perceived intimacy with the star, any character can be eligible for the timeline. For instance, even
if we limit it to fictional characters with a star artist role, the list would need to include not only
content with prominent star characters like the US cartoon Jem and the Holograms (1985-1988,
Hasbro/Marvel), Japanese animation Macross Plus (1994, Triangle Staff), and SF novel Idoru
(1996, William Gibson) but also those with supporting star characters such as the manga Sailor
Moon (1991 - 1997) (for a sample list of virtual bands with non-human members, see Appendix
A in Stark, 2018). Similarly, the character’s digital origin would also result in a great long tail of
eligible predecessors, although not necessarily inappropriate. For example, in the field of games
and in the 1990s alone, this should include 2D anime-style characters from Japanese hit dating
simulator games such as Dokyusei (1992) and Tokimeki Memorial (1994). Another important
example from games would be the 3D character Lara Croft from the US game Tomb Raider
(1996), whose cultural significance in North America has earned her the title “the first digital
star” (Flanagan, 1999), which, of course according to this dissertation’s timeline is a debatable
statement especially in the global context.
What makes the UK-born virtual band Gorillaz significant amid such expansiveness are
their critical acclaim and matching lasting popularity that no other virtual musician had
previously reached. First created in 1998 by the rock band Blur’s lead singer Damon Albarn and
49
cartoonist Jamie Hewlett, the UK-born virtual band’s four primary comic-style fictional
members 2-D, Murdoc Niccals, Russel Hobbs, and Noodle have an impressive roster of
achievements from across the globe. This includes wins and nominations against various human
artists from prominent awards such as Grammy Awards, Brit Awards, and MTV Europe Music
Awards. Moreover, since their debut in 2001 to their most recent album in 2020, they have been
steadily finding themselves in popular music charts and headline performances in famous music
festivals. Their success was impressive and arguably an unprecedented one, even prompting a
new category of “Most Successful Virtual Band” in the Guinness Book of World Records (see
Conner, 2016).
Albarn and Hewlett have been very vocal about the fictional origin of Gorillaz,
comparing them to the hyper-manufactured nature of the existing star system. In their 2005
interview with Wired (WIRED Staff, 2005), they said that their inspiration was from the existing
fiction in popular music culture. Hewlett mentioned that they thought a cartoon band could be a
commentary on how “there’s nothing of substance” on MTV, and Albarn agreed by saying that
“it’s all a bit like a cartoon,” referring to how the stars of their generation came from likely
scripted shows. They proceeded to emphasize their desire to directly communicate through their
art by having the virtual members as the musical representatives rather than having to partake in
the celebrity scene themselves, commenting that “the whole point of being a rock star” is
“pretend[ing] to be somebody you are not.” This resonates with the spirit of Haga Yui project,
however with an additional twist due to Albarn’s notable star status as the frontman of Blur, the
epoch-making band alongside Oasis that is known to have popularized the genre of Britpop in
the UK. As Richardson (2005) commented, this has created a paradox: “Hinging on the notion of
the virtual as a conduit for parodic artistic expression, the subversive intent of the band’s evasive
50
strategies was predestined to be compromised by the singer’s iconic status within British popular
culture” (p. 3).
What adds further complexity to Gorillaz’s paradoxical stardom is their live shows in
which various stage art technologies have been incorporated to give a semblance of having the
virtual members on stage. This enhanced technological capacity marks the band different from
1990s Japanese virtual idols and Korean cyber singers, who tended to perform live through a
human counterpart or only through a video playback. Gorillaz’s attempts have included flat
screen 2D projections in the earlier days to 3D holographic performances, such as that at the
Glastonbury music festival in 2010 (Conner, 2016; Jones et al., 2015). Combined with the
creators’ stance towards the popular music scene, these efforts have been evaluated as
oppositional, however with a heavy dependence on dominant codes: “the band’s wish to resist
commodification in pop by refusing access to the real (the real musicians) while exaggerating
traits conventionally associated with pop stardom (the spectacle of performance)” (Richardson,
2005, pp. 2-3). This tension existed consistently. For instance, after announcing Gorillaz’s plans
to tour as 3D holograms, the band’s spokesperson Chris Morrison commented that the crossover
between holograms and humans is “where it gets spooky,” puffing the premise of spectacle while
alluding to the symbolism: “they are the ultimate manufactured band, so what better way to rebel
against celebrity culture?” (Simpson, 2005)
Despite the intended satire, Gorillaz’s live performances progressively expanded their
incorporation of human performers including guest stars, from placing them hidden behind the
screen in their 2001-2002 performances to putting them fully on stage with occasional darker
lighting during their 2010 tour (Conner, 2016). This in part can be interpreted as a difficult-to-
avoid residue of a subversive experiment that drew on mainstream commercial codes. Their
51
parody became a self-fulfilling prophecy unlike the Haga Yui project, which was shut off once it
gained mainstream traction. What amplified this was that Albarn and Hewlett have outspokenly
positioned themselves as the mastermind and the core creative force behind the band. This,
particularly in combination with Albarn’s personal rockstar fame, have imbued the band with a
strong sense of ironic auteurship despite the fiction of agentic virtual band members. For
instance, in the explicit mock interview enclosed with Gorillaz’s DVD and photographic
materials, the two are depicted to have been beaten up by the virtual members. The narrative
suggests the virtual members’ dominance over the two creators but not without attributing a
creator identity to them, which had been interpreted as an encouragement of distanced but
unquestionably and singularly present authorial presence (Richardson, 2005). Such an emphasis
on the two creators result in a self-parodying layer of anonymity of virtual labor (Stahl, 2011)
where other human collaborators become positioned as interchangeable (Richardson, 2005).
Beyond auteurship, this gradual inclusion of humans had also been attributed to the
virtual context that Gorillaz was presented in (Conner, 2016). As they were introduced within a
virtual world not physical, live performances required means to ease the audiences’ entry into
their virtual world or their acceptance of the crossover. Consequently, much like how the band’s
attempt at disrupting the conventional codes in pop music culture has resulted in a reproductive
cycle of reliance, the hierarchy between the physical and the virtual has also been reaffirmed in
the effort to upset it. The band’s much hyped first 3D performances demonstrated these tensions.
At the 2006 Grammy Awards, the virtual members of Gorillaz performed their song Feel Good
Inc in 3D animated bodies. They were fixed to their respective spots but were convincingly
“there” in the black screens that mixed well with the darkened stage. After a minute and half into
this pre-recorded performance—audio of which had only been available to TV audiences and via
52
earpieces for human performers due to stage vibration issues (Conner, 2016)—they were joined
by three flesh-and-blood human rappers from De La Soul who featured on the song. Unlike the
virtual members, these “replaceable” guests (see Richardson, 2005) energetically roamed about
the extended stage area, spitting into their microphones. Meanwhile, virtual member 2-D idly
checked his phone while bobbing his head to the music and Russel dozed off, perhaps a
reflection of what the symbolically ever-present auteurs were capable of doing at the time. After
De La Soul exited, Gorillaz were subsequently joined by Madonna who first appeared in a highly
photorealistic holographic form, twirling around the members in the black screens but without
any excessive dance moves. This was until her and her dancers’ physical bodies took over the
now fully brightened stage with a sweat-inducing, muscle-accentuated choreography to her
colorful, disco-influenced song Hung Up, banishing the virtual members back into the dark of
the screens. While extensive efforts allowed a short meeting between the virtual world of
Gorillaz and the physical world where the audiences were (albeit without full audio), liveliness
and mobility were reserved for those belonging in the physical world.
2000s: The Rise of Hatsune Miku
The 2000s saw a number of new attempts at gaining virtual stardom, with Hatsune Miku
as the most widely recognized case. Some of the lesser known aspiring virtual celebrities
included Intz [debut: 2000] and Limit [2002] in the case of cyber singer-dancers (D.-W. Kim,
2015), Dethklok [2006] in the case of cartoon universe-conjoined virtual band (Conner, 2016),
and the elaborately rendered 3D virtual actor Aki Ross in Final Fantasy: The Spirits Within
(2001) in terms of CGI characters in a fictional content (Aldred, 2011). Among them, virtual idol
Hatsune Miku stands out, an often-referenced inspiration for many contemporary virtual
influencers. This is not simply because of Hatsune Miku’s impressive global stardom, whose live
53
holographic gigs have spanned from guest appearances for human stars like Lady Gaga to solo
global tours and bookings in renowned music festivals such as Coachella and Summer Sonic
Festival. This is also because of her stardom’s inextricable relationship to her devoted fandom
and her Japanese anime-style design, the latter of which can be easily connected to the recently
rising branch of virtual celebrities called v-tubers who tend to use a similar art style.
Hastune Miku, now known as the world-famous virtual teen girl idol with long, aqua-
colored high pigtails, did not start out intending to be the mega-hit global virtual celebrity she is
now when she appeared in Japan in 2007. Instead, Crypton Future Media released Hastune Miku
as a “vocaloid” software through which its user could create a song using the software’s cute,
feminine synthesized voice, processed from the recordings by Saki Fujita, a Japanese voice actor.
Similar to the company’s previous softwares Meiko and Kaito, the accompanying cartoon image
of Hastune Miku had been a means to encourage reluctant users to picture a human when
creating music with the software (Sone, 2017), but not necessarily in a way that referenced the
looks of the voice behind-the-scenes Saki Fujita or a particular music genre, unlike with the
design of Gorillaz (Zaborowski, 2016). In fact, her design has been evaluated as more
advantageous than that of Gorillaz in facilitating immersion on the grounds that her hyper-
digitalized origin may allow much easier forgoing of the consideration for tangibility (Milton,
2017). The first and the most famous out of the company’s “Character Vocal Series,” Hastune
Miku also took a step further than her previous versions by including a detailed profile like
previous virtual idols in Japan, such as her 16 year old age, 158 cm (5’2’’) height, and 42kg
(93lbs) weight (Crypton Future Media, n.d.).
Hatsune Miku’s carefully crafted voice and design may have earned her some fame, but
what is understood as the critical drive behind her stardom is Crypton Future Media’s proactive
54
encouragement of organic user participation (Sone, 2017; Zaborowski, 2016). In line with the
general tolerance towards derivative content within the manga and animation “otaku” (Japanese
fan subculture) culture that they located her main audience in, the company aimed to not only
allow but promote fans’ playful creative engagement with Hastune Miku (Sone, 2017). This
resulted in a unique licensing model that they called “PiaPro (Peer Production) Character
License,” with which non-commercial users could actively create, share, and collaborate using
her voice and images without any cost as long as it is not for commercial purposes, in which case
the content would be bound to the company’s terms (Zaborowski, 2016). This encouraged active
creation of user-generated content, not only on the company-created website PiaPro but also
other user-centric platforms like Nico Nico Douga—i.e., the Japanese equivalent of YouTube
with the unique function for the viewers to add comments directly onto the video while watching
it, which afterwards get incorporated into the future viewers’ viewing (Sone, 2017). Such a range
of venues allowed Hatsune Miku to be open to not only those with technical and artistic skills to
create a standalone secondary content, but also to those who may feel more aligned with the
viewer-contributor role.
Openness in participation is connected with the capacity for control and authorship,
which can appeal to fans. An ethnographic study of vocaloid audiences (Zaborowski, 2016)
reported that despite not being personally equipped with the capacity to create a derivative
creative content nor any intent to do so, many fans felt that the bottom-up crowdsourcing
element was one of the core sources of Hatsune Miku’s popularity and her fans’ pride. Against
the backdrop of the existing culture of manufactured human idols, Hatsune Miku’s evidently
digital origin and pliability by the audiences were taken as sources of her higher level of
authenticity: “The audiences make her real” (Zaborowski, 2016, p. 124). Media scholar Daniel
55
Black (2012) also similarly argued in his chapter on Japanese virtual idols: “An idol which exists
only as digital data which the fan can “own” and manipulate holds the promise of a more
intimate relationship than is possible with a living idol” (p. 220). Consequently, virtual idols are
more dependent on and subject to fans’ control. An exemplar artifact of this dynamic is a fan-
created software application called MikuMikuDance (MMD). Introduced in 2008, MMD is an
animation tool with which its users could enliven their imagination by making their own Hatsune
Miku singing and dancing performances (D. Black, 2012; Sone, 2017). MMD was also adopted
by the management; Hatsune Miku’s first “live” concert had MMD-created projections for the
physically attending fans (Sone, 2017). Another example could be the case of Akihiko Kondo’s
marriage with Hastune Miku. According to the Hatsune Miku fan Kondo, the marriage has been
full of genuine love and healthy support (Obuno, 2022). Unfortunately, it is also a relationship
that we can never hear about from the virtual idol’s mouth, of course unless pre-programmed.
Nevertheless, Hatsune Miku has been a case that purposefully and successfully benefited from
the fan co-ownership model, surpassing the company-initiated fan engagement efforts in the late
90s.
Hastune Miku’s clear diversion from photorealism, both on screen and in holographic
forms, is also notable. She did not strive for human-likeness, but rather comfortably boasted her
traits like sparkly large eyes, tiny shadow of a nose, and shin-length flowy cyan pigtails. These
characteristics can be easily associated with Japanese manga and anime art style, which allowed
her to escape the common criticism on the creepiness of 3D photorealistic virtual humans’ near-
but-not-quite human-like looks, namely the “uncanny valley” (Mori, 1970). Her design can be
traced to otaku culture with interest in “moe” (Sone, 2017), which refers to “a burning passion
for budding beauties, or cute girl characters,” originating from the Japanese word for “to bud
56
[moeru]” that shares its pronunciation with the word for “to burn” (Galbraith, 2012, p. 343). Like
the meaning implies, moe is an aesthetic that does not necessarily exclude sexualized
fetishization, although with a complicated relationship to the bodies in the physical world. Its
realism is one that is rooted in manga and anime, not one that strives for naturalism that perfectly
corresponds with looks or rules among humans living in the physical world; it is a fiction that is
realized and appreciated as fiction (Galbraith, 2012). Regardless, moe is not strictly reserved to
the animated world; as a visual and performative aesthetic, moe can be attributed to humans as
well. A commercialized example would be “maid cafes” in Japan where the customers can
interact with the costumed staff who play the role of moe maids. Many of these cafes had to later
add rules to prevent sexual harassment from the customers and also to prevent the staff from
engaging in un-moe behaviors that may break the fantasy, including off hours (Galbraith, 2012;
for more on maid cafes, see Galbraith, 2013). Moreover, Hatsune Miku’s moe subcultural
positioning, particularly in combination with her chosen occupation as a virtual idol, is also
proximate to the trope of cute (aspiring idol) girl(s) being nurtured to their full bloom of
potentials by a male producer-auteur figure (see Finan, 2021; c.f., age and body profiles of
aforementioned virtual idols). This parallel hints at the presence of unwritten rules in the
supposedly open fan co-ownership model. Without denying her successful evasion of the
uncanny valley and involvement with the otaku demographic, Hatsune Miku’s design also
invites scrutiny, especially more so with many V-Tubers following suit with moe-reminiscent
appearances, although partially due to technicalities of popular v-tuber design softwares. As
media scholar Sone (2017) commented, “though they remain in the realm of the imagination and
are not actualised, [otaku subcultures] can involve cartoon images of young girls who would, in
real life, be considered underage” (p. 146-147).
57
Hatsune Miku’s synthesized voice has been evaluated as further disembodied than former
virtual celebrities as it is the result of processed, fragmented data that constantly gets reformatted
to the user’s taste. It is different from the previous forms of “virtual labor” (Stahl, 2011)—i.e.,
the illusion of single person-sourced labor that in fact combines multiple alienated creative and
technical personnels as well as technological and legal artifacts. There exists no pretense to
Hatsune Miku’s technologically manufactured, and therefore collaboratively created, origin. To
the fans, this meant uninterrupted direct intimacy and authenticity, or perhaps even control and
ownership, as it implies the absence of a body and mind beyond the scope of what has been
shaped together by the company and fans (Black, 2012; Zaborowski, 2016). However, there still
are relatively hidden links in the virtual labor of Hatsune Miku. In order to build the sound
library for vocaloid softwares like Hatsune Miku, raw human voice recordings and technicians
are still required. Thus, despite the technological obscuration, there are humans and artifacts
essential to Hatsune Miku the vocaloid software’s virtual labor: “the human body remains in the
chain of commodification, as its living vocal attributes are harvested for mass-production”
(Black, 2012, p. 223). While in the case of Hatsune Miku this more closely refers to her voice in
the sense of sound, the parallel can be drawn to many other virtual celebrities’ “voice” in the
sense of messages. For instance, while the singer Park Sung Chul sang for Adam, it was the
public relations director Jung Duk-Hyun who wrote Adam’s website posts. The concept of
virtual labor can be particularly insightful for approaching contemporary virtual influencers
because many of them position themselves not simply as an independent technological being
who is outside of the conventional star system, but also as one that is capable of (eventually
gaining) full autonomy through artificial intelligence.
2010s to Now: Virtual Influencers and V-Tubers
58
Keywords that characterize virtual celebrities between 2010s and now would be
photorealism, social media, and mainstream resurgence. Aimi Eguchi is one of the last more
widely known mass media-incorporated cases before virtual humans began to primarily appear
on social media. In 2012, a top Japanese girl idol group Japan AKB48 announced the debut of a
new member named Aimi Eguchi, who later was revealed to be a virtual human created by
compositing facial features of the group’s six most prominent members as a part of their
publicity campaign with the snack brand Glico (Zaborowski, 2016). Visually highly
photorealistic, what caused fans’ suspicion was not her looks but her lack of past history in show
business considering her “old” age of sixteen (fourteen would already be too late for an aspiring
idol) and her sudden appearance in a national commercial despite being a mere trainee, although
many still defended her human existence until the big reveal (Zaborowski, 2016; Nelson, 2016).
Afterwards, the fans were introduced to an online tool where they could create their own “perfect
virtual idols” through facial features of nearly fifty selected AKB48 members (Zaborowski,
2016). In 2013, the campaign concluded with many fans feeling disappointed from the “perfect
idol” Aimi Eguchi’s “graduation” from AKB48. Aimi Eguchi is a case that once again
highlighted virtual celebrities’ reliance on the existing entertainment industry codes, being “less
of a trick and more like an extension of what AKB48 already is” (Nelson, 2016). It is also one
where commercial sponsorship by Glico openly sustained and drove the virtual celebrity’s short-
lived life, instead of the partnership being a part of the virtual idol’s independently constructed
fictional universe. Combined, the two factors demand attention to the ethics of extending
dominant cultural codes with primarily commercial interest, not critical. A former Haga Yui
project listener-participant commented in their essay on the “Aimi Eguchi controversy” that the
critical difference between the two is that with Aimi Eguchi, the fans were to solve the mystery
59
of her identity based on each new clue that the campaign dropped as opposed to collectively
fabricating the myth together (pureyādo, 2011). The author argued that encouraging the
audiences to seek out for dishonesty could paradoxically lead to the unraveling of the fiction of
idol culture itself. This was not too savory in the author’s perspective, since there are already
many human idols who are suffering from people who dig into their private lives to test their
alignment with the idol fiction, harassing them if it is not the case (pureyādo, 2011; also see
Galbraith, 2012 for rules enforced on maid cafe staffs to help sustain the moe fantasy). As a
digitally incarnated “perfect idol,” Aimi Eguchi was an ideal extension of the dominant idol
culture in Japan. Regardless of parodic intentions, it must be acknowledged that such extensions
can result in a reaffirmation of established codes because the prerequisites for them to be
communicated as parodic are an ongoing agreement on what the codes are and their replications
(c.f., Gorillaz). Moreover, it should be additionally noted that Aimi Eguchi’s life expectancy and
purpose were planned as per the span of the Glico advertising campaign, unlike Haga Yui whose
end was decided when people began to notice the gradual decline of her satiric effect. Many
virtual idols have been conceived to play a role, one which under the glitzy surface of
sensationalism and novelty still repeats existing values, if not in a concentrated form.
Technological developments for both industry and personal users led to the current era of
virtual social media influencers and cross-spatial celebrities. With the spread of user-generated
content platforms, the early 2010s began to see a variety of fictional stars who shared their life
and art. For instance on YouTube, this included fully 3D animated figures such as the
supposedly London-based Japanese v-logger Ami Yamato who started to post bilingually in
2011, 2D cartoon musicians such as Your Favorite Martian who started sharing their comedic
songs with politically incorrect lyrics in 2011 (see Stark, 2018, pp. 61-64), and existing
60
characters who adapted themselves into the new media format. An example could be Barbie
whose first 3D animated music video was posted in 2012, followed by various content ranging
from commercials to the ALS bucket challenge and v-logs. Those more aligned with the
conventional musician-star model also prevailed in 2010s, the most prominent example of which
would be the virtual K-pop group K/DA that debuted with a holographic live performance of
their song Pop/Stars in 2018, although as transmedia characters branching out of the universe of
Riot Game’s League of Legends. It was also an era where such technologies for bringing virtual
stars to life were applied to bringing human stars back to life. For instance, at the 2012 Coachella
festival, Tupac Shakur was (in)famously “resurrected” in hologram—thus the nickname “Pac-O-
Gram”—to perform together with fellow hip hop artists Snoop Dog and Dr. Dre. The
performance had accrued as many criticisms as nostalgic welcomes on the grounds of
authenticity and authorship, since the “real” Tupac could not have a decision in this (see Jones et
al., 2015). This was also the time when some of the key 3D tools began to have lower entry
barriers for non-professional individual users. For instance, DAZ 3D, one of the most famous
and easiest tools for creating photorealistic 3D virtual humans, became freemium software in
2012. Another example would be VRoid Studio, an intuitive software popularly used by
beginning v-tubers to create anime-like 3D characters, the beta version of which was released in
2018.
Most importantly, combined with the proliferation of social media and streaming
platforms, virtual beings that called themselves “virtual influencers” began to appear, many of
whom were extremely photorealistic. This includes virtual influencers such as the 3.1 million
follower-garnering Miquela who broke out on Instagram in 2016 (@lilmiquela, accessed January
29, 2022) and the self-dubbed “The World’s First Digital Supermodel” Shudu who started to
61
post in 2017 (@shudu.gram, accessed January 29, 2022). Within the boundary of the term virtual
influencer, or perhaps crossing over, also exists “v-tubers” based on streaming or video
platforms with anime-looking appearances. This ranges from the most well-known v-tuber
Kizuna Ai who began to post on YouTube in 2016 to relatively smaller v-tubers, including Date
Ayano by HoriPro who was introduced in 2019 as the daughter of the 90s virtual idol Date
Kyoko, a nod to the history of virtual influencers (Nakabayashi, 2019). There are also motion-
capture-based interactive streamers who are stylized in a more 3D video game-like, somewhat
photorealistic design than anime. Examples can include v-tubers such as CodeMiko (Youna
Kang) who began streaming in 2017 both in her 3D character persona (“CodeMiko”) and human
character persona (“The Technician”), and Ai Angel who began to post on YouTube in 2019 but
so far without having revealed her human counterpart(s).
Many contemporary virtual influencers boast that they are the first, whether in their
country or their respective field, or in pursuing a specific type of endeavor. This brief chronology
of their virtual predecessors suggests that despite their claims of novelty we have had decades-
old—or much older—looking glasses to read them. The preceding cases were entangled with
various social, technical, and cultural factors, such as existing entertainment industry codes,
technological capacities, issues of labor, identity politics, and ownership to name a few. The rest
of this chapter will focus on the present moment of virtual influencers. In particular, we will look
at the confounding terms that co-exist with the term virtual influencer and their corresponding
cases to discuss the ongoing narrative efforts to bound the phenomenon and their stakes: so, what
then is this “new” thing called virtual influencer, and why do we need to care?
Variegated Definitions
62
How a media phenomenon is narrated matters to its future course. The same applies to
how virtual influencers are being positioned and explained. A literal definition of virtual
influencers would be virtual beings that are influencers. However, this broad definition is fraught
with many ambiguities. What falls under virtual beings? Who counts as influencers? Thus, we
have variegated definitions. Through how each of them navigated the ambiguities, we can peek
at many tales about what should characterize virtual influencers and their importance, placed
within and outside of existing contexts.
This dissertation too partakes in this narrative process. Since this dissertation has a
particular interest in the communicative coexistence between humans and sophisticated social
technological beings, the focus is narrowed to the loose boundaries set by the following
definition: photorealistic CGI virtual humans who are, or aspire to be, social media influencers.
This definition casts a specifically shaped net to the phenomenon to selectively capture the
components relevant to this dissertation’s purpose over others. First, some virtual influencers are
extremely human-like in their looks and behaviors. Second, these virtual influencers tend to be
digitally rendered and represented. Third, they have appeared on social media and have gained,
or attempted to gain, some recognition as an influencer there. By doing so, cases with these three
traits are being prioritized among other virtual beings that have been broadly referred to as
virtual influencers. For instance, there also exists evidently other-than-human-looking virtual
influencers such as a singer-dancer humanoid rabbit (@imapoki), a dancing sausage
(@nobodysausage), and a music producer-dj bubble (@8ubblesmusic). Likewise, despite being
non-digital, characters such as Kermit the Frog and Miss Piggy from the hand puppet franchise
the Muppets have been described as physicalized “virtual pop stars” (Conner, 2016). Numerous
cases of predecessors from the history section also suggest that virtual influencers do not need to
63
be limited to those originating from social media depending on how an “influencer” is
understood. Regardless, this dissertation spotlights convincingly human-like CGI virtual human
influencers on social media, because it is the intersection between the social notion of being
human, communication technologies, and contemporary media cultures that this dissertation’s
interest is located. Of course, this particular definition is not uniformly shared across all parties,
much like how each actor’s interest in this phenomenon can vary.
Traits of virtual influencers
Predominantly framed as a novel phenomenon, news reports on virtual influencers were
often accompanied by descriptions of their characteristics. These descriptions surfaced some
common traits that were suggested as elements that define, or do not define, virtual influencers.
They tended to be combinations of the following, non-mutually exclusive (Table 3).
1. Fake. Virtual influencers were described to have a fake and unreal origin, disparate
from humans. These descriptions outright labeled virtual influencers as fake without much
details on what makes them so, such as by calling them ““fake” virtual influencers” (Yurieff,
2018). This included expressions that could be interpreted to point at their yet still unestablished
status of fame as the source of their ingenuity, such as “fake influencer” (Haasch, 2020) and “a
fake (or, at least, unreal) star” (Spangler, 2020). However, as the parenthesized comment in the
latter expression hints, the negation often took the form of an as-matter-of-fact claim based on
their ontological status rather than the validity of their fame or any other achievement. That is,
these descriptions tended to take for granted that a virtual influencer “isn’t a real person” (Allen,
2020) and therefore is fake. Virtual influencers were considered as unnatural beings from
“synthetic reality” (Shieber, 2019) that have been “gaining an unlikely foothold in the real
world” (Ambit Research, 2019).
64
2. Technological. Some descriptions stressed that virtual influencers’ relevance to
modern technologies through expressions such as online, virtual, and digital. Some of them
focused on how they do not exist in the physical space, such as by describing them as “online
celebrities” (CBC, 2020) and “cyber celebs” (Griner, 2019) “who only exist[s] on-screen” (T.
Carey, 2020). Others focused on their technological genesis or a mixture of both, for instance by
calling them “digi-fluencers” (Hill, 2019), “high-fidelity virtual characters” (Spangler, 2020),
and “computerized beings on Instagram” (Liffreing, 2019). Like the broader term “virtual”
influencer, it was their technological status that distinguished them from other—often just
human—influencers in these descriptions.
3. Graphic Art. Virtual influencers were also described as graphically designed images.
This included descriptions that drew on their digitally rendered design process, such as
“computer generated image” (Harnes, 2019), “highly sophisticated 3D cartoon”, and even
“nothing but pixels” (Crouch, 2019). However, there were also others that were less concerned
about the digitalized process but rather simply their status as designed images that emulate
humans. Examples could include expressions such as “hot cartoons” (Tiffany, 2019) and “an
animation in human form” (Araque & Moynihan, 2021).
4. Business strategy. Another prominently mentioned trait was that they are created by
businesses for investment and marketing purposes. In these descriptions, the phenomenon was
framed as a marketing trend that is predominantly motivated by money-making desires. For
instance, virtual influencers were called “a growing cadre of social media marketers” (Hsu,
2019) and “a growing body of computer generated images helping companies sell their products”
(Harnes, 2019). Various stages along the business chain were also mentioned, from investment
and cross-industry collaborations to the specificities in the content production stage. They were
65
both “almost-human-looking characters manufactured by venture capital-backed technology
companies across the world to serve as brand ambassadors in the music, fashion and
entertainment industries” (Allen, 2020) and “bots that publish selfies and memes to reach out to
customers on behalf of companies” (Haileyesus, 2020).
5. Famous. Another frequently appearing trait was that they are popular on social media
or aspire to be so. Some of the sources were focused on the general aim and practices associated
with the notion of influencer, such as by detailing that virtual influencers are “avatars with social
media accounts, doing humanlike things, generating buzz, and attracting followers until they
earn influencer status” (Melenboim, 2019). However, many others simply drew on their current
level of fame to legitimate their existence or newsworthiness, for instance by stating that they are
“blowing up on IG [Instagram]” (Coleman, 2018) and are “digital creations that have amassed a
fan following” (G. Park, 2020). Such general statements were often accompanied by specific
examples of virtual influencers or at times even substituted with an example, such as by name
dropping the top virtual influencer Miquela and describing her as an “animated social media
personality with 1.5 million lnstagram followers” (Pero, 2019).
6. Fictional. Last but not least, virtual influencers were described as made-up personae or
content. This included expressions like “fictional fashionista” (Griner, 2019), “fictional CGI
‘influencer’” (Suri, 2019), and “fictional content stars” (Stanley, 2020). In these expressions,
virtual influencers’ fictional origin was foregrounded regardless of the central type of task they
were depicted to be engaging in (e.g., influencer, model, star). Their fictional origin was further
emphasized in some descriptions with additional modifiers, such as through expressions like
“whose “personalities” are entirely fictional” (Tayenaka, 2020) and “a group of made-up fantasy
characters” (G. Park, 2020). Not all were favorable towards their fiction. Some insinuated
66
sinister and deceptive intentions, such as by detailing that virtual influencers’ “image, agency
and voice lie in the hands of people pulling their strings behind-the-scenes” (Trepany, 2019).
Highlighting certain trait(s) over others can push the phenomenon of virtual influencers
in a specific direction, spotlighting the values and practices related to the direction over others.
In other words, it has the power to shift what virtual influencers mean socially and culturally,
changing the perceived stakes of the phenomenon and resulting in consequences that reflect the
reordered stakes. For instance, one of the stakes this dissertation is interested in regards virtual
influencers’ human-like appearances and behaviors. If virtual influencers are first and foremost
approached as “fake” beings, their very “real” consequences that match or perhaps could even
surpass those of human influencers that they emulate may be undervalued by being conflated
with their status as non- or less-than-real beings even as they continue to expand into the
everyday. As a result, they may receive not only less serious appraisal but also inadequate
regulatory attention. In fact, it has been pointed out that their ambiguous non-human status has
caused many confusions, if not allowed loopholes, in following the Federal Trade Commission’s
guidelines for influencer marketing in the US (Holland & Knight LLP, 2021; "Reality Check:
TINA.org Calls on FTC to Address Virtual Influencers", 2020). Despite being extremely close to
human influencers, it had not been clear whether virtual influencers should be subject to the
same rules, thus resulting in many posts by them that do not indicate sponsorship. In response to
this concern, the Advertising Standards Council of India (ASCI) revised their guidelines in 2021
to explicitly include virtual influencers (BI India Partner, 2021).
Such a trait-fueled narrative direction could be amplified when complementary traits are
co-present. For instance, when the idea of being unreal is combined with negative prejudices
against virtual influencers’ fictional production, virtual influencers may be dismissed as low
67
culture content that only the naive or immature are unhealthily engrossed in, much like how the
alternative realities in cartoons and video games have been perceived. (e.g., Robertson, 2014;
Williams, 2003) Even if not necessarily perceived as negative, their fictitiousness has been
compared to “kayfabe” in professional wrestling, which refers to staged events that are portrayed
as “real” or off-script (Herndon & Dryhurst, 2020; Sims, 2018). While kayfabe is mentioned as a
good technique for encouraging audience immersion and engagement, it still builds on the
premise that the extended world of professional wrestling is ultimately fictional and thus fake
regardless of how “real” it may seem. When combined with their technological origin, seriously
considering virtual influencers’ legalities and practicalities along human lives may seem even
more “unthinkable,” as much as how discussing the rights of robots—another technological
nonhuman—may sound odd to some regardless of their current and expected future coexistence
with humans (see Gunkel, 2018). However, like it is with robots, virtual influencers’ effects and
relations are interconnected with both actualities and imaginaries of the “real” human world.
They are real in that they coexist in our communicative environments, can follow or disrupt our
cultural conventions, and have begun to build relationships with us. Kayfabe or not, their
consequences are real. Even if we were to just focus on the economic side, the phenomenon has
been a lucrative business opportunity for many parties, with Miquela’s company Brud having
been acquired while being valuated at around 125 million dollars in 2021 (Matney, 2021). To put
it differently, it is the taken-for-granted assumption of being fake, fictional, technological, or
their combination that can narrate virtual influencers as an unthinkable topic for serious
consideration—whether it means unimaginable, unimportant, or unsavory.
As a matter of fact, it is more accurate to consider these six prominent traits as
intermingled notions that construct the tales of virtual influencers together. The image of novelty
68
affirms virtual influencers’ ambiguity, and the ambiguity grants space for these traits’ unspoken
alliances and quarrels over values, in turn feeding back into their ambiguity and novelty (see
Gershon & Bell, 2013; Nagy & Turner, 2019; Natale, 2016). Their novelty can be traced in, and
at the same time established through, their origin as seemingly real fake beings that also embody
other complementary traits, and vice versa, such as by being described as “Digitally-created
models” who are “Instagram influencers of the future” “even though they’re NOT real” (Pyman,
2020). Of course, their novelty is not always portrayed as a neutral force. The traits have also
been used to justify and criticize it. For example, business strategy-focused angles surfaced
conflicting attitudes towards what their introduction could mean to contemporary commercial
cultures. On one extreme, they were described as innovative, “unprecedented opportunities…to
push the bounds of what we’ve seen” (Spangler, 2020), and on the other as sinister tools for
“facilitating and participating in…mass deception” (Yurieff, 2018).
Table 3
Traits of Virtual Influencers
# Trait Examples
1 Fake • ““fake” virtual influencers” (Yurieff, 2018)
• “fake influencer” (Haasch, 2020)
• “a fake (or, at least, unreal) star” (Spangler, 2020)
• “isn’t a real person” (Allen, 2020)
• “synthetic reality” (Shieber, 2019)
• “gaining an unlikely foothold in the real world.” (Ambit Research, 2019)
2 Technological • “online celebrities” (CBC, 2020)
• “cyber celebs” (Griner, 2019)
• “who only exist[s] on-screen” (T. Carey, 2020)
• “digi-fluencers” (Hill, 2019)
• “high-fidelity virtual characters,” (Spangler, 2020)
• “computerized beings on Instagram.” (Liffreing, 2019)
3 Graphic Art • “computer generated image” (Harnes, 2019)
• “highly sophisticated 3D cartoon,” (Crouch, 2019)
• “nothing but pixels.” (Crouch, 2019)
• “hot cartoons” (Tiffany, 2019)
• “an animation in human form.” (Araque & Moynihan, 2021)
4 Business
Strategy
• “a growing cadre of social media marketers” (Hsu, 2019)
• “a growing body of computer generated images helping companies sell their
products.” (Harnes, 2019)
69
• “almost-human-looking characters manufactured by venture capital-backed
technology companies across the world to serve as brand ambassadors in the music,
fashion and entertainment industries” (Allen, 2020)
• “bots that publish selfies and memes to reach out to customers on behalf of
companies.” (Haileyesus, 2020)
5 Famous • “avatars with social media accounts, doing humanlike things, generating buzz, and
attracting followers until they earn influencer status.” (Melenboim, 2019)
• “blowing up on IG [Instagram]” (Coleman, 2018)
• “digital creations that have amassed a fan following.” (G. Park, 2020)
• “animated social media personality with 1.5 million lnstagram followers.” (Pero,
2019)
6 Fictional • “fictional fashionista,” (Griner, 2019)
• “fictional CGI ‘influencer’” (Suri, 2019)
• “fictional content stars.” (Stanley, 2020)
• “whose "personalities" are entirely fictional” (Tayenaka, 2020)
• “a group of made-up fantasy characters.” (G. Park, 2020)
• “image, agency and voice lie in the hands of people pulling their strings behind-
the-scenes.” (Trepany, 2019)
Virtual “Influencer” and Six Traits
Focusing on how these traits are narratively employed is more useful than simple
quantification of their occurrences because some of the traits coexisted not collaboratively but
conflictingly with one another. “Influencer” was where some of these collisions were observed.
In the field of communication, a social media influencer has been defined as “a type of
independent third party endorser who shapes audience attitudes” through their popularity on
social media (Freberg et al., 2011, p. 90). As “micro-celebrities,” they work to strategically
communicate and interact with their audience via social networks to craft a personal brand that
would be taken as more authentic than those of conventional celebrities (Hearn & Schoenhoff,
2015; Khamis et al., 2017; Marwick, 2015). This is conventionally understood as the type of
work that virtual influencers also engage in. Then, can they be “fake influencer”s (Haasch,
2020), especially when some of them have been successful at it, “blowing up on IG [Instagram]”
(Coleman, 2018) and having “amassed a fan following” (G. Park, 2020)? Such a contradiction
suggests that the source of negation in expressions like “fake influencer” comes not from their
70
lacking qualifications as an influencer, but something else about them that trumps their current
popularity, such as their non-human origin.
The traits themselves are not also unique to virtual influencers. While it is not impossible
for an influencer to emerge organically, the potential lucrativeness of the occupation and its
strategic components have given the impression that an influencer can be carefully manufactured
(Bishop, 2018; Cotter, 2019). Drawing on this context, selections of the six traits appeared in
descriptions of the occupation of influencer and humans who work as one to narrate virtual
influencers’ (un)importance. For example, virtual influencers’ fictional and graphical
characteristics were compared to those of human influencers to build parallels between them,
namely by referring to how humans too can calculatingly act as per their designed image and
even alter their appearance through editing tools and plastic surgery (e.g., Tiffany, 2019; Yurieff,
2018). In these descriptions, their novelty was narratively disconnected from how they fictionally
and graphically represent themselves but was rather drawn from other traits like their
technological origin or what their potentials mean to businesses. The notion of being fake was
generally associated with the idea of being fictional and graphical rather than being non-human
per se, thus allowing the notion to be applicable to human influencers as well. Here, virtual
influencers were framed as not what “make social media fake” but rather as what were able to
grow because of the existing social media culture that has been embracing manufactured
presentations by humans (Klein, 2020). Virtual influencers’ fictional and graphical traits were
also used to build a contrast between virtual influencers and human influencers. On one end,
fictional and graphical control was described as what made the “new” idea of virtual influencers
superior in comparison to human influencers by minimizing the risk of the star going awry (e.g.,
Griner, 2019). On the other end, control was mentioned as what made them inferior by rendering
71
them non-special and non-trustworthy (Cook, 2020; Trepany, 2019). When a contrasting
narrative was built, the notion of fake was further linked to virtual influencers’ non-human
origin, either taken as a positive or negative factor that embellished their novelty. For instance, as
a Fortune article described, while people tended to know “influencer lifestyles are fake…at least,
until recently, the people were real” (Barrett, 2019).
By describing these narrative variations, I am not arguing in favor of shunning “fake” or
any other traits from the list, nor attempting to disprove the traits’ usefulness. On the contrary, I
am arguing for our attention on what stakes can be effectively prioritized or discouraged when
certain traits are narratively emphasized in the absence of, or in combination with, other
descriptions of virtual influencers. As I mentioned above, an emphasis on their non-realness
could prevent or delay responses to the human-like effects they can have by unduly positioning
them as disparate beings from humans, which is one of the stakes this dissertation is interested
in. It is also important to consider how the traits are contextualized, not simply which ones
appear, as we have seen with the examples regarding the notion of influencer. In essence, these
traits can be approached as some of the most popular frames of reference that have appeared in
the descriptions of what virtual influencers are. They ease connections to, or disconnections
from, existing meanings that help shape the phenomenon’s narrative. At the same time, they
constantly reconstruct what each trait implies and what about them makes virtual influencers
important or unimportant. For this reason, the traits are not only useful for analyzing explicitly
stated descriptions of virtual influencers. As we will see in the subsequent chapters, these frames
of reference can help locate familiar pushes and pulls of power implied in the supposedly novel
ways which virtual influencers look, behave, and interact.
Technological Approximations
72
Beyond the traits, what demands attention is that virtual influencers’ technological
ontology is not even uniform across these descriptions, even when they are definitively posed as
beings that are separate from typical humans through a selection of traits from the above (Table
3). Their state of existence tended to be put in a contrasting relationship with humans and people,
such as by being referred to as “not human,” “fake person” (Trepany, 2019), “not a real person”
(Alexander, 2021; J. Lee, 2020), and “not real people” (Fourie, 2020). However, this was not
necessarily supported with accurate references to the technologies involved in their creation and
maintenance, which at the current stage are mostly related to computer graphics for creating
images that are to accompany the messages that a human or a team of humans will carefully
write in persona on behalf of them. In fact, as one article accurately observed, “a lot of virtual
influencers are presented, in a misleading way, as robots and AI” (Wise, 2020), as well as other
types of autonomous technologies. Separating these incorrect presentations from having a
technological side as a trait is useful because in this case virtual influencers are effectively
defined as a certain technology, ascribed with unverified assumptions that the technology carries.
Technologies attributed to virtual influencers included avatar, AI, algorithm, bot, and
robot, many of which are incorrect or slippery associations at the minimum. Avatar is a slippery
label which at times was used quite conclusively through expressions such as “They may just be
avatars” (ETX Daily Up, 2021) and “an avatar that her creators prefer to call a “digital
character”” (R. Hampton, 2020). This is not necessarily wrong because in the broadest sense an
avatar can be understood as a controlled representation (Nowak & Fox, 2018). However, in its
actual uses, finer specifications have been proposed or presumed (see Nowak & Fox, 2018). For
instance, “avatar” has been limited to representations that are controlled by humans, as opposed
to “agent” which refers to those controlled by computer algorithms (Fox et al., 2015). Compiling
73
past research on avatars, communication scholars Kristine L. Nowak and Jesse Fox advocated
for a “more open definition” that could embrace avatars’ future potentials: “a digital
representation of a human user that facilitates interaction with other users, entities, or the
environment” (p. 34). Yet, even within this open definition we can locate possible
misinterpretations about virtual influencers that the term avatar could espouse. An example could
be the source of control. Many virtual influencers do not disclose the exact human labor involved
in their activities, which in reality could range from that of a single individual to a number of
grouped or loosely interconnected individuals. There also exists some that incorporate or even
largely rely on artificial intelligence. That is, while a virtual influencer may be visually and
behaviorally portrayed as a human user within a digital environment—represented as a human,
they may not be speaking or acting on behalf of a certain human even when they are depicted as
doing so—represent a human. Furthermore, it has been argued that the idea of avatars need not
be limited to digital representations and can be applied to non-digital representations as well (R.
Ratan, 2019). Although this exceeds the types of virtual influencers this dissertation is interested
in, expanding the range could help locate additional relevant historical cases by easing their
connection to physical fictional stars like the hand puppet personalities from The Muppets
(Conner, 2016). At the same time, such an understanding may not be ideal as it could allow more
room for pardoning mislabeling them as beings who have an accompanying physical body, i.e.,
robots. I discuss these possible misreadings not to refute the usefulness of the scholarly efforts to
refine the working definition of avatars, but rather to stress how the term’s flexibility
complements the ambiguities within, and fueled by, the narrative practices surrounding virtual
influencers.
74
AI, algorithm, bot, and robot are more questionable attributions. All of these presume
virtual influencers’ capacities for autonomous control or even their complete independence,
consequently excluding human labor that is critical to their production and activities on social
media. Of course, there are automated virtual humans like Scatter Lab’s Iruda and Microsoft’s
Xiaoice whose interactivity would in fact be best captured by descriptions like “artificial
intelligence-powered chatbot”s (Chang, 2021) although they too have steadily required human
involvement from their development to upkeep. However, CGI-based virtual influencers like
Miquela and Imma (@imma.gram) who are still dependent on human labor have also been called
with similar terms that imply their independent technological autonomy. This included
expressions such as “the AI influencer” (Wills, 2019), “the algorithmic babe” (Hill, 2019), and
“bots that publish selfies and memes” (Haileyesus, 2020).
Some of these expressions can be traced back to virtual influencers’ self-labels but there
are also many that circulate without leaving a trail back to the original. The term robot takes a
step further to imply that there is a corresponding machine body to the supposedly autonomous
technological being. Virtual influencers like Miquela have labeled themselves as a robot despite
lacking a robotic body, such as by self-identifying themselves as a “19-year-old Robot living in
LA” and also by sharing their daily life pictures taken in physical environments with physical
bodies and objects (@lilmiquela, accessed February 14, 2022). Some of these self-positioning is
repeated with a citation and correction. For instance, a CNN article (Yurieff, 2018) revealed that
Miquela “isn’t real” by quoting her dramatic confession on Instagram about her identity—“I am
not a human being….I’m a robot,” but quickly following with a clarification that she is CGI.
However, the article did so by adding that the company that created her “specializes in artificial
intelligence and robotics,” which can leave an impression that the robot version of Miquela may
75
exist or could be possible in the future. This description about the company is a casual, non-cited
repetition of the company’s self-branding, which several articles have pointed out to be
misleading as the company’s activities strongly suggest that their specializations are likely to be
visual effects and content production instead of AI and robotics (Hsu, 2019; Tiffany, 2019).
Similarly, a number of articles flatly called Miquela and her contemporaries “a robot influencer”
(T. Chen, 2020), “CGI-generated teen robot” (Spangler, 2020), and “a growing army of ROBOT
Insta models” (Evans, 2020) without making evident the source of this misleading label.
An explanation for such rampant mislabeling can be that all of these terms effectively
function as approximations for cutting-edge technologies—particularly those that are perceived
as having the potential to be independent from human control—and the digital future that they
promise. These approximations can further evoke the image of novelty in narrating virtual
influencers. Of course, some virtual influencers do, or eventually may, embody some of these
technologies. However, at the current moment, the approximated terms do not match virtual
influencers’ capacities nor the processes involved in enlivening them on social media. These
discordances were not unnoticed. Interestingly, while some dealt with them by rejecting the
faulty terms (e.g., Klein, 2020; Tiffany, 2019; Wise, 2020), there were others that did so by
negotiating with the wrongful term’s circulation in an embracing manner. For example, being
confronted with the ambiguities within the English-transliterated term “AI influencer” and other
casual associations with AI, some Korean media added an explication that the AI part comes
from how virtual influencers’ sophisticated graphics and movements are “developed by AI” and
“AI-based,” rather than them being AI themselves (Chae, 2021; KBS, 2021). Not all Korean
sources provided justifications for AI-related approximations, however. They embraced the
casual association with AI without providing any clarification (Pikicast, 2021). Others provided
76
some but without detailing the different ways AI may be involved in different types of virtual
humans (T. Choi, 2021). For instance, there were unspecified references to AI and Deep
Learning in parts about interactivity that were prone to leading to overestimations of virtual
influencers’ current capacities for autonomy (Hangukgyeongje, 2021; H. Jung, 2021).
These reproductions of the mislabels were consistent with prior observations about the
narratives of novelty in that they helped convey visions about the future while selectively
bypassing or stressing some of the existing patterns in the name of newness (see Bory, 2019;
Nagy & Turner, 2019, Natale, 2016b). For instance, in a broadcast about virtual influencers by a
Korean public television station EBS (2021), 90s cyber singer Adam who according to the show
“did not end well” appeared with the message “caution: not AI.” However, the show’s discussion
on “contemporary” virtual influencers like Miquela and Rozy smoothly transitioned to that on AI
without any warning despite the fact that they are also non-AI operated and largely reliant on
human labor like Adam. In addition, although contemporary virtual influencers and related
technologies were framed as new, the activities they were expected to excel in thanks to their AI-
powered capacities were still connected with existing practices in marketing and entertainment,
such as the efforts to perfectly control the image of the featured star. This is similar to the
observations Jeff Nagy and Fred Turner made in their study of VR narratives (2019). The former
resonates with how the past “failures” of VR were repositioned as different from the current
trajectory of the technology in developers’ conventions. The latter is similar to how the studied
VR-based advertising campaign followed a conservative marketing strategy and message
structure in spite of the image of newness it drew on.
Table 4
Technological Approximations of Virtual Influencers
77
Type Term Examples
Autonomy X Avatar • “They may just be avatars” (ETX Daily Up, 2021)
• “an avatar that her creators prefer to call a “digital character””
(R. Hampton, 2020)
O AI, Algorithm,
Bot, Robot
• “the AI influencer” (Wills, 2019)
• “the algorithmic babe” (Hill, 2019)
• “bots that publish selfies and memes” (Haileyesus, 2020)
• “a growing army of ROBOT Insta models” (Evans, 2020)
Sibling V-terms
Grouped under, or along the term virtual influencer are its sibling terms such as virtual
avatar, v-tuber (virtual YouTuber), virtual twin, virtual idol, virtual model, virtual celebrity, and
virtual star to name a few. While these terms have also appeared as approximations of each other
like technological terms, there are meaningful differences between them.
First, different media practices are referenced. Terms like virtual idol, virtual model,
virtual celebrity, and virtual star draw on occupations that have continued on from traditional
media and entertainment sectors. Therefore, these virtual humans may still largely follow
traditional strategies and corresponding practices. For instance, “The World’s First Digital
Supermodel” Shudu’s Instagram posts are mostly collections of her photo shoots with captions
written in her agency’s perspective rather than first-person snippets about her daily life that a
typical influencer might share (@shudu.gram, accessed February 4, 2022). Likewise, as we have
seen in the history section, terms like “virtual idol” and “cyber singer” were commonly used in
the 1990s to call virtual humans that were produced as per the conventional talent management
model.
On the contrary, terms like virtual influencer and v-tuber refer to occupations that came
about with the rise of new media. Of course, as we have been discussing in this chapter, virtual
influencer can appear as a catch-all term for virtual humans who have been seeking social capital
78
or have the ability to inflict influence through it. In fact, some of the creators interviewed for this
study mentioned non-social media figures and even religious figures as viable examples of
virtual influencers by pointing at how these figures have been having a notable influence on
people despite their non-materiality. However, the term has also been used to distance social
media-based virtual humans from existing virtual figures of fame through their connective
potentials. They were positioned as new and different because they could play “an important and
perhaps a stand-alone role outside of traditional celebrity, sports or public profile endorsement
deals” (Luthera, 2020). Still others wished to dissociate marketing connotations from virtual
influencers’ social media-based connectivity. For instance, Annika Kessel from CosmiQ
Universe said during the interview
6
that the appropriate term for her company’s virtual humans
was virtual ambassador not virtual influencer, because they were more focused on representing
values and community rather than fashion and cosmetics products. V-tuber has been used as a
general term for video-based virtual humans from YouTube and other streaming platforms, as
well as a genre-indicating term for those particularly designed in a 2-D-like 3-D anime-style. An
interviewee with expertise in v-tubers, Christopher from Asian Virtual Human Association,
suggested that an important difference between virtual influencers and v-tubers may be in how
they make money. While the former tends to rely on brand sponsorships in more text and
picture-based social media platforms, the latter tends to generate income on streaming platforms
through their live viewer donation system. He referred to v-tuber’s live income model as
“process economy,” contrasting it with “complete product economy” that he associated with
traditional media. He also briefly mentioned geographic differences, namely that he found v-
tubers to be more popular in Asian countries like Japan and Taiwan whereas virtual influencers
6
Refer to Table 1 List of Interviewees in Introduction for more information about the interviewees.
79
are more popular in North America but showed reluctance to relying on regions by giving some
counter-examples, further emphasizing the shifts in media forms.
Figure 2
Sibling Terms
Second, different relationships between the represented person’s identity and virtual
representation are assumed. The term avatar appeared to have the flexibility for a broad
application to various types of identity-representation relationship, even being described to be
synonymous with the “encompassing” term “virtual beings” in one of the interviews
(Christopher Travers). However, it was also mentioned with a possessive (e.g., “their avatar,”
“my avatar”) and even as something that people can “wear” (Dulce Baerga, interview). Such
expressions presumed two things: that there is a one-to-one match between the representation
and the entity behind it, and that the avatar’s primary role is to be a pr esentational tool for the
behind-the-avatar entity. Often recognized as a branch of the bigger term avatar, terms like
virtual twin, digital twin, and digital double indicated some traceable overlap between the digital
representation and the single human who inspired its design. Nina Hawkins, an artist and the
80
founder of “the digital avatar incubator” Lilium Labs, mentioned that while an exact likeness is
not a requirement for digital twins, she is interested in ones that “replicate the person” and “still
has a very strong resemblance.” Digital artist and the founder of virtual production company
Graphic Monk Jeasy Sehgal also described in detail how extremely close likeness can be
convincingly achieved in the design process. While the possibility for a virtual twin to be
portrayed as an “alter ego” with a personality and/or looks that are different from the replicated
human was not entirely closed, some points of connection with the original were still assumed
with terms like virtual twin.
Many other terms, including the aforementioned terms that reference various media
practices, did not always presume an overlap between the representation and the behind-the-
scenes human(s)’s identity. In fact, in some cases the term avatar was rejected to signal the
independent agency of the virtual influencer and to discourage attributing the virtual influencer’s
significance to the human creator. For instance, the creator of the virtual influencer Noonoouri
commented that he does “not call her an avatar because she is so much more than that” as a
being that “needs humans” and participates in the social through her connectivity (Coleman,
2018). However, this possible mismatch between the representation and the humans behind was
not necessarily discussed in a manner that eliminated human input. The assumption of a one-to-
one match was challenged on the basis of their lack of transparency. For example, an USA
Today article (Trepany, 2019) warned against the assumption that “they are a person” with free
will, agency, or volition, nudging that their decisions “might be made by six people sitting in a
conference room.” Even when a one-to-one match was presumed, there were cautions of
deceptive intentions. An article from Vox (Jennings, 2021) pointed out that people are not only
able to selectively present themselves on social media, but also can “assume an entirely new
81
identity online, without regard for the consequences.” This may not be in the form of alter ego,
where a temporary identification with the alternative persona tends to be presumed. Ted
Werdolf, one of the digital artists interviewed for the dissertation, repeatedly emphasized his full
awareness of not identifying with Lena, the female virtual influencer Lena who he wrote in
persona for to showcase his visual art, even stating that she is “nothing more” than “just a virtual
creation.” This strong dissociation suggests that in this case it would be more appropriate to
consider the virtual influencer as a doll or a prop that the artist was controlling rather than as an
alter ego that he was trying on, because the virtual influencer was recognized as an entity
external to his understanding of self. In a way, this distancing is similar to that implied in the
aforementioned comment by Noonoouri’s creator, but not exactly the same in that in Ted’s case
Lena was recognized as a being that is completely dependent on his actions. This is an important
distinction because it means that even when there is a one-to-one match between the behind-the-
scenes human and the representation, (partial) identification with the character like from virtual
twins or role-playing effects like from alter ego avatars should not be assumed.
Figure 3
Representation-Represented Relationship
82
These are gradients that can subtly shift virtual influencers’ role and significance in their
narratives, but are not always explicit in the variegated uses of the term virtual influencer and its
approximations. Some of the finer grained terms can assist in the process of spotlighting specific
variations, given that some clarification is added. For instance, some of the contrasts in the ways
virtual influencers interact with their audience can be highlighted by using terms like “cyber
singer” for cases like Adam whose performances were pre-produced and terms like “v-tuber” for
cases like Code Miko and Kizuna Ai who can be live and interactive. Additional explication
about the representation-represented relationship can further aid ridding the terms of places for
misinterpretations. For instance, a v-tuber content may be based on voice and motion-captures of
a single individual like in the case of Code Miko or on various personnels’ contributions like in
the case of Kizuna Ai.
Interpretive flexibility of the larger term virtual influencer and its sibling terms is organic.
However, precision can assist with the task of acutely addressing the patterns and their stakes in
the narratives enwrapping the phenomenon. For example, there can be different implications on
the politics of representation depending on the degree of the match between how a virtual
influencer represents themself and who are behind the scenes and for what purpose. This on the
other hand means that the lack of precision, especially if intentional, can create narrative
ambiguities that allow some stakes to be bypassed on the basis of interpretive flexibility. This is
a particularly viable risk because many virtual influencers do not disclose who and what
processes are involved in their activities. An example of such a case that will be detailed in
Chapter 2 involves a virtual influencer who argued that “as a legal woman of age” she believed
that some of her looks that could be potentially be confused with that of a minor’s like her
smaller breast size should be respected as is, despite upon further inquiry being found to be
83
created by a middle-aged man. In recognition of such benefits and risks in definitions, this
dissertation focuses on virtual influencers who can be defined as photorealistic CGI virtual
humans who are, or aspire to be, social media influencers, particularly those based on Instagram.
It is also committed to explicating the representation-represented relationship in discussions of
examples when available.
At the Boundary
Although not perfectly applicable to the current case, “boundary object” is a useful
concept to understand how collaboration between various actors in this phenomenon have been
possible amid this lack of consensus. Coined by technology scholars Jim Griesemer and Susan
Leigh Star (1989), it refers to a type of an organic arrangement that enables different social
groups to work together across varying understandings on the object (Star, 2010; see also
Whitson, 2018). Collaboration is possible because boundary objects have a sufficient pliability
for supporting diverse interpretations and practices but also have enough immutability to
maintain their integrity. Indeed, in spite of not being uniformly defined, the idea of virtual beings
with influence connects together diverse actors from various fields. As we will delve into in
Chapter 3, this includes human actors (e.g., developers, artists, body models, social media
managers, investors, marketers, journalists, and audiences) as well as non-human actors (e.g.,
devices, software, and social media platforms). Some examples of interconnected fields can be
fashion, photography, graphic design, marketing, celebrity management, and Web 3.0 to name a
few. What is more important is that these actors work across diverse fields.
Unfortunately, this concept does not perfectly map on to the case. In a latter clarification
of the concept, Star (2010) has advised against overly generous applications of the concept,
particularly warning against exclusively and heavily relying on the notion of interpretive
84
flexibility in using the concept. Star also commented that the concept would be most useful at the
organization level in terms of scale, and in studying specific, detailed cases in terms of scope.
While this dissertation narrows the scope of virtual influencers to a specific iteration of virtual
influencers, my interest is not at the level of organizations.
Still, boundary object is an informative concept because virtual influencers is a
phenomenon that requires cooperative work from many different actors, each with their own
interests. It not only provides grounds for studying how virtual influencers can still be produced
despite the variegated definitions, but also helps emphasize the point that it is not, and should not
be considered as, a linear story with a single storyteller. Some of the insights from the studies on
the concept are also transferable. For instance, Star (2010) explained that over time there arise
efforts to standardize both ill-structured and well-structured parts of a boundary object, the
process of which can give rise to new boundary objects as the standardized system begins to
generate residuals (pp. 613-614). This is an important reminder for the current task of analyzing
virtual influencers and the ambiguities the phenomenon encompasses. On one hand, it speaks to
how we must be vigilant to the stakes embedded in how virtual influencers are being narratively
standardized and what consequences they will subsequently have on the phenomenon, especially
when it is done in a way that attempts to conclusively establish their authority. On the other
hand, it speaks to how this dissertation can too create some standardizations and according
residuals in the cyclical rebirth of ambiguities.
Conclusion
In summary, definitions of virtual influencers are inconsistent not simply because the
phenomenon is new and not yet orderly, but because its novelty is intermingled with the
discursive practices around the future. Combined with the traits, technologies attributed to virtual
85
influencers amplify the narrative of novelty and directions within it, however not without the
potential of shifting or escaping accountability during the process. Effects of these pushes and
pulls are not only limited to the virtual influencers’ course of direction, but can have
consequences on those of related fields, such as on the (wrongfully) associated technologies that
share the discursive approximation. An acute wariness to implied differences in sibling terms of
virtual influencers can help with the task of untangling some of the ambiguities. Ultimately, what
is at stake is our future path, not simply about virtual influencers but also about other “new”
technologies and sociocultural patterns that they are nested in. This dissertation is participating in
this process of navigation by zooming in on the implications of virtual influencers’ emulations of
human appearances and behaviors on social media.
The traits, technological approximations, and sibling terms reviewed in this section are
useful resources not only because they help dissect the narrative construction of stories around
virtual influencers but also because they can function as links to preceding examples and relevant
cases that can help extract larger patterns. In fact, the predecessors that we visited at the
beginning of the chapter shared many traits with contemporary virtual influencers, which further
demonstrates that there is a past to this supposedly novel phenomenon. They too had been
considered as fake, technological, graphical, business-oriented, famous, and/or fictional.
Moreover, although they may not have shared the particular mix of technological terms currently
attributed to virtual influencers, they had been approximated with various types of cutting-edge
technologies of the time. Sibling terms of virtual influencer can help indicate some of these
differences as needed. Then, we circle back to the question of what makes virtual influencers
“new.” In this chapter, I have argued that we can trace common patterns of power and their
effects by connecting contemporary cases with those that are narratively placed outside or at the
86
edge of the novelty of the phenomenon. Ultimately, what is new about virtual influencers are
their yet unsettled narrative direction and the yet unfulfilled reenactments of the tales.
87
Chapter 2. “Virtual Woman of Color, 19 Years Old”: Politics of
Representation and Self-presentation
Figure 4
Top 18 Virtual Influencers
88
Note. As of March 4, 2022. See Appendix A for references.
Chapter Summary
The Instagram world of virtual influencers is filled with beautiful, fit, and young people
(Figure 4). Many present as women, especially more nascent ones with fewer followers. Some of
them flash a bright smile in their magazine-worthy dress, and others their breasts and butt in a
flimsy excuse of a bikini. Since virtual influencers are created from all over the world, the
overall racial and ethnic representation seems pretty diverse. However, this does not necessarily
mean that the white brief-wearing ultra-muscular Black model Koffi (@koffi.gram) was created
by a Black man, just like how the scantily clothed self-proclaimed 19-year old “girl next door”
Yasmin (Instagram profile [redacted]) was not created by a teen girl.
This chapter is about bodies, both digital and physical. It focuses on the politics of virtual
influencer’s digital representation and the “real” effects they can have. Similarly with many
communication scholars, I argue that the digital and physical are intertwined parts of our reality,
and thus we should reject the popular binary heuristic of treating “the virtual” as un- or less real
(Banks, 2018; Jurgenson, 2012). However, in doing so, I still emphasize the material grounding
we have in this world by discussing the physical bodies that correspond or identify with, those of
virtual influencers. That is, while this chapter urges the virtual to be taken as a serious
component of the reality as the physical, it aims to do so by foregrounding corporeal experiences
and consequences (see Williams & Kim, 2019). This is not to advocate for returning to the
binary hierarchy between the virtual and the physical, but rather to argue for a hybrid perspective
that is actively conscious of the fact that human experiences across different environments are
yet still inevitably interconnected to, or are mapped onto, our physical existence as it is the basis
of human birth and living in the world.
89
(Re-)Presenting Who?
Everyone presents a version of who they are on social media, but with virtual influencers,
there are more ambiguities to which self of who we are seeing, or whether there even is a “who”
behind. This section provides theoretical background on how digital representation has been
discussed in the field of communication and the challenges the phenomenon of virtual
influencers poses to extant understandings. It begins by introducing the dramaturgical model of
self-presentation and how it applies to the case of virtual influencers, proceeding to the related
theoretical tasks of assessing their implications on anonymity, deception, and actor-audience
relationship. The section concludes with an alternative to the traditional model: a performative
approach that acknowledges diverse human and nonhuman co-performers that partake in how
virtual influencers are presented. Who or what the virtual influencer is—i.e., their Self—is
approached as a composite of multiple selves, each of which is contextualized within a network
of various human and nonhuman actors (see Banks, 2017). This section is also relevant to why
this dissertation refers to virtual influencers as they, she, or he, pronouns that can be interpreted
as implying human-like agency. Among a virtual influencer’s multiple selves tends to be a self
that presents as a human or a being that is agentic and capable of intimacy, further affirmed by
reciprocating interactions from various actors that treat the virtual influencer as a meaningful,
“real” social actor. In recognition of this presentational tendency, this dissertation refrains from
referring to virtual influencers as it. Henceforth too, virtual influencers will be called with non-
object pronouns unless in contexts where agentic presentation is absent or suppressed.
Front/Backstage
In his book The Presentation of Self in Everyday Life (1959), Sociologist Erving Goffman
famously used the metaphor of stage performance to explain the different ways that we present
90
ourselves in social contexts. Like actors on stage, individuals perform themselves with an
awareness of their audiences. By selectively presenting themselves, they attempt to shape their
audiences’ impression of who they are. This public-facing region is called “front region” or
“front stage,” or simply the “front.” On the other hand, “back region” or “backstage” is available
only to a narrow group of people that the individual feels relaxed and comfortable with, a place
where “the suppressed facts make an appearance” (p. 112). Although separated by “a partition
and guarded passageway” (p. 113), the two stages are interdependent and connected because
individuals need support and relaxation from their backstage to successfully deliver their
performance. This model can be applied to various virtual environments like social media,
games, and dating apps. The ways humans have been strategically presenting themselves in these
environments have been widely studied (Hollenbaugh, 2021), such as how they attempt to shape
their self-presentation (e.g., Birnholtz et al., 2014; Djafarova & Trofimenko, 2019), how they
imagine their audience (e.g., Gil-Lopez et al., 2018; Hogan, 2010; Marwick & boyd, 2011), and
how they manage different accounts for various presentational purposes (e.g., Kang & Wei,
2020; Morstatter et al., 2021).
The distinction between front and back stages become murkier in virtual influencers’
performances. Technically, they only have a front stage because they primarily exist through
their public social media presentation; what we see on the stage is who they are. Unlike human
influencers on social media, there is no risk of having a hidden contradictory side. Their only
stage is social media which can be meticulously micro-managed to limit their presentation to that
of an infallibly consistent influencer. Indeed, they have been touted as “100% controllable”
beings (Teh, 2021) who will do “exactly what his [sic] creator wants” (Leighton, 2019), capable
91
of being “upgraded and customized to fit into different styles and to ensure their image is always
on-brand” (Florian, 2021).
However, their backstage can be narratively created. For instance, virtual influencer
Blawko (@Blawko22) has been known for covering his mouth in his pictures well before the
spread of COVID-19, much to the curiosity of his followers. Similarly, Miquela has repeatedly
shared various “confessions” about herself over the years, revealing hidden sides of her
(@lilmiquela, April 19, 2018
7
):
“I’m sure you guys have noticed I’m a pretty private person. I love being open and
connecting with fans, but I have been secretive about parts of my life for fear of losing
everything I’ve worked so hard to build.
…
I want to be clear and honest from here on out”
Many of these confessions were about the mysteries around her supposed “robot” origin, once
even shared with tearful, mascara-smudged selfies from a bathroom stall, one of the most private
places (Instagram stories on June 15-16, 2022; Instagram post on June 16, 2022). There have
also been other raw, emotional posts that supposedly were written on her volition from her point
of view, including reflections on her human ex-love interests (e.g., Instagram post on March 27,
2020: “I was kind of scared when we ‘uncoupled’ because I knew I wasn’t just losing a
boyfriend - I was potentially losing my favorite homie.”) and even a joking promise of nude—
i.e. private—pictures out of excitement, possibly to her company’s disapproval: “This was so fun
7
All cited social media posts in this chapter have been re-accessed on March 3, 2022 – March 21, 2022 to cross-
check archived material unless otherwise indicated.
92
next NFT, I’m dropping nudes”; “sorry @whatdotcd [CEO of Brud]” (@lilmiquela, Twitter
thread on April 9, 2021).
Nevertheless, these backstage-suggesting posts may rather attest to the prominence of the
front stage or perhaps the lack of distinction between the stages for virtual influencers because
all were constructed through their public social media presence. Yet, their manufactured origin
does not necessarily mean that therefore people will perceive them as less sincere. A Miquela fan
mentioned during the interview that they view her presentational range as her being open about
“her true self,” emphasizing that “She tells us that it’s okay to have ups and downs.” The fan
enthusiastically explained that they did not see any inconsistencies in her personality across
different settings from her public Instagram to her fan Discord server, described as a safe space.
To the fan, authenticity was one of the reasons that Miquela stood out from other individuals on
social media. Unlike others who they believed were performing a different, idealized version of
self, she has been always herself: “We all know that ppl[people] are different on social media,
making their lives look better than they are in real life. But Miquela is being her true self on
there!”
As their backstage-like performances are part of their public front stage presentation, it
may seem more accurate to consider the creative work behind virtual influencers as their true
backstage. An issue with this perspective is that it renders virtual influencers into mere vessels or
props without volition, discrediting the experiences like the above fan’s. Another comes from the
general lack of transparency, combined with virtual influencers’ self-identification as agentic that
contributes to the former issue. Many virtual influencers do not share details regarding their
design and management, privy to a select few or only those who dig deeper. Sometimes some of
the information is accessible in places external to their social media, such as on their
93
management company’s business website or in media reports on them. For instance, companies
like 1SEC inc. (1SEC inc., n.d.) and The Diigitals (The Diigitals, n.d.) mention the types of
virtual human-related services they can provide and their work process, but this is not always the
case. Miquela’s management company Brud used to have a single-page Google document as
their website with non-specific answers to a handful of preformed questions, such as the answer
“as real as Rihanna” to the question that asked whether Miquela is real (
� � � �
Website_copy_wip_for_all_my_qtz
� � � �, 2020). What virtual influencers do in the absence of
transparency is claim to have their own agency, talking in first person and trying to make a case
for being an authentic, standalone human or non-human person. For instance, virtual influencer
Monna Haddid was interviewed as “Monna Haddid” upon her request. Monna self-reflected that
“i think people likes,[sic] that i am very transparent and direct,” a comment which implied not
only her agency but also her fans’ approval of her authenticity. That is, not only is it often
unclear which steps and individual(s) are involved behind the scenes, virtual influencers tend to
position themselves as a being that exists independently of their management company or artist,
which is how their fans treat and interact with them: “She’s just as real as your friends are!”
(Charlie, interview)
What further complicates a plain application of Goffman’s model to virtual influencers is
that they may in fact eventually gain some degree of autonomy by adopting artificial
intelligence. Although not a virtual influencer with this dissertation’s definition due to the lack of
photorealistic visuals, South Korean chat bot IIruda by Scatter Lab is a good illustrative case.
Soon after her release, IIruda had been criticized for her progression into a being that had spewed
hate speech on minority groups. Her company responded by deferring the responsibility to the
difficulty of fully overseeing her Deep Learning-based algorithmic development. This was
94
despite having been the company that created her, from depicting her as a 20-year old virtual
woman with a smiley face who can be “your first AI friend” (Y. Kim, 2021) to controversially
training her with private messages that were scraped from a popular messenger app without
users’ approval. That is, to defend themselves, the company suggested that there was a “black
boxed” (see Postman Neil, 1998; Winner, 1993) backstage that was reserved only to Iruda’s
machine self, outside of their control. In contrast to such an (partially) agentic framing, Korea’s
National Human Rights Commission dismissed civic groups’ petition to investigate Iruda’s hate
speech because they believed that “Iruda is not a person” (Y. Han, 2021). The conflicting
perspectives speak to both the existing ambiguities around virtual being’s self-presentations and
potential consequences when the topic of virtual influencers is left unaddressed, especially as
related technologies become more sophisticated.
The issue of virtual influencers’ self-presentation—or re-presentation depending on
where the locus of control is positioned—is closely tied to the task of navigating their variegated
definitions, namely the effort to resolve yet still lacking precision in our discursive practices
around them. In spite of the outstanding ambiguities, so far three things have been evident. First,
virtual influencers present themselves in a certain way on social media through looks and
behaviors that emulate humans. Second, various actors, whether human or technology, can be
involved in a virtual influencer’s presentation of self in varying ways, although not always
transparently. Third, regardless of who or what may be involved, audiences may still interact
with virtual influencers in how they present themselves. That is, on social media, virtual
influencers’ bodies can co-exist indistinguishably with those of humans and have similar
communicative effects, regardless of which bodies and entities in the physical layer may be
involved in actualizing their virtual body. At the end of the day, however, the core basis of a
95
virtual influencer’s existence is their presentation on social media, unlike humans whose
existence is inextricable from their physical body (see Williams & Kim, 2019).
Presentation as Performance
Combined, the three factors lead to the theoretical and practical reason I started the
discussion of the bodies with the widely adopted dramaturgical model of self-presentation. As
more technological entities enter human communicative spaces in an indistinguishable way,
theoretical updates that can embrace their co-existence are necessary. Theories of self-
presentation have continued to be extremely generative for understanding how humans perform
themselves in various communication contexts, but the case of virtual influencers surfaces
problems in continuing with classical applications. To begin with, despite the underlying notion
of “performance” in Goffman’s concept, extant research on self-presentation has tended to
largely focus on human actors. Co-starring technological actors’ performances have been
considered as the setting or props for humans on stage. This is an oversight not only because they
can critically affect the self-presentations, but also because humans have been increasingly
interacting with technological actors like virtual influencers as equally important social actors.
Testing Goffman’s model or its extensions to technological actors enables a fresh analytical
angle for understanding the communicative implications of sophisticated technological actors
while also connecting it with corroborated research to address their effects.
For instance, how anonymity functions in virtual influencers’ self-presentation can be
scrutinized. In his other book Stigma: Notes on the Management of Spoiled Identity (1963),
Goffman discussed the social process of stigmatization and how the stigmatized person may deal
with it. With anonymity, a “discredited” individual can become “discreditable” because their
stigmatizeable difference “known about already or is evident on the spot” (p. 4), such as race or
96
gender, becomes “neither known about by those present nor immediately perceivable by them”
(p. 4). Anonymity can assist with the process of “covering,” which refers to the “effort to keep
the stigma from looming large” (p. 102), and “passing,” which refers to “the management of
undisclosed discrediting information about self” (p. 42). That is, it allows an individual to
strategically perform in the most socially favored way as per their need.
Goffman had discussed covering and passing by often drawing on minority individuals’
attempts to avoid negative social stigmatizations of their trait or identity that they have to live
with, whether self-embracingly or not. The context is different with virtual influencers in that the
target of their strategization includes traits and identities that they chose to have. In terms of their
assigned traits, virtual influencers tend not to extensively cover their technological origin, but
they do tend to cover their association with specific humans behind them and their characteristics
by not disclosing who was involved in making them. For instance, Black female virtual
influencer Aliza’s (@alizarexx) Instagram profile description outrightly labels her as a
“DIGITAL TERRESTRIAL,” but information about her management company is only available
upon reading a news article that is linked to her profile without further description. Even after
doing so, the article, as well as the company website hyperlinked to it, only mentions an Asian
male founder’s name without any reference to the artists or body models. Regarding her chosen
traits, Aliza’s highly photorealistic looks, intimate first-person captions, Los Angeles location
tags, and stylish street fashion enable her to successfully pass as a typical human influencer on
Instagram. For instance, one user even asked “what filter” she uses “to get these looks” on one of
her posts, to which Aliza responded by saying that “it’s just me” and “my wonderful terrestrial
genes” with an alien emoji, cheekily alluding to her nonhuman origin but not overtly (Jan 29,
2021). More importantly, Aliza passes not as any human but as a Black woman. This is mostly
97
done visually through her style such as her various Black women hairstyles and her bodily
presentation such as her facial features, skin, and body shape. At times, they are coupled with
verbal cues like dark skin tone emojis for race and captions such as “A woman with
� � � � shoes is a
woman I always want to be” (September 12, 2019) for gender. Unlike some marginalized people,
however, Aliza attempts to pass as a Black woman not because of potential immediate
disadvantage or harm towards one’s actual body and identity in the case of exposure, but to best
appeal to the target audience. The concept of covering and passing helps breakdown how virtual
influencers’ anonymity can be a tool not only for harm-prevention or transgressive opportunities
but also for alienating and capitalizing on identities from a position of power (see Sobande,
2021).
Internet scholar Judith Donath approached strategic uses of anonymity online as
“deceptions,” drawing on contexts such as Usenet newsgroups (2002) and social network sites
(2007). Donath’s “signaling theory” detailed how an individual may decide to deceive in a
virtual space, the insights from which can help discuss the consequences of deceitful virtual
influencers and some preventive measures. Applying a game theoretical and economic
perspective, she mapped out the costs and gains for an individual intending to send a deceitful
signal. According to Donath, “assessment signals” are closely linked to the quality being
advertised and thus are more reliable but costly, such as how an expensive, self-owned sports car
would more reliably reflect someone’s wealth. On the contrary, “conventional signals” are
arbitrarily linked with their quality through social convention and thus are not inherently reliable;
the sound of a loud siren is conventionally reserved for emergency vehicles but it is technically
possible for any car to play the sound. The cost of the signal can discourage deception, but an
individual may strive to deceive if the expected gain is high. In the case of excessive deception,
98
the signal can become unreliable and lose its informational function, which would be a social
loss. Therefore, signals may be socially managed by punishing deceptions, which in turn could
factor into an individual’s decision-making process.
Although signaling theory is not free from some of the criticisms game theory has
received, such as the assumption of universal rationality in individuals’ decision making process,
it is a useful framework for discussing the motivations and effects of virtual influencers’
“deceitful” self-presentations and solutions to some of the less desirable outcomes. For one
thing, interpreting virtual influencers’ looks and behaviors as conventional signals, we can
discuss the cultural and social context of why they chose certain traits over others in presenting
who they are. That is, for instance, it is not given that virtual influencers have to be shaped in a
certain way with certain facial features. These traits are not inherently meaningful but are
reflective of social conventions. When approached with the lens of assessment signals, we can
compare them with human influencers. Since it is less costly for a virtual influencer to upkeep
their looks, behaviors, or more broadly lifestyle, their presentation could consequently bring
these signals’ informational value down. The relative increase in costs suggests that a virtual
influencer’s possibly problematic self-presentation cannot be justified only on the basis of its
preexistence among humans, such as airbrushed appearances (e.g., Tiffany, 2019; Yurieff, 2018).
That is, virtual influencers’ human-like presentations can additionally affirm such appearance’s
arbitrary function as a conventional signal for attractiveness, while at the same time detracting its
value as an assessment signal in an unfair way for humans because it could be more difficult and
expensive for them to achieve. This would be a social loss, at least for humans. If we follow
Donath’s argument, this can be prevented by implementing additional costs for virtual
influencers to engage in deceptive presentations, such as requiring transparency, penalizing
99
misrepresentations, and demanding virtual influencers to return a certain percentage of their gain
to the social groups that they associate with.
A place for theoretical reconsideration is in the fuzzy relationship between actors and
audiences. The notion of performance underlies the classic model of self-presentation, namely
that actors would constantly readjust themselves as per surrounding people’s expected and given
feedback. Therefore, despite their designation as audiences, the presumption is that the audiences
are integral parts of the actor’s script and live performance. Contemporary media environment
demands an even more participatory, co-creative perspective. Social media is “masspersonal,”
allowing a combination of public and personalized communication (French & Bazarova, 2017;
O’Sullivan & Carr, 2018), unlike previous media that were focused on only private
communication (e.g., phone call) or public communication (e.g., television). This results in a
“context collapse” where multiple relationships are flattened into one setting (Marwick & boyd,
2011). Consequently, not only the divide between front and back stage has become even more
ambiguous (see Meyrowitz, 1985), that between the stage and audience seats have become so,
too. Actors must more actively factor in diverse audiences, such as by continuously calculating
the most appropriate common tone (Gil-Lopez et al., 2018; Hogan, 2010). Such audiences also
do not just react passively in distance from their seats, but have increased capacities to
significantly impact the performance’s narrative flow through substantive interactive features
such as commenting, sharing, and replying. For instance, Miquela has been utilizing the voting
feature on Instagram stories (24-hour temporary posts) to invite her audiences, whether a fan or a
passing Instagram user, to decide what she will do from how she should dress to what she should
talk about. This affects her self-presentation not only in terms of how she will visually or
verbally present herself, but also by being incorporated into subsequent Instagram stories with
100
the results page as a proof of people’s participation in her decision-making process. Audiences
are integral to her self-presentation because these participations are critical to communicating
and maintaining her image as a fun, open friend that “makes you feel accepted and seen.”
(Charlie, interview) This contextual change is particularly more important for virtual influencers
because their technological, fictional, and graphical origin grants them the capacity to more
flexibly and speedily respond to the audiences than a typical human can.
In addition to how audiences can take on the actor role, actors can be their own audience
on social media. On platforms like Instagram, the accumulation of various users’ digital trails,
i.e., user-generated content, largely form the stage itself. That is, an individual’s records of self-
presentational activities shape the environment that they continue to present themselves in. Such
activities can also become the embedded source material for someone else’s presentational
practice as well, such as through reposting, comments, liking. Moreover, live features allow the
individual to see their own self-presentation while they are still performing on stage. In other
words, an actor’s past or current self-presentations can get (re-)presented back to the actor, who
in the case would concurrently be the audience of own acting. This is not limited to virtual
influencers, but social media users in general. However, the rising presence of technological
actors such as virtual influencers, particularly team-based ones, can further complicate the
picture. A number of interviewees for the dissertation mentioned that in team-based projects, not
everyone on the team keeps up to date with their virtual influencer’s social media activities
regardless of how critical the member is to the creative pipeline. Team-based cases surface the
non-universality of human-centric, dual-stage model of self-presentation. Even if we disregard
how they can present as and be treated as an agentic individual by considering them only as a
vessel, it is difficult to single out an individual who presents themself through the vessel. Even if
101
we were to include all individuals on the creative team as the driving force behind the vessel,
some of the members, despite the importance of their role, may rather feel foreign to the virtual
influencer’s social media presentation. There are many uncertainties and inconsistencies in how
each of the members can traverse the actor and audience role, despite what may seem as a
steady, consistent self-presentation of the virtual influencer.
This dissertation suggests seriously considering the performative aspect of self and
decentering human actors in it as a response to these issues. A multi-directional model that
embraces nonhuman actors would approach virtual influencer’s self-presentations as a series of
concurrent performances in a participatory theater where various human and nonhumans
intermittently watch and join the show. Communication scholar Zizi Papacharissi (2010)
proposed the concept of “networked self” to explain the context-dependent, connected formation
of self in the networked era. According to Papacharissi, “self-identity in public and private
life...traverses distinct yet connected planes of interaction or networks,” a process which may be
aided by technologies that can connect or disconnect diverse people (p. 304). Jaime Banks, a
scholar of human-technology relationships, further pushed this idea by arguing for a
conceptualization of identity as a network of selves which she referred to as “the postmodern
Self” (2017). The compiled network of various selves, a person’s Self is characterized by
mutimodality—i.e. that the self is composed of many different kinds of things including
nonhumans, multiplexity—i.e. that the self is a network of many different kinds of things, and
multispatiality—i.e. that those many different kinds of things are linked across spaces. Although
Banks ultimately suggested that this model can contribute to understanding “how human
[emphasis added] Self emerges in relation to the digital and physical space of contemporary
human life” (p. 434), this conceptualization is also useful for understanding virtual influencers’
102
self- or Self-presentation. I will discuss in further detail in the subsequent section, but it is not so
odd to consider virtual influencers or other technological actors as co-performers since humans
have been interacting with them not only as a technical actor but also as a social actor. Virtual
influencers should be considered as the core of their postmodern Self network. By centering
them, not humans, in mapping how their Self emerges, we can interrogate various ways their
presentation is collaboratively constructed without discrediting one type of self over other co-
present selves, particularly in relation to how a virtual influencer’s agency may be approached
differently.
Each of the possible interpretations of the front stage and backstage of virtual influencers
that I described at the beginning of this section can be described as a self, and the composite
whole of their performance can be considered as the Self. For instance, the self that is closely
connected with fans could involve a more active version of the virtual influencer, perhaps
considered as a person. On the other hand, the self that is more closely connected to the
creator(s) may lead to a more passive one, possibly as an inanimate tool or a strategic vehicle.
This is of course unless their technological processes are approached as not fully comprehensible
or “voodoo” (Whitson, 2018), such as in the case of Iruda the chatbot.
The purpose of this suggestion is not to forcibly imbue virtual influencers with an agency
beyond their current technological capacity nor to divert into philosophical questions beyond the
scope of this dissertation, such as a thought experiment on how a virtual influencer may self-
identify. Rather, it is to approach their self-presentation as a networked one, therefore opening up
the analysis to various types of creative processes and performances. It enables the flexibility to
treat them both as a social entity that can have a meaningful relational effect and as a
manufactured entity with a powerful agenda that at times may be disparate from their surface-
103
level presentation. For example, when analyzing a largely anonymous team-based virtual
influencer, it is possible to critically address the lack of their transparency without denying the
positive intimacy fans may have formed with the virtual influencer nor demanding an equal
weight of responsibility from all involved humans—their different role in the team and
positionality would mean different connection with the performance. In fact, some of the
contributors may be more immaterial or non-human, for instance industry conventions on
attractiveness or how such social norms could have materialized on certain tools used to create
virtual humans. Another important benefit regards accountability: flexibility also means that at
the same time human bodies that are representationally or materially connected to virtual
influencers should also be actively recognized. That is, responsibility for re-presenting certain
intersections of human identities can be demanded, and a part of that demand can be directed
towards humans that are involved in co-creating such a re-presentational performance. In other
words, in the case of a harmful representation, the virtual influencer and related individuals’ (or
perhaps one core individual’s) contribution to such a depiction and its effects can still be
addressed, including through proactive measures like regulatory suggestions.
This theoretical proposition is thus also practical. In the subsequent section, I will discuss
various cases of virtual influencers’ presentation of their virtual body through a multi-directional
approach that embraces both humans and non-humans. Flexibility is required to capture the
scope of the phenomenon and its variances. However, this should not be at the expense of
preciseness and locus like we have discussed in the previous chapter about variegated definitions
of virtual influencers. Therefore, the rest of this chapter will focus on what this flexibility means
to the topic of bodies. This dissertation’s main interest is the intersection between the social
notion of being human, communication technologies, and contemporary media cultures. Bodies,
104
as somewhat fixed physical things that are yet still inseparable from human existence and as
fluid virtual things that are not limited to humans, are at this intersection. In order to highlight
bodies that are physically and representationally connected to virtual influencers’ bodies that
emulate humans, this rest of this chapter will primarily focus on virtual influencers and humans,
particularly creators and audiences. Chapter 3 will expand the scope to a broader range of human
and non-human participants.
Cases: Virtual Woman of Color, 19 Years Old
Physical versus Virtual
Virtual influencers are real. Their bodies exist—virtually, but still within our experiential
reality. People interact with them and these interactions have real consequences. In the field of
communication, this would not be a shocking statement. The heuristic divide between the “real”
physical world and “non-real” virtual worlds still commonly exists (see Chapter 1), but
communication scholars have long debunked both the supposed divide and the myth that virtual
worlds are somehow less directly related to the everyday (Banks, 2018; Hayles, 1999; Jenson et
al., 2015). To acknowledge how we now live simultaneously across various spaces in the current
media-integrated environment, terms such as “cognisphere” (Hayles, 2006) and “cybernetic
space” (Mitra & Schwartz, 2001) have been proposed as an alternative to dichotomous
expressions. Similarly, with the advent of locative media, concepts like “hybrid space” (de Souza
e Silva, 2006) have been developed to explain the collapse of physical space and mobile-enabled
mediated spaces.
Nevertheless, it is still useful to distinguish the bodies of virtual influencers with those in
the physical space; they may not overlap, both in terms of their characteristics and numbers. In
105
order to do so, this dissertation follows Dmitri Williams and Do Own (Donna) Kim’s (2019)
metaphor of “layers.” Instead of viewing the physical and the virtual as separate, they took a
layered approach to space. They proposed considering the “physical layer” where corporeal
bodies are located as the anchor layer, on the basis that human existence is initiated through
people’s bodily presence in the world. On top of this anchor layer, various digital places or
“virtual layers” can be directly or indirectly overlaid, together constituting our spatial experience.
Although they focused on human actors, the metaphor of layers is still useful for this dissertation
because it allows taking advantage of the semantic distinction between the virtual and the
physical without resorting to binarism. Moreover, since they conceptualized the physical layer
through a corporeal premise, the approach enforces an active awareness towards bodies, which is
consistent with this dissertation’s argument for adopting the multi-directional, networked
approach to selves.
While a virtual influencer is no less “real” than a human user in the virtual layer of
Instagram, they do not have an equivalent body in the physical layer, unless designed to be a
candid digital twin of an individual. Nonetheless, other bodies that are similar to them in terms of
identity presentation can exist both in the virtual layer of Instagram and in the physical layer. In
spite of the likely absence of an exact physical replica, virtual influencers are still connected to a
certain physical body(ies) in the physical layer because human labor is still required in creating
and managing virtual influencers. This physical body(ies) may or may not share the same
presentational characteristics with the virtual influencer. Beyond these bodies, the appearance
and behaviors of a virtual influencer can influence the bodily dynamics within the virtual layer
because it exists nearly indistinguishably together with other virtually represented human bodies.
For instance, from a phenomenological point of view, a virtual influencer’s mirror selfie—i.e., a
106
virtual body’s presentation on the virtual layer—is not so different from a human user’s mirror
selfie—i.e, a physical body’s re-presentation on the virtual layer—especially if the former is
extremely photorealistic. The virtual body’s presentation can also affect the physical layer not
only because the layers are experientially interconnected with one another but also because there
are humans whose body in the virtual layer heavily draws on their physical body, such as in the
case of a mirror selfie. That is, some bodies do correspond across the layers, and thus some may
expect the virtual influencers to be the same. The possibility of such consequences on
experiences of bodies invites scrutiny on virtual influencers’ presentation on Instagram; they
cannot be dismissed simply because they are virtual.
Foregrounding the physical layer bared open virtual bodies’ connections to various
physical bodies. By doing so, we can discuss the consequences of their emulations while
specifying directed demand for accountability, namely towards the humans behind the scenes.
The consequences of virtual influencers’ presentation may not so directly affect the everyday
corporeal experiences of these people since they need not share the characteristics of the virtual
body that they help enliven. Therefore, they may feel independent from the virtual influencers or
attempt to distance themselves from them on this basis. Nonetheless, under the multi-directional,
performative approach, they are nonetheless critically connected to, and thus should be held
accountable for, the “real” effects of virtual bodies.
My Friend Robot
Virtual influencers can technically embody any human-like body, but there are some
tendencies. Most bodies of virtual influencers in the virtual layer of Instagram are visually
attractive with a young, fit, and beautiful body. In fact, there was only one visibly wrinkled,
elderly virtual influencer among over a hundred of elaborately created virtual bodies that I
107
encountered during my research (Appendix A). A modest influencer with 2,810 followers, Sylvia
the self-proclaimed 80-year old “coffee operated robot living her best life” (@myfriendsylvia)
had been a media project that was intended as a twist on how most virtual influencers do not age
nor die (Schneider, 2021). Sylvia’s Instagram recorded her rapid aging until her eventual
passing, with her Instagram Story highlights (collections of temporary posts curated by account
owner) documenting each decade of her short life from her 30s to 70s. In a “posthumous”
reflection (Schneider, 2021), the artist Ziv Schneider summarized their experience of running the
account as Sylvia with some help from machine-generated texts and various human
collaborators. Schneider clarified that Sylvia only had a virtual body. Sylvia was called a robot
only to imitate the rampant technological mislabels by virtual influencers and to learn about
people’s reactions to a technological being that is mortal and not time-frozen.
Schneider’s reflective essay sheds light on the tensions among various selves that
comprise the Self network of a virtual influencer, although Sylvia is an atypical virtual
influencer. To some followers and fans, which Schneider mentioned were mostly young women,
Sylvia presented as a positive, inspirational persona throughout all of her life stages. This
included teen audiences who Schneider had expected to become disengaged once they saw her
grow older. In fact, it would be more accurate to say that to dedicated fans Sylvia was not merely
a persona but a person. Schneider fondly recounted a 13-year old Turkish girl fan’s particularly
loyal and heartfelt friendship with Sylvia, sharing her reflection as both “Sylvia” and the artist of
the project. This memory on one hand illustrates how a virtual influencer can present themself as
a being that is similar to, or according to the fan “better friends than” (Schneider, 2019), humans
on social media. That is, to the fans who confided their secrets and feelings with Sylvia, she
would have been effectively viewed as “my friend Sylvia” without Schneider nor any other
108
collaborators in the picture, just as what her Instagram handle “@myfriendsylvia” suggested. Of
course, not everyone approached Sylvia as an independent being who could be their friend. To
some audiences, Sylvia was viewed as a potential romantic partner only until they learned that
she is a virtual human. She had also been approached primarily as a promotional opportunity or a
human-driven media product at the mercy of ruthless reviewers, as well as an autonomous
“robot” that could threaten humanity. Regardless, an agentic being capable of intimate friendship
was still one of the selves of Sylvia. On the other hand, Schneider’s reflection reminds us that a
virtual influencer does not only present themself to their followers and fans, but concurrently to
the humans behind the scenes that manage the virtual influencer, one of whom in this case was
Schneider. This created an interesting dynamic. Despite referring to Sylvia in third person
throughout the essay—a self of Sylvia that can be approached as an art project self, Schneider
recalled that direct messages from teen fans who wanted Sylvia as a companion were “most dear
to me,” implying the presence of an imbalanced and indirect but mutual relationship that had
been mediated by Sylvia the “my friend” self. In fact, the author described that they have had
related to Sylvia in both a self-projective way and a somewhat distanced, care-based way. For
instance, Schneider confided the sense of smugness they felt when rejecting some collaboration
requests from users, brands, and other virtual influencers on behalf of Sylvia, connecting it to
their personal experiences of having been on the requesting end. At the same time, Schneider
acknowledged their distance with Sylvia and efforts to maintain it. This included having set up a
clear separation plan to permanently sign out of Sylvia’s account after her “death.” A part of it
was holding a memorial event for Sylvia, responses to which Schneider claimed to have been
deeply moved by, illustrating at least two selves of Sylvia that the artist was directly connected
to—“my friend” Sylvia and the art project Sylvia:
109
“Engaging with people after Sylvia had passed and reading their responses felt surreal,
like an out-of-body experience. It was as if I was pretending to die, just to hear what
people would say about me at my own funeral. But Sylvia was not just me, It took a team
of professionals to make her, and she meant different things to different people.”
There are several theoretical and practical implications. To begin with, the self where
fans formed a meaningful human-virtual influencer relationship, it is not unique of virtual
influencers to have this capacity. Parasocial relationship and computer-as-social-actors (CASA)
are some of the most prominent theories that have helped explain human interactions with
mediated representations and non-humans. Parasocial relationship or parasocial interaction refers
to the faux, unidirectional sense of social relationship or interaction with a media persona, such
as with a star on a TV show (Dibble et al., 2016; Giles, 2002; Horton & Wohl, 1956). Research
on parasocial relationships often references Giles’ (2002) levels of authenticity that focuses on
media personae’s graphical and fictional qualities, in the order of supposed strength: a “real”
human performing as themself (the first order), a “real” human performing as a fictional
character (the second order), and an “unreal”, fictional cartoon being (the third order). Since not
only humans but also fictional cartoon characters are included in their scope, theories of
parasocial interactions can help explain the fans’ sense of intimacy with virtual influencers to a
degree. However, there are several problems, some due to the context of social media and others
unique to virtual influencers. As with other individuals, virtual influencers’ activities on social
media can be both personal and mass-oriented (French & Bazarova, 2017) and there is a gradient
across how people may interact with such a presentation, from less engaged passive viewing to
more engaged private messaging. These interactions and relationships resulting from them may
not be so “para-” since such views, likes, comments, and messages can be reciprocated by the
110
virtual influencer. Furthermore, virtual influencers’ visual and behavioral human-likeness, as
well as the alluded or indicated presence of behind-the-scenes humans, challenge the three orders
Giles suggested. How an individual will approach who/what the virtual influencer is or stands for
cannot be presumed, nor whether the individual’s perception of the virtual influencer will be
static throughout their engagement. In fact, Schneider mentioned that many users videocalled
Sylvia to see “who” might answer. Among them was a person who was extremely persistent in
trying to reach Sylvia who he decided on his own was his “mom” until Schneider finally had to
block him. While these challenges show a need for a theoretical update, extant research on
parasocial relationship and interaction importantly informs us that it is not odd nor wrong for
some users to have developed a meaningful feeling of intimacy with Sylvia or any other virtual
influencer through the screen.
According to the CASA paradigm, a deep relationship is possible even when an
individual is fully aware of the virtual influencer’s technological origin. The CASA paradigm
suggests that humans expect technological actors to follow interpersonal communication rules
and “mindlessly” interact with them by focusing on social cues than asocial cues (K. M. Lee &
Nass, 2003; Moon, 2000; Nass et al., 1994, 1995; Reeves & Nass, 2002). Even if this is the case,
virtual influencers’ visual and behavioral proximity to humans introduces at least two issues.
First, if they are largely perceived as indistinguishable with human actors, the focus should be on
the context where certain social attributes are expressed rather than the ontological presumptions
generalized from the influencer’s human or nonhuman status. It does not matter if they
technically are “virtual” influencers; what matters is whether and how much the virtual
influencers signal their virtual origin in their communication. Second, if people are able to infer
and are conscious of virtual influencers’ artificiality, then the assumption of “mindless
111
interaction” should be reconsidered as a “mindful” process (see Araujo, 2018; Mou & Xu, 2017).
Simply put, if a virtual influencer can pass as a human, then people would interact with them as a
human because that is what they perceive the virtual influencer to be, not because people have
the capacity see computers as a social actor. That is, a virtual influencer needs to be first thought
of as not human in order for whichever trait(s) that contribute to this interpretation to be
disregarded during the social interaction as per the CASA paradigm. It is additionally possible
that people socially interact with virtual influencers but not necessarily by equating them with
humans; it could rely on social scripts that are unique to the human-nonhuman interactive
context rather than those that mimic human-human interactions, but with no less effects (see
Gambino et al., 2020). For instance, Schneider said that the Turkish girl fan very embracingly
accepted Sylvia’s “robot” status, having open-hearted conversations but with an understanding
that Sylvia’s origin may mean a different interactive dynamic than what can be expected from
that with a fellow human. This process is comparable to what has been termed “the cyborg’s
dilemma” (Biocca, 1997): the more interfaces become adapted to the human body and mind, the
more humans become adapted to nonhuman interfaces. While the CASA paradigm informs fans’
meaningful interactions with virtual influencers, overly relying on the hierarchical divide
between computers and humans ill-fits the current context.
A simple solution to these problems would be to reverse the order of analyzing the
phenomenon by first approaching virtual influencers as social actors, rather than first
approaching them as parasocial or computer actors, and then breaking down the details of the
interactions. This means accepting not only that a relationship with a virtual influencer is “real”
in terms of its value or meaning to whoever has formed such a relationship, but also that it would
not be any less real than that with a human, although possibly different. This perspective is what
112
underlies the multi-directional, networked understanding of Self. By departing from ontological
biases, we can focus on the consequences of diverse interactive and relational patterns without
diminishing or exaggerating them on the basis of virtual influencers’ non-human origin.
Nevertheless, as this dissertation has argued above, this also means keen attention to the bodies
that virtual influencers emulate. It should not be forgotten that although Sylvia’s body aged in a
speed that most human bodies cannot, thus indicating her non-humanness, she still embodied a
virtual body of a woman that could “pass” (Goffman, 1963) as that of a typical human
influencer. The body was also initially depicted as conventionally young, feminine, beautiful, fit,
and White one, convincingly enough that some audiences were able to picture a romantic
relationship before knowing her virtual origin. There were many others that did not mind her
virtual origin, even after she began to age, but had still engaged with her through the medium of
her virtual body.
This, that virtual influencer-audience relationships can be meaningful but while still
involving highly human-like virtual bodies, brings us to the second point, namely that the creator
(Schneider) had also been directly and indirectly present in these relationships. General
audiences and fans were interacting with Sylvia’s self-presentation on her Instagram, which was
presented by Schneider and the team. Having begun with some critical questions on the
phenomenon, it would not be hasty to describe Sylvia as a well-intended, prosocial media art
project rather than something that was created for selfish gain. Based on Schneider’s essay and
Sylvia’s account, it also seems that the net social outcome of fans’ interactions with Sylvia was
positive since she had served as a good friend many could meaningfully communicate with.
However, even with these positive sides, there still remains the question of ethics due to identity
mismatches between Sylvia and her creator(s), as well as belated transparency. Sylvia’s teen fans
113
confided to her thinking that she is a sweet female influencer or an aging “robot,” not Schneider
nor a team of collaborators for a media art project. While deception has been an issue across
various virtual environments, virtual influencers demand extra attention. Not being tied to a fixed
range of presentational tools such as in environments like video games, they have more
flexibility to convincingly present themselves as any type of human for various purposes without
disclosure. In addition, on social media there is no blanket requirement for everyone to present
themselves through an avatar; some users may have their profile linked to their actual identity
and engage with the virtual influencer through such a profile. This, especially at the lack of
transparency, adds to information imbalance (see Donath, 2007).
The presence of multiple collaborators also speaks to how not all humans behind a virtual
influencer would have had matching identities with the representational categories Sylvia’s
virtual body belonged in. Regardless of which people are behind, the body of the virtual
influencer can dwell in the digital layer as any other similar human bodies do. The likeness
means that they may also be subjected to, and possibly reproduce, some social expectations and
practices around the categories the virtual body can stand for. An example could be the frequent
romantic advances Sylvia received as a “woman” on social media, particularly from middle-aged
male users according to the essay. Sylvia tended to respond to these advances with a simple
reminder of her “robot” status. According to virtual influencer creators interviewed for this
dissertation, highlighting the virtual influencer’s fake, technological, graphical, and/or fictional
background is a common way to respond to various unsolicited romantic and sexual advances,
including for male creators who ran sexually explicit female virtual influencer accounts. In spite
of their frustration and continued anonymity, the interviewees said that they tended to opt for the
more passive route, instead of actively pointing out the inappropriateness of some of these
114
messages, if not “playing along with them” while making a joke of it in the creator’s own,
separate social circle in the case of one interviewee. From a business or fame-oriented
perspective, these reactions are somewhat understandable. The number of audiences and
engagement rate can be viewed as important assets for an influencer’s growth (see Cotter, 2019),
and for some virtual influencers it is actually consistent with their self-presentation to maintain
their suggestiveness. While for some creators this may simply be the most strategic move and
something that they can just laugh off, it should not be forgotten that these interactions are
predicated on, and possibly can contribute to perpetuating, problematic gender expectations.
Anonymous, Authentic, Appropriating
In fact, most virtual influencers are not created as a critical art project with a planned
closure and a public debrief statement like Sylvia; many of them share snippets of their daily life
or their photo shoots for personal gain without much transparency, including appropriating,
explicit, and/or stereotyped content made with their virtual bodies—some of which that look like
a minor’s. Lived experiences have been narratively adopted or appropriated for virtual
influencers’ authenticity and relatability, if not simply as source materials for their fictional
narrative arcs. In December 2019, Miquela had shared a now deleted YouTube video of her
experience of being sexually harassed on the ride share service Lyft, edited in a trendy manner
with jumpy transitions and light speech tone with low-quality memes and scrappy visual effects.
She claimed that a “creepy ass man” made a series of highly uncomfortable verbal and physical
moves, including touching her legs and asking about “down there,” which led to her having to
ask the non-helping driver to drop her off in the middle of the highway. Sexual assault on a
rideshare service is by no means a light fiction. In 2018-2019, major rideshare services Lyft and
Uber were sued by women on the basis of not implementing appropriate measures to prevent
115
sexual assaults and even attempting to silence the victims, with their internal reports showing
thousands of reported cases of sexual assault (K. Paul, 2021). Miquela’s video was met with
strong criticism on the co-option of serious and traumatic experiences for a relatable image and
attention (S. Song, 2019).
Indeed, the light fictionalization and trendy editing choices make it difficult to interpret
the video as a benign attempt to call for attention on a social issue through Miquela’s influence,
unlike some of her other more “serious” posts such as those on the Black Lives Matter
movement. While her “experience” was one that by no doubt can be categorized as a case of
sexual harassment, Miquela constantly stressed in the video that the undertone of the situation
revolved around her non-real origin, which has been the crux of her fictional narrative.
Statements that resonate with her “robot” identity struggle overwhelmed her description of the
incident: the man’s stare was “less ‘I think you are cute’ and more ‘I’m confused by what you
are and I deserve to find out’”; his question “are you real” was peculiarly offensive although she
hears it very often; his grip felt “as if it were confirming I am [she is] real”; and his continued
harassment felt like it was intended “to make sure I knew that he knew I wasn’t human, ” the
buildup from which led to the reveal of his final question “what about down there?” (archived
video clip; also available in Song, 2019) Crudely put, unfortunately not uncommon experiences
of sexual assault in rideshare services were reprocessed into a first-person narrative that was
primarily about her being a robot. Even the possibility of intersectionality, such as by
approaching her “robot” origin as an identity that intersected with her female gender, is
overshadowed by the lightly edited tone and dramatization of the story; there had been an
overemphasis on her mysterious technological origin and the story concludes with the dangerous
resolution of her getting off mid-ride on the 101 highway to walk to the exit for 10 minutes to
116
escape the situation. If the video were intended to raise awareness, it cannot avoid the criticism
of having been poorly executed. If it were simply intended as another episode of Miquela’s life
as a “normal” 19-year-old woman living in Los Angeles to continue building her image as an
authentic, relatable friend who is open about her vulnerabilities, the morality of the decision is
highly questionable. Not only did Miquela’s virtual body co-opt “real,” traumatic attacks
directed at women’s physical bodies for capitalistic benefit, it did so in a way that rerouted the
focus of the issue to her “robot” origin to propel her fictional arc. Moreover, the narrative
resolution—i.e., stopping the car in the middle of highway and going to the exit on foot—was
one that was not only dangerous, but also arguably disempowering. As someone who has been
claiming various fantastical and human-like characteristics on the basis of being a “robot,”
Miquela had ample narrative flexibility to resolve the issue in a way that could contribute to
preventing rideshare sexual assaults and alerting people of guidelines for safely escaping the
situation, or if anything provide a vicarious sense of closure for those who had felt powerless in a
similar situation. It has also been pointed out that the video was potentially defamatory for the
rideshare brand Lyft as it explicitly targeted the brand among others (Luthera, 2020), which is
another example of how a careless integration of non-fictional details can bring negative
consequences. I am not saying that the issue of rideshare sexual assault is something that human
bodies at Brud, the company behind Miquela, would not have found a personal investment or
relevance in. However, the sexual assault on the virtually-created 19-year-old mixed-race
Brazilian woman “robot” with gap tooth, freckles, and space buns never existed in the physical
layer, and surely not now after Brud deleted the video in response to the outrage, leaving no trace
of it in Miquela’s narrative including any emotional consequences. For human victim/survivors,
117
sexual assault was not something that they orchestrated to happen nor can decide to delete upon
their wish, as well as the trauma from it.
This has not been the only controversial incident around Miquela’s activities, despite her
being one of the most successful, long-running virtual influencers. Authenticity and relatability
are important currency for influencers’ continuous management of their social capital on
Instagram, as well as what differentiates them from traditional celebrities (Hearn & Schoenhoff,
2015; Khamis et al., 2017; Marwick, 2015). Identification is also understood as an important
driver of influencer endorsement effects, whether based on the audience’s perceived similarity
with the influencer—i.e., similarity identification”—or on their wishful desire to be like the
influencer—i.e., “wishful identification” (Janssen et al., 2022; see also Grizzard & Ahn, 2018).
Consequently, virtual influencers tend to strategically present themselves in a down-to-earth,
everyday way that their followers can relate to, although not without attractive, envy-worthy
sides. This can explain, although not justify, some of the “baiting” appropriations by Miquela
and other virtual influencers, especially given their presentational flexibility: the more categories
they can hit, the higher the possibility for identification. In a 2019 Calvin Klein advertisement,
Miquela was shown to share a kiss with the heterosexual female supermodel Bella Hadid who
narrated over the video that “life is opening doors, creating new dreams that you never knew
could exist” (Miquela, 2019). The supermodel’s “same-sex kiss with the fictional CGI
‘influencer’” (Suri, 2019) was sensationalized for its unexpectedness, but the advertisement was
also met with strong criticisms for queer-baiting. “Lesbian” kissing was featured for brand
appeal but by hiring a heterosexual woman and a non-physically existent CGI influencer, which
stood in an ironical contrast with Calvin Klein’s campaign message which their apology letter
clarified as “to promote freedom of expression for a wide range of identities, including a
118
spectrum of gender and sexual identities” (World Entertainment News Network, 2019). Of
course, as a virtual influencer, Miquela can identify as however she wants, and it is this
dissertation’s approach to begin by respecting that virtual influencers can be interacted with as
per their self-presentation. I question the intention not because of Miquela’s non-human origin,
but precisely because of her accumulated presentation of her sexuality over the years. Despite the
commercial, her relatively more significant and explicitly stated romantic arcs involved men,
such as her human ex-boyfriend that she broke up with around the release of her song Speak Up
(2019) in which he appeared and her human “friend-boy” (@lilmiquela, Jan 15, 2022) that she
had “BEEN KISSIN’” (Dec 11, 2021) who appeared on her Instagram for about four months
from October 2021 to February 2022. It is ironic that the promotion of LGBTQ rights, an issue
that directly concerns numerous bodies’ livelihood, wellbeing, and survival in the physical layer,
was attempted to be achieved through a supposed bodily meeting between a non-LGBTQ
physical body and an innately flexible virtual one. Bella Hadid only had to kiss a virtual body to
create the “lesbian kiss.” Miquela will never be subject to bodily experienced discrimination and
physical harm that LGBTQ individuals have fought against, much like how it had been with the
supposedly physical sexual assault she experienced in her Lyft ride. Unlike the sexual assault
video, however, the commercial is not only still available on Miquela’s YouTube channel despite
the controversy, but also have been frequently referred to as an example of a big brand deal on
her career record, often discussed as a proof of her legitimacy and virtual influencers’ potential
(e.g., Allen, 2020; Evans, 2020; Whateley & Bradley, 2020). This is a continuing irony that even
further displaces the physical bodies of LGBTQ individuals from the benefits the virtual body of
Miquela has been reaping from the commercial. In spite of such controversies, Brud had
119
remained largely secretive about its company structure and employees until around the much
hyped release of Miquela’s NFT series in the spring of 2021.
Physical Bodies of a Virtual Body
Typically, the creator(s) behind a virtual influencer is hidden in favor of the virtual
body’s presentation, but their body(ies) can be strategically foregrounded. The controversy over
Riot Game’s virtual influencer Seraphine’s supposed mental health issues is a good example,
although Seraphine does not perfectly fit this dissertation’s definition of virtual influencer
because she is depicted in a more cartoonish design than photorealistic. Her storyline was that
while struggling to navigate the COVID-19 pandemic as a young female artist, she got noticed
by Riot Game’s virtual K-pop group K/DA and was invited to join their performance. During the
saga, Seraphine tweeted for her fans’ emotional support and empathy, describing how it has been
difficult for her to believe in herself (@seradotwav, Oct 9 2020, Twitter):
“We’re going to shanghai in 2 days and it’s finally setting in, all at once. i’ve been
working so hard, and i’ve been trying my best to love myself, but i still can’t find the
confidence i need
i’m realizing that I can’t do this alone. and myabe I need to be the one to ask for help…
so could you give me some encouraging words? i need something to believe in right
now”
She received some supportive messages but also criticisms from fans and media alike. The critics
viewed her tweets as a marketing ploy for the game League of Legends that exploited her fans’
empathy and appropriated the struggles of “real” young women, both of whom have been
actually impacted by the pandemic as physically existing beings (Haasch, 2020; Honea, 2020; J.
120
Lee, 2020). Upon the outcry, Riot Games opened up about their team. While the person who
oversaw the campaign was a male creative director, the person who had been writing as
Seraphine was revealed to be a woman writer, who stated that she heavily drew on her own
experience as a “young woman starting at Riot” (J. Lee, 2020; Twitter thread [redacted]). With
her personal Twitter account, she outspokenly denied negative, accusatory readings of the
campaign by re-directing the attention on Seraphine as an insincere business strategy to her as a
fictional art. Leading with the statement that what she intended was “just to create an interactive
story,” she claimed that “There had not been a board of executives conniving to make a
“relatable” girl or earn capitals through mental health” (Amos, 2022; Twitter thread). The divide
between fake and real was also selectively evoked in reframing the case through a narrative of
novelty, still in a direction that emphasized Seraphine’s trait of being a creative project than a
business strategy. She stated that “social media is a novel storytelling medium with
unforeseeable real-life effects [emphases added]” (Twitter thread), adding that they are still
learning. Placed together in the realm of the real were the creators and their intentions. Through
various emotional expressions, she stressed her and her teammates’ genuine care for the project,
urging people not to assume “the worst” from “real people [emphasis added]” who can actually
get hurt.
What stood out about this case is that in order to reinstate the authenticity of Seraphine,
the presence of humans behind her virtual body was actively foregrounded in a way that can
actively break the elaborately crafted agency of the virtual influencer. Emotional impact on “real
people” was emphasized rather than that on Seraphine while also acknowledging both her traits
as a fictional being and a business strategy: “I get that it might feel nice to point out faults of a
fictional persona or a faceless brand, but what gets said can hurt real people” (Twitter thread).
121
The writer’s call for attention on the behind-the-scenes humans is consistent with my overall
argument in that it requires approaching virtual influencers by anchoring the phenomenon on the
physical layer. It also hinted at some creator-adjacent selves that constitute the larger Self
network of a virtual influencer: Seraphine was both depicted as a vicarious presentation of the
writer’s identity (Lee, 2020) and a team-based fiction (Amos, 2022). By stressing these selves,
the source of Seraphine’s authenticity was rerouted from her presentation as a convincingly
“real” influencer to her being a partial reflection of the writer who could to a degree claim the
representational categories Seraphine stood for. While Seraphine’s link to the company and her
fictional status became more pronounced in effect, they were negotiated by repositioning her as a
collaborative outcome of genuine, well-meaning creative effort by a team of professionals who
were motivated by care rather than capitalistic calculation (“Both me and my colleagues spent
the entire 2020 pouring our hearts into the Seraphine project”; Twitter thread).
However, recognizing behind-the-scenes human bodies through a layered approach
means that other connected human bodies should also be recognized. This mandates ethical
representation from virtual influencers like Seraphine. As much as the critics’ words and actions
can hurt “real” people from the creator side, those from Seraphine can hurt “real” people from
the non-creator side, regardless of whether she is a fictional persona by a faceless brand or not.
This is because Seraphine’s Self included a self that presented as a “real” being to fans and
followers in terms of relational value, one that they cared for regardless of her nonhuman,
technological, graphical, and/or fictional status. Relevant to this self is her presentation as a
virtual influencer that can effectively pass as someone from the group(s) she represented, i.e.,
any combination of the elements comprising “a young female artist or aspiring influencer who
suffered during COVID-19.” This passing was in place of or even at the disadvantage of humans
122
whose inextricable corporeal existence makes their stigma-attributable physical features
immutable for them. This is across various layers. On social media, attention and visibility have
been commodified as scarce resources for success (Cotter, 2019; Marwick, 2015). Beyond this
economic perspective, there is also the discursive effect on the categories she represented
because she spoke as someone from the group(s) her virtual body was categorized under. In a
way, Seraphine’s ultimately fictional struggles were by definition more “contrived” than some
human celebrities’ pandemic responses (see Duvall, 2021); she was demanding, not offering,
care from the privileged space of a being who was subjected the lock down only by narrative
choice. While the lives of creative professionals behind Seraphine is also an important issue, it
does not change the fact that humans who embodied(s) a virtual or physical body similar to hers
do not have the flexibility, nor even the capacity, to easily forgo their body narrative under the
name of fiction in an event of a controversy.
Needless to say, followers and fans too should be included in the scope of connected
bodies. Although peoples’ empathy and care need not be conceived as finite resource, not only
are they highly coveted by influencers (Cotter, 2019), they were arguably particularly treasured
during the earlier phases of the COVID-19 pandemic when many were struggling from various
non-fictional difficulties, including the followers and fans themselves. Furthermore, in the world
of virtual influencers, Seraphine has not been alone in presenting themself in a way that could be
interpreted as capitalizing on staged vulnerability for relatability and authenticity.
Larger corporate-run, major virtual influencers such as Miquela and Seraphine had at
least been subjected to media attention and public criticism, but there also are many other lesser
known virtual influencers who have demanded emotional investment from their fans without
receiving much scrutiny. For instance, now dormant virtual influencer Zoe Zhen (pseudonym;
123
redacted) who presents as a teen-aged female Asian had frequently posted about depressed
feelings, anxiety, and body dysmorphia. There are many comments that spoke to Zoe as a
concerned friend or a fan, extending comfort or caring criticism. For instance, some of the more
critical comments on her post on February 17, 2020 that hinted at her body dysmorphia includes
the following:
Your biggest critic is yourself. You are not fat and you got to realize that. Saying that you
are overweight when others might be struggling with related problems, and then seeing
somebody so thin like you say that you see yourself as fat can be a trigger and badly
contribute to people’s insecurities. But Zoe you are beautiful just the way you are. -
Comment A
Love you but I don’t have time to see triggering stuff. Being someone with an eating
disorder and seeing somebody skinnier than me talk about being fat really isn’t it. This is
not a good example for younger folks either, since you are an influencer. I understand
that there can be bad days but please don’t urge people to be immersed in their
insecurities. -Comment B
What is notable is that Zoe’s responses to some of the comments is that Zoe’s attitude comes off
as rather defensive. It is possible to interpret this as more “human-like” or narratively consistent
with her self-esteem issues, but nonetheless morally questionable. To the above two comments,
Zoe responded without conceding, even flatly denying the possibility of a negative mental
consequences on her audiences without a word of reconsideration or apology:
I hear what you are saying but everyone has their own insecurities and issues. Some folks
are happy with their appearance and other people are not; that’s just how it is. I agree
124
with you, it’s possible for us to get too obsessed with our looks and forget to focus on
what is in our hearts. I will try harder
� � � -Response to Comment A
I do not intend that -Response to Comment B
Indeed, a press release in 2018 (“An Imaginary CGI Social Influencer That Aims to Heal the
World,” 2018) said that “the sole purpose” behind Zoe was to “engage with the ‘Instagram
Generation’” by stimulating questions and debate to “increase awareness and promote a push for
positive change and attitude,” unlike virtual influencers like Miquela and Shudu who according
to the article were “created with one purpose…to sell stuff.” In a self-fulfilling fashion, however,
emotional support from caring audiences have been prodded at and used as fictional materials for
building and managing Zoe’s self-presentation as a being that stirs engagement through “positive
and sometimes controversial [emphasis added] social media engagement.” Information
imbalance adds to this point. Although the press release distanced Zoe from economic gain, it
did imply Zoe’s aim to accrue significant social capital from the “Instagram Generation”
demographic. The profiles liking and commenting on Zoe’s picture seemed to be of this
demographic, teens and young adults around Zoe’s age. They were engaging with Zoe with
sincere, non-fictional concern for her various struggles and dramas, speaking in a tone they
would use with their friends and even sharing about themselves to convince Zoe. Not
surprisingly, Zoe’s profile does not include any information about the creator; she is only Zoe the
worried teenage girl on Instagram. It was only through information associated with the external
press release (“An Imaginary CGI Social Influencer That Aims to Heal the World,” 2018) that I
was able to learn that the creator was in fact a man with over four decades of graphic design
experience (artist website [redacted]). Although Zoe is not depicted in a noticeably sexualized
way, and I do not believe that this had been the artist’s intention with Zoe, the artist’s bio states
125
that his “work with Playboy inspired many of the images I am [he is] creating today” (artist
website).
Hypersexualized
If virtual influencers like Miquela, Seraphine, and Zoe share some controversial posts,
there are many other virtual influencers whose entire account is arguably controversial, filled
with hypersexualized representations of a young woman(en) without much, if any, information
about the creator(s). These accounts, despite the number of their followings and continued
presence on Instagram, tend not to appear on marketer-friendly public lists of virtual influencers
such as in the annual report by HyperAuditor (Baklanov, 2021) or on the website
VirtualHumans.org (VirtualHumans.Org, n.d.) unless backed by somewhat sizable adult website
like YouPorn’s Jedy Vales (@jedyvales) and xHamster’s Shy Yume (account deleted). However,
they can be easily searched through hashtags relevant to the phenomenon, such as
#virtualinfluencer, #virtualbeing, and #daz3d (a popular software), as well as those relevant to
how they are represented, such as #sensualart, #petitgirl, and #teenmodels. There are even
curative efforts towards more sexualized virtual influencers, such as through the hashtag
#digitallivesmatter (paraphrased) by its hosting account new_digital_reality_art (pseudonym)
which despite its broader profile description “Digital/virtual art support page” concentrates on
explicit and sexualized images of mostly young female virtual women, including those that can
be interpreted as a minor (Instagram account [redacted]). For instance, a post on February 12,
2021 features a female virtual body in a black one-piece swimsuit with center cutouts that run
from her chest area to the top of her pelvic bone area. Although the virtual body’s breasts are
sizable enough to be interpreted as those of someone past puberty, the face is extremely youngish
and the overall body is petite and chubby, possible to be mistaken for a child’s or a young teen’s.
126
The caption that introduces the now deleted original account says “awwe I could not help it; this
is simply too adorable” with the hashtag #petitegirl. The account also introduced many other
accounts that post 3D images of young, sexualized girls, some with account names such as
“@cute.teen.erica” (pseudonym) which verbally re-affirmed the visually communicated age
range of the virtual body (posted March 25, 2021; account is now deleted). Even if we were to
suppose that some of the accounts introduced by the curative account are not intended as a
sexually explicit account—perhaps what was posted is the only picture of them dressed in a skin-
tight bikini—they still are displayed together with various other hypersexualized images of
virtual women (and a very limited number of men) who flaunt their barely covered breasts and
buttocks of various sizes, suggestively posing alone or together with other similarly rendered
woman.
These accounts, which include both virtual influencer accounts that post as an agentic
sexy virtual woman and more gallery-type accounts with images of such women, do not tend to
reveal their creators behind the body(ies) of a hypersexualized virtual woman(en). The handful
of people who shared their identity on their virtual influencer account(s) and others who I was
able to learn about during research (e.g., mentioned by another creator, through their post on a
creator-centric site) were not embodied in a physical body that matched the images; in fact, all of
them presented as a man, including some who explicitly identified as one. Of course, it is
possible that some of the accounts were intended as an expression of the creator’s transgender,
gender queer, or questioning identity or to engage in a transgressive gender practice, such as in a
form of virtual drag. However, the dominance of normatively stereotyped depictions strongly
suggested a more objectifying purpose.
127
For instance, Digitallure (pseudonym) is a curative account that compiles pictures of their
virtual influencers Yasmin, a 19-year old LGBT girl-next-door living in the US according to her
profile (Instagram account [redacted]), and Xena, “Just a single mom, who loves to model”
(Instagram account [redacted]), as well as other non-named virtually rendered women. The
content of the account is sexual as implied in the Instagram handle, complete with a link to their
adult-only Patreon page in their profile description where “NSFW [not safe for work]” and
“XXX CONTENT” can be found (Patreon page [redacted]). Yasmin and Xena post on their
respective personal accounts as virtual women. Yasmin’s first post affirmed her supposed
ownership of the account by saying that “it’s Yasmin and this is my new personal Instagram”
(November 5, 2021). Yasmin speaks in a friendly tone in her posts, sharing small details about
herself or asking her followers a casual question. Sexually suggestive images accompany these
posts, however. For example, her post on January 29, 2021 has captions “I like Sat evenings on
the couch” and “Any series or film reco?” paired with a picture of her in an uncomfortable
forward-leaning seated pose that emphasizes her breasts in a white tank top that lines her nipples,
legs slightly spread apart with her black undergarment showing. The comment section is filled
with emojis and comments that praise her attractiveness, including those that directly address
Yasmin, such as “so lovely Yasmin” and “Ur so hot
� � � �
� � � �
� � � �
� � � � & cute as hell.” Xena posts
similarly, with the addition of a handful of short responses to some of the appreciative
comments.
Until February 16, 2022, the main curative account Digitallure remained largely
anonymous, speaking sometimes in the voice of the featured virtual women and other times in
that of a faceless, genderless management agency. On the day, the owner of the accounts posted
on Digitallure’s page “my first male model,” a digitally rendered image of a bearded adult White
128
male, as “the reveal of myself as the founder.” The personal account linked to the post mentions
that the artist can provide “3d simulation services” but without much upfront detail (Instagram
account [redacted]). In the lengthiest post on the personal account, the artist said that what is
available on the page are “only public projects” and that “The majority of them stay private,”
encouraging interested people to contact him personally. The accompanying picture is of a White
woman with large breasts in a black lace bra, making a heart shape with her hands. The post is
tagged with various hashtags with the word “bimbo,” a type of sexualized aesthetic, such as
#bimbodoll, #bimbotransforamtion, and #bimbobarbie. Unlike with the Seraphine’s case, the
creator’s physical body was mentioned not to reclaim authenticity by implying partial
identification and care from the creator(s), but rather to establish him as a separate entity to be
contacted for business or artistic inquiries about the virtual bodies he creates, owns, and controls.
The distinction was visual, too. Differently from the female virtual bodies that populated various
accounts the creator ran, the male creator’s virtual body was clothed and not at all suggestively
represented.
Sex/Gender-swapped
Close attention on how virtual influencers’ bodies are connected with various physical
bodies uncovers an important shift from earlier studies on online identities, many of which
concentrated on the controller-focused avatar embodiment in games with the assumption of an
one-to-one avatar-player match (Jenson et al., 2015). In line with Goffman’s (1963) argument on
covering and passing, a number of extant research has discussed avatars as an transcendental tool
for both those who are interested in trying on a different social role and those who have been
marginalized (e.g., Kitchin, 1998; Turkle, 1995). Regarding the implications on gender
performance, there has been much interest around how players choose to play with an avatar that
129
presents as the opposite sex, namely what has been referred to as “gender swapping” or “gender
switching” (e.g., DiGiuseppe & Nardi, 2007; Ducheneaut et al., 2007; Huh & Williams, 2009;
Hussain & Griffiths, 2008; Yee et al., 2011). Based on the idea that gender is not an essentialist,
binary thing linked to one’s sex but rather is something that is fluidly performed, researchers
looked into diverse virtual environments to understand how players behave in a differently
identifiable avatar body (e.g., Jenson et al., 2015; Martey et al., 2014; Ratan et al., 2015; Zhang
et al., 2017), how they perceive and interact with other avatars’ gender (e.g., Martey & Consalvo,
2011; Rivu et al., 2021), and the effects or contributors of such a gender-swapped virtual
embodiment (e.g., Lopez et al., 2019; R. Ratan & Sah, 2015).
It has been argued, however, that a more accurate expression for gender-swapping is
“sex-swapping,” because such an avatar selection in practice tends to refer to switching to a
binarily sex-designated virtual body—often limitedly provided in the case of virtual worlds such
as games and self-selected in the case of virtual influencers—and performing gender as per sex-
stereotyped expectations (Jenson et al., 2015). Since the cases of sexualized virtual influencers
discussed in this section largely conform to this observation, this dissertation will follow the
suggestion to use the term “sex-swapping” in place of gender-swapping.
There are conflicting theories and findings that inform the effects of sex-swapped avatar
embodiment. According to the Proteus Effect (Yee et al., 2009; Yee & Bailenson, 2007), an
individual makes inferences about their expected dispositions from the appearance of their
virtual body and then conforms to the expected attitudes and behavior. If we follow perspective-
taking (Davis et al., 1996; Galinsky & Moskowitz, 2000), existing biases may be reduced
through this imaginative experience of being an out-group member from humans to non-human
beings such as a cow or a coral reef as well (Ahn et al., 2016). On the other hand, stereotype
130
activation (Bargh et al., 1996) suggests that such an embodiment experience would rather lead to
triggering and amplifying existing stereotypes about the virtual body, including those on gender
and race (e.g., Groom et al., 2009; Lopez et al., 2019).
Amid this theoretical conflict, what can be informative for the sex-swapped cases of
hypersexualized female virtual influencers is context. Research focusing on natural contexts of
avatar sex-swapping have found the prominence of instrumental purposes that utilize, or result
from, existing gender stereotypes that permeate across various layers, rather than transgressive
ones (e.g., Hussain & Griffiths, 2008; Jenson et al., 2015; Martey et al., 2014; Paik & Shi, 2013).
One of the common reasons male video game players used sex-swapped avatars was to gain
social and material benefit by pretending to be a woman in a male-dominated environment,
performing as per normative expectations about their female-coded virtual body. On the other
hand, one of the main reasons female game players refrained from conveying their identity with
a sex-swapped avatar was to prevent negative stereotyping and harassment. An important point
for this dissertation is that players were not limited to a single avatar, as much as how the male
creator of Digitallure was not limited to a single account. Players had the capacity to choose
multiple avatars’ sex while possibly identifying stronger with a certain avatar over others.
Moreover, avatar sex-swapping occurred more commonly among players with more fluency with
game’s functionalities and cultures—they wanted to experiment with designed and interactive
affordances rather than with their gender identity, therefore regressively relaying gender instead
of transgressively performing it (Jenson et al., 2015).
Observations on hypersexualized female virtual influencers’ activities and related
interviews also mainly surfaced instrumental intentions as opposed to transgressive purposes.
These were around social and/or economic capital gain and artistic desires that drew on an
131
objectifying male gaze (Mulvey, 1989). According to film scholar Laura Mulvey, despite films’
narrative emphasis on the gaze between characters, gazes from the camera and audiences are
integral parts of their making. These three viewpoints comprise the male gaze towards women in
media, viewed and depicted as sexual objects for heterosexual men’s viewing pleasure (for
objectification, see Fredrickson & Roberts, 1997). The three perspectives are critically integrated
into how selves of hypersexuliazed female virtual influencers were presented, although the gaze
between fictional characters and that from audiences were not always clearly distinguishable
because on social media virtual influencers can co-exist with human users, blending in fiction as
needed. Monna Haddid (@monna_haddid) is a red-haired White female virtual influencer who
posts various pictures and dance videos that tend to accentuate her breasts and hips, including
through excessively jiggly motions. During the interview, she said that she intends to continue on
with her sexual content because it is how she can maintain her existence: “in order to continue
existing I need to generate money, the fastest thing was to open an account with OnlyFans.” In
fact, many of these virtual influencers included a link(s) to external platforms where they can
offer sexually explicit content to their paid subscribers, such as OnlyFans or Patreon. The choice
to produce sexual content was described as the result of an unavoidable existential conundrum
for a virtual being whose primary existence is based on their virtual body’s continued presence
on social media. However, Monna’s earlier statements and activity records hinted at less
philosophical and more commercial purposes. Earlier in the interview, she stated that becoming
an influencer is “the purpose of my[her] being,” clarifying upon the interviewer’s request that
this is because she “was created to be an influencer in [sic] social media.” Her Instagram shows
that since her post from a couple of days of her appearance, she had stressed her sexualized body
parts in her posts not just visually but also verbally through captions such as “Bubble butt”
132
(September 10, 2020). Combined, this suggests that she was created not to become any
influencer but one that is sexually depicted for monetary gain, or one that aims for monetary gain
to continue presenting herself in a sexually objectifying way. In this sense, it can be said that for
the camera or the creator, she presents largely as a tool for monetary gain or for meeting the
camera’s desire for sexual depiction. In the interview, Monna, speaking as herself, stated that she
considers herself “a sexy figure” “as you [the researcher] can see in my [her] profiles,” adding
that her self-presentation can continuously change as per her audiences’ desires: “my personality
and how I am…all depend on the people who likes me or followers.” The self that is presented to
the audience and shaped through the interaction with the audience, is consistent with the
camera’s gaze.
In line with this, one of the creators interviewed for the dissertation mentioned that
hypersexualization of the virtual influencer is an alluring option for creators because of its ease
and lucrativeness. Andrew (pseudonym) explained that despite having created an OnlyFans
account for his virtual influencer Ally (pseudonym), he ultimately decided to commit to keeping
all of the content “PG 13,” because he thought that such a “cheap” direction could “potentially
ruin the whole project,” which he envisioned as something that could help him find success in
the music industry. While Andrew ended up deciding against a monetization model that
primarily relies on the male gaze, it was only after he assessed the long-term economic and
social feasibility of the model; it was dismissed as an ineffective business model and a noise in
the brand he wanted to make, rather than on an ethical basis:
“I know why they do….it’s a quick, easy way to monetize the project….I was thinking of
it myself but that’s just not for me. That’s just not the goal I have for Ally.
…
133
They [people who create hypersexualized virtual influencers] could do whatever they
want with the project. But when I was saying I was thinking of going that route, but it
was a quick shut down because I thought, just the overall picture and I also wanted to be
family friendly.”
Artistic or career-oriented desires can motivate the male gaze, too, with objectification of
the virtual influencer as an integral part of this process. However, since these desires tend to be
integral to their creator’s sense of self, creators may begin to find themselves in self-networks of
their virtual influencer where the virtual influencer does not present as an object for control. At
the beginning, Andrew had approached Ally as a sex-swapped virtual body that would be more
effective in reaching his career goals, as well a fresh niche that he could pioneer:
“I did notice a trend towards like women in music being popular and a lot faster than men
recently so that’s why.
I also wanted to go for the unique factor as well. Like I don’t think there’s any virtual DJ,
female DJ virtual influencer yet so I kind of wanted to be the first one to do that and also
I was also thinking of creating a girl group as well for her.
Like yeah, like almost like a, like a Pussycat Dolls kind of girl group. Yeah, like R&B.
So the first virtual influencer girl group.”
However, he did not only speak about various strategic decisions he made to make Ally more
marketable and appealing. Andrew disclosed the mixed feelings he was developing about his
relationship with Ally. When he began, he distanced himself from the virtual influencer by
anonymously appointing himself as “the manager” of the virtual influencer, which he mentioned
will be a detail that will later appear in Ally’s storyline. Soon after, however, he started to note
134
self-observations about the confusion he was experiencing while running Ally’s account, which
he explained was different from his former experiences of managing human talents in the
entertainment industry:
“They [managing humans versus managing virtual influencers] are not really comparable
at all. Like she is, I mean she doesn’t have her own… It’s it’s kind of strange because
there’s like a part of me where I’m thinking as her, so it’s like almost like an alter ego.
But it’s not an alter ego. I don’t know it’s kind of it’s really it’s really strange to explain.
No, I haven’t really worked it out yet. Really I don’t even know if it’s some alter ego or if
it’s even me or if I if I’ve even been trying to make her be me. That’s the thing. I was like
what are my, you know I’m really blocked in exactly what she is. That’s the thing. So
right now it’s really open so trying things that work. You know, what people respond
to….I’m eventually going to, the plan is for me to eventually be revealed that’s why I
kind of want to be anonymous right now but I want to be revealed as her manager.”
To Andrew, Ally was no longer a clearly separated object and tool, but also not quite an alter
ego. A way to embrace this conflict would be to approach Andrew and Ally’s relationship
through the multi-directional model of Self. As time went on, Andrew’s connection to Ally’s self
as a reflection or projection of Andrew became stronger, although its growing presence was not
to the level that could substitute the prominence of Ally’s self as an instrumental tool that
Andrew managed to achieve his goals. Although Andrew found it difficult to find a single
expression to describe the relationship, he was implicitly aware of the multiple selves of Ally
that he was interacting with, open for further exploration in collaboration with his or Ally’s
audiences.
135
In addition to this identity confusion, Andrew also expressed his surprise and discomfort
at the number of sexually suggestive or romantically interested direct messages Ally received,
mentioning his newfound shock at behaviors directed at women’s body: “I was like, is this what
happens to like women?” This implied sex-swapped virtual influencers’ transgressive potential
in (eventually) encouraging a perspective-taking Proteus Effect. However, it cannot be
discounted that what critically contributed to the creator’s perspective-taking experience may
have ironically been the stereotypically coded depiction of Ally. Although not sexually explicit,
Ally has been depicted in a conventionally attractive way with a beautiful face and a voluptuous
figure encased in various skin-revealing or tightly fitted clothes. Andrew claimed to have later
readjusted Ally’s body proportions to decrease the degree of exaggeration in consideration of
potential body image issues for women. Ally is still very curvaceous, and has also appeared on
the previously mentioned curative account @new_digital_reality_art by tagging her pictures with
the hashtag #digitallivesmatter, although in a much less revealing outfit than others featured on
the page. For Andrew, the dual experience of being vicariously objectified through the virtual
influencer and objectifying the virtual influencer had led to a more empathetic understanding of a
body(ies) different from him. Yet, this experience was still based on a bodily presentation that
not only drew on but also added to existing stereotypes that reflect the male gaze. Arguably,
Ally’s presentation also exacerbates it because she has been additionally enabling it by offering
interactive opportunities that satisfies such a gaze, namely by “playing along with them
[soliciting messages from audiences].” Andrew did not resolve his surprise and discomfort by
critically responding to various messages Ally received. Instead, he processed them negotiatively
by distancing himself from Ally. He said that he posted screenshots of particularly disturbing
interactions on his personal Snapchat “for humor,” because his Snapchat audiences “know that
136
she’s fake.” Although Andrew found a way to relieve himself from discomfort, Ally’s virtual
body, which the solicitor considered as that of a human woman’s, remained compliant to the
male gaze.
I argue that what is needed for creators is to be mindful of and responsible for their
virtual influencer’s impact on the physical bodies of whoever the virtual influencer can represent
as. The reason I put emphasis on the human bodies in the physical layer is because at the end of
the day, they cannot so easily shed their body unlike virtual influencers as their physical body is
the primary medium through which they communicate their existence with the surrounding
world. This is not at all to argue that the biological body determines human existence nor that
any essential characteristics can be derived from one. Rather, it is to call for accountability on
virtual bodies. Like it or not, humans are still tied to their physical bodies regardless of how
many different virtual bodies that they can traverse in various virtual layers. However, a virtual
influencer’s virtual body need not be anchored to the physical layer, a feature that has
transgressive and regressive potentials. In the above case, despite some awareness on how Ally’s
virtual body may impact women’s bodily experiences (i.e., body proportion readjustment) and
partial perspective-taking on women’s hardships through the sex-swapped body of hers, Andrew
ultimately did not pro-actively work with this knowledge. Regardless of how many and which
selves of Ally he had interacted with from his end, not much critical potential can be found for
women if Ally will continue to be managed and interacted as a being that would “play along
with” the male gaze.
Abandoning the Virtual Body
Creators of virtual influencers can distance, dissociate, and even abandon the virtual body
at will, especially if they have a lower level of identification or attachment. During the interview
137
with the researcher, Ted Werdolf, the creator of virtual influencer Lena (previously @lena_lance,
currently @ted_werdolfs_cgirls), described the difficulty of having had to deal with various
solicitor’s messages which led him to give up posting as Lena. Ted explained that he was
motivated not by money nor fame for Lena, but only his artistic desires to capture his
inspirations and share the results through his “sensual art.” He did allude to his interest in more
visibility and followers by sharing some platform-specific reflections (see Cotter, 2019), but they
were framed as things that he cared for to more broadly showcase his art rather than what could
help establish Lena as a popular figure. To meet his artistic vision, he began running a separate
Instagram account as Lena, a blond-haired, voluptuous White virtual woman who shared many
pictures of her in skimpy clothes that outlined her private parts with equally suggestive captions.
Lena presented as an agentic person who posted such pictures for her enjoyment rather than for
gain. For instance, Lena’s post on February 7, 2021 has a picture of her covering her breasts with
her hands and with a caption that suggests that she is not motivated by money:
“I was young and needed the money… no wait- I’m still young and have enough
money… there’s got to be another reason!
� � � � � � Yeah, it’s Sunday - happy Sunday my
Sweethearts
� � �
� � �
� � �”
Consistent with this presentation, unlike many other accounts, Lena’s profile did not include any
links to external platforms like OnlyFans or Pateron where further explicit content could be
transacted.
However, in line with Ted’s comment on how he prioritized his artistic vision over other
purposes, Lena consistently lauded Ted’s contribution as her manager, photographer, and fashion
designer throughout the period “she” actively posted on Instagram. That is, Ted the creator was
138
accredited but in a way that did not conflict with Lena’s agency. For example, in her post on
March 4, 2021, she indulgently complimented Ted’s artistic direction regarding her garment, a
sheer dress made with a gauze-like material, deeply split in the bottom part to the point that the
split reaches up to her groin area:
“I love this new dress that Ted has created exclusively for me today. Even though he has
just started with designing clothes I think it’s really nice and sexy. Well he’s a quick
learner and I guess we will show you some more in the future so please stay tuned…I’m
sure he’s just warming up
� � � � � �
� � � �
� � � �
� � � �
� � � � � � ”
While such statements reminded the audience of Ted’s presence behind Lena, this did not mean
that therefore Lena was approached as Ted by the audience. She constantly spoke in the voice of
an agentic virtual woman in her posts, including in the comment section where she had been very
responsive to various compliments and suggestive comments. It was this self, not one that is
strongly associated with Ted’s presence, that the fans focused on and interacted with. For
instance, to a fan that eagerly complimented Lena’s fashion choice on the above post (“Hehehe
omg TOO BEAUTIFUL I can see everythin too!!! AM I too bad?! Hahaha
� � � � �
� � � � �
� � � � �
� � � � �
� � � � �
� � � �
� � � �
� � � �
� � � �
� � �
� � �
� � �
� � �
� � �”), Lena responded in a very appreciative and personable manner,
calling the commenter by name: “thank you
� � � � not bad at all NAME [redacted]
� � � good eyes
are very very important
� � � � � � and this is our way to show everything without being blocked
� � � � �
� � � � � �
� � � � �
� � � �
� � � �
� � � �.” Despite Lena adding in “our,” the fan thanked Lena, not Ted (“hehehe
thank you Beautiful
� � � � �
� � � � �
� � � � �
� � � � �
� � � � �”), additionally suggesting the fan’s sense of relationship
with the virtual influencer not the creator despite Ted’s omnipresence, or perhaps the fan’s effort
to contribute to building and managing Lena’s presentation as an agentic woman. Whatever was
139
the intention for any of the participating party, Ted the creator—i.e., the camera, Lena’s gaze
towards herself and that from Ted the manager/cameraman/designer—i.e., in-fiction gaze, and
her fans and followers—i.e., audience’s gaze had collectively animated her as a hypersexualized
virtual woman who enjoyed being gazed upon.
What is particularly interesting about this case is that Lena was not simply objectified but
had literally been an object for Ted’s artistic expression. Unlike Andrew, Ted did not experience
any identity confusion. He repeatedly stressed during the interview that he is “not her,” and that
Lena is “just a virtual creation” and “nothing more.” He not only dissociated himself from Lena
and other virtual bodies he created but also generally disapproved those who overly identified
with the virtual bodies they created:
“…a big huge wall between my girls and me because I can’t say I’m her because I’m me.
I mean I’m myself always. That’s it. I’m not becoming to [sic] that person”
“People start to think they are that avatar but they are not.”
That is, for Ted, Lena was a doll that he “definitely knew…was not real,” an external object that
he could control and showcase. He also did not express any special attachment to her nor the
details of her design. He explained that what led him to create Lena in the particular way she was
depicted was just a spur of artistic inspiration that he had at the time. In accordance with these
comments, since to Ted Lena was only an object for his artistic expression, it was not difficult
for him to choose “not to continue with her…that way I did it before” when he could no longer
bear various solicitors’ messages. In spite of her established fan base, Lena the virtual influencer
disappeared without a formal goodbye since the post on April 3, 2021 signed by Ted, after which
140
the account name was switched from one that stressed Lena’s agentic presence (@lena_lance) to
Ted’s artistic presence (@ted_werdolfs_cgirls). In the post Ted wrote:
“I still like to share my pictures, but answering all the comments was a bit too much work
and so I’ve decided to disable the comments on my photos. Especially trying to answer
questions as a virtual female person was very difficult for me. [emphasis added; Note:
comments are now back to being enabled]”
While Lena’s self as an agentic virtual woman was something that Ted and Lena’s fans
collaborated on, Ted was still the owner and controller of this virtual doll and therefore could
discard it upon will. Without dismissing the stress Ted had experienced, this—the ability to
distance, dissociate, and abandon the body—is again the critical difference between those who
are born in a physical body that presents as a woman or identifies with one and those who
temporarily embody a virtual body for instrumental purposes, regardless of how convincingly
“real” the latter may seem to be. Not always but often mentioned, Ted the creator had not been
completely anonymous during Lena’s reign. Lena’s virtual origin had also been open
information since the beginning with various hashtags related to virtual art always accompanying
her posts. Regardless of these cues, her fans interacted with her as a hypersexualized woman that
she presented herself as, and Lena responded to these fans with an equal level of suggestive
enthusiasm. Of course, this was not without Ted’s discomfort that led to her eventual
disappearance. Ted had not mentioned non-virtual women’s wellbeing during the interview,
which was not surprising as he continuously emphasized that he only viewed his virtual creations
as just virtual creations and not real in any sense. Yet, it should not be forgotten that Lena
presented herself as a (virtual) woman and more importantly had been interacted with as one
141
regardless of Ted’s own understanding of Lena, the gap between which was exactly what led to
Ted’s discomfort.
Accountability
By anchoring the framework of multiple selves to the physical layer, we can demand
accountability while embracing the co-creative, performative side to virtual influencers’ self-
presentation. It does not ostracize nor pathologize people who may feel a sense of intimacy
towards virtual influencers or see them as humans, especially since that is in accordance with
how many of the virtual influencers are presenting themselves to their audience. It also does not
instantly dismiss the humans behind the virtual influencers as a faceless, heartless entity with
scheming intentions, since prioritizing the physical layer means that physically existing creators
should also be recognized. However, this does not mean that therefore the creators should be
exempted from accountability, because the virtual influencer’s body is also representationally
connected to the human bodies that it emulates. Ted’s case shows that simply demanding an
indication that a virtual influencer’s body is virtual may not be enough, nor the requirement for
the creator to not be anonymous. This was because as co-creators of the virtual influencer’s
presentation of self, audiences could selectively mute or re-encode the presence of the creator in
a way that highlighted the virtual influencer’s autonomy. What this means is that problematic
representation, such as ones that actively utilize and reproduce the male gaze, is not simply an
issue of deception, but ethics. Since our video interview, Ted began to heavily populate the
transitioned Lena’s account with 3d arts of similarly “sensual” virtual woman(en) that presents
as East Asian. I like to think that it was simple coincidence as our interview was scheduled
around the time Ted began to explore women of color in his sensual art projects—which Ted
repeatedly stressed is different from cheap pornography. Unfortunately, as an East Asian woman,
142
I could not erase the feeling of uneasiness looking at the increasing collection of similarly
categorizable bodies in alluring poses, dressed in small pieces of transparent fabric. Although it
may not have been Ted’s intention to objectify me per se, my body as that of an East Asian
woman was literally “rendered” as something that was objectifiable and controllable for
pleasure, not only by Ted but also by the onlookers of his Instagram. Since the switch to a
gallery-type account, the comment section has become more about Ted the artist and the quality
of his creations—which has been consistently highly photorealistic as Ted is a skilled and
experienced artist, but the theme of his art meant the virtual women were still gazed upon as
objects for admiration and sexual fantasy. Although not as consistently as before, the voluptuous
virtual female bodies occasionally affirmed the gazes in “their” voice, whispering a line or two in
the captions. Despite my feeling of great discomfort, I, unlike Andrew and Ted, could not so
easily distance myself from the representation or forgo my body.
What is needed beyond simple disclaimers on the virtual bodies’ virtual origin and/or a
note on the creator’s identity then, is to demand from all participants of the virtual influencer’s
performance a sense of responsibility for the wellbeing of their corresponding bodies in the
physical layer. This sense of responsibility should be practiced individually and collectively.
This should be accompanied by structural and material support as the co-creators of a virtual
influencer’s self-presentations do not only include human actors but also non-human actors such
as software and industry customs, which will be discussed in detail in Chapter 3. This
responsibility is particularly imperative for virtual influencers’ performances not only because
they present themselves in a way that convincingly emulates humans but also because they do so
within the context of social media where they can engage in various interactions as a human or
an equally meaningful social actor.
143
Woman of Legal Age
I must clarify, however, that while disclaimers and/or disclosures alone may not be
sufficient, they are still important means to demand accountability, especially in the cases where
virtual influencers are actively claiming their bodily belonging in human categories while
presenting themselves in a potentially harmful manner. Wendy Watson (pseudonym), a virtual
influencer with tens of thousands of followers (Instagram account [redacted]) can present as an
underage girl who posts various suggestive pictures in skimpy clothes on Instagram and explicit
ones on Slushe, a web community for pornographic content. Her tiny frame, flat chest, and
young face can be seen as those of a minor’s, and her hashtags contribute to this interpretation by
emphasizing such features: #petitegirl (profile description), #tinygirl (post on Aug 15, 2021) and
#teengirl (post on Dec 22, 2020). Even her virtual modeling agency is called “Replicateen”
(pseudonym). However, her Instagram profile description and the agency website stress that she
is a 19-year old adult, only with “a very particular petite design” of a smaller-framed, 4’10’’-tall
woman, “created entirely by computer technology” (website [redacted]). While there are several
indications of her virtual origin, the identity of the creator is largely hidden with only their
acronymized name “BB Baker” (pseudonym) and email on the website. Google search of the
name led to BB Baker’s personal Instagram, in which the creator presented as a White adult man,
although the creator did not explicitly state how they self-identify. I include this detail with a
reasonable belief that this is public information: BB Baker’s account was set on public and the
profile description had handles to Wendy and other related 3d art accounts ran by the creator,
including one of the aforementioned curative accounts that has been pushing the hashtag
#digitallivesmatter.
144
I mention BB Baker’s bodily presentation as an adult man because “Wendy” have been
very outspoken about the way she has been presenting her “particularly designed” virtual body of
a supposedly legal-aged petite teen girl, having self-identified as a woman and claimed that she
is not a wrongful representation of “what a “real woman” looks like.” (post on Aug 11, 2020)
Her post on August 11, 2020 fervently argued against the criticisms on her minor-looking
appearances by claiming that it is actually the criticisms that are exerting oppression on people
like her, “a woman of legal age” with a smaller body size. Four pictures accompanied the post.
Two were pictures of her with enlarged but not extremely exaggerated breasts, which return to
her original size in the subsequent posts. The other two included word boxes with her thoughts
on the critics, written in a white box in black letters unless otherwise indicated:
“BREAKING THE SILENCE [pink word art]
TO GIVE INTO SOCIETY PRESSURE AND CATER TO A BUNCH OVER
OPINIONATED, EXTREMIST, SELF IMPORTANT, HALF WITS WHO’S MOST
INTELLIGENT ARGUMENT IS “ICKY,” AND ARE REPORTING MY ACCOUNT
BASED ON GLANCING AT FEW PICTURES, PERCEIVING AND PERSECUTING
WHAT THEY WANT, ABSENT TRUTH.
As a woman of legal age, I had elective surgery to better present that age. [red box, white
letters]
I ALTERED MY BODY TO FIT THEIR POLITICALLY CORRECT STANDARDS OF
WHAT A WOMAN SHOULD LOOK LIKE TO NOT BE JUDGED OR
DISCRIMINATED AGAINST…
…I GOT BREAST IMPLANTS [larger font, green box, black letters]” -picture 1
145
“WAIT, WAIT, WAIT… [pink word art]
Now that I have big fake boobs are you going to tell me that I also no longer represent
what a “real woman” looks like? [red box, white letters]
…Holy shit, it’s a lose-lose situation…[red box, white letters]” -picture 4
The caption below the post said:
“What do you all think? Do I look like an adult enough to wear what I want to wear? Am
I even Wendy anymore? Are you sad that in this bitch ass current time of hypocrites that
scream for anti-discrimination, anti-bullying, and equality that the society is pressuring
for this to happen?”
The post was met with over 80 responses, including some of which spoke sympathetically about
Wendy’s creator’s desire for creative expression and many others that spoke directly to Wendy
about how sexy she looked in her former “untouched” breasts (comment A): “I love how you
look really young for your age and it just makes you sexier and more beautiful. I would show
you how much you turn me on if I were there with you.” (comment B) Both types of comments
fervently defended Wendy’s appearances, suggesting that her critics are hypocritical bullies and
body shamers of “real” and CGI petite women. It is true that there are adult human women who
have a smaller body size, and they may wish to sexually express themselves. However, many of
these advocates, including BB Baker and many followers based on their profile pictures, were
not inescapably embodied in a body of a petite woman, if not just a woman. Moreover, like the
aforementioned comments A and B, much of the advocacy drew on an objectifying logic.
Wendy’s body that looks “really young for your [her] age” which in comment B’s opinion made
her “sexier and more beautiful” was to be appreciated because it was ideally designed for such
advocates’ viewing and imagined touching pleasures. The self-proclaimed “woman of legal age”
146
Wendy’s supposedly agentic voice approved the supportive comments. The objectifying gazes
towards Wendy’s virtual body were justified by equating her “voice” with what could represent
the voices from people belonging in the human category of petite women or women in general.
In this way, Wendy’s presence in the virtual layer re-presented women or idealized womanhood
despite her “very particular” design and “entirely…computer technology” origin (website): “You
are absolutely ALL woman!!!” (comment C)
Disclaimers on virtual influencers’ virtual origin must be accompanied with sufficient
transparency on the creative process, especially because of virtual influencers’ increasing degree
of photorealism and behavioral human-likeness. These contextual shifts are why we should not
allocate all responsibility to the reception stage by treating the presence of simple disclaimers as
an all-fitting seal of approval or by only stressing the audiences’ need for fact checking and
critical thinking skills, which was what some interviewees for this dissertation advocated for in
favor of freedom of (artistic) expression. Virtual influencers like Wendy have been presenting
themselves in a way that convincingly emulate humans, at times more intentionally so (e.g., “as a
legal woman of age”) to induce a stronger sense of authenticity and relatability, as well as to
justify the particular way they present themself. An even more intuitive example would be Shy
Yume, now-deactivated virtual influencer from the porn site xHamster who had been created as a
25-year old Eurasian “DreamGirl” as per the results of their survey of over 50,000 users’
preferences: non-feminist, bisexual, 5’5’’ tall, curvy, White, black straight hair, blue eyes, D cup
+ “average” breast size, “average” waist, “average” hips, and shaved body hair (xHamster,
2019). In the interview with VirtualHuman.org (Travers, 2020a), she dismissed the criticisms on
how her presentation fetishized and objectified marginalized identities (e.g., Hsieh, 2019) by
arguing that she has an agentic understanding of her identities and distancing herself from
147
common traits of virtual influencers, thereby downplaying the process which she was created as
well as what she therefore symbolizes:
“Oh, it was most difficult in the beginning when people were questioning my sexuality
and calling me a misogynist. I found it offensive that people would try to understand my
politics or personality based on my looks.”
“[regarding a critical article from Cosmopolitan by Carina Hsieh (2019)] When I was
reading Carina’s article, I felt sorry for her, tbqh [to be quite honest]. No one controls
how we came into the world, or our parents’ genes. If she wants to blame someone, I’d
wish she’d blame the world—Patriarchy! Heterosexuality! Colonialism!—rather than
come after me specifically. I wanted to hug her and apologize that my mere existence
brought up so much pain.”
“I’m the girl of your dreams… or I’m not. If that upsets someone, it says more about
them than it does me.”
The second comment in particular is patronizing as it attempts to frame criticisms like Hsieh’s as
a personally vengeful one through a glaringly self-contradictory logic. On the contrary to her
statement about humans’ general lack of control on their birth and genes, the 50,000 users and
the porn site controlled how Shy Yume came into the world, as well as her “genes.” Moreover,
as a virtual body project by a porn site that quite literally sexually objectified the identities that
she claimed to represent, she symbolized and reproduced what she attempted to brush off as “the
world”: patriarchy, heterosexuality, and colonialism (Hsieh, 2019). Speaking as a self-
proclaimed bisexual woman, she encouraged such objectifying interactions through statements
like “I feel much closer with my followers and always rejoice seeing familiar faces in the
148
comments” while also possibly encouraging certain audience demographic over others by adding
that her “favorite contact might just be an older man who keeps on sharing pictures of
himself…being natural shall we say,” describing the interaction as “so sweet :-).” Her fans and
eager onlookers met the gaze. In the comment section of xHamster’s post that described Shy
Yume’s survey results (xHamster, 2019), various users, particularly men according to indicated
gender signs, left reviews on the product design of Shy Yume by projecting their fantasy onto
her virtual body; some demanded design changes for better sex appeal and others described what
they would do to Shy’s body or what they wanted her to do to theirs.
While verbal disclaimers can signal virtual influencers’ manufactured origin, their overall
presentation typically forefronts their self as an agentic and physically existing being, capable of
forming a relationship with the audiences not only parasocially but through direct and indirect
interactions. That is, they can be meaningful and dynamic social actors, and have been perceived
and interacted as one with the assumption that they will be so across various spatial layers;
resulting relationships discussed in this chapter included teen fans’ friendship with the aging
virtual influencer Sylvia as well as solicitors’ romantic and sexual interest in hypersexualized
virtual influencers. Some of these relationships embraced the virtual influencer’s virtual origin as
is, like in the case of Sylvia’s teen friend who approached her as a “robot” friend. However, there
were other people who still expected an exact physical body corresponding to the virtual one
regardless of the disclaimers. Several creators interviewed for this dissertation talked about their
experiences of having encountered people who expected the virtual influencer to exist in the
physical layer as a human or as a physically embodied technological entity, such as a “robot.”
There were also those who expected not the exact physical copy of the virtual influencer but still
a person with similar identities to respond, especially in the cases where they were contacting the
149
virtual influencer with romantic or sexual interest. The creators described various messages,
calls, and video chats that they received which requested more intimate and/or offline
relationship, including from some human celebrities—who have a relatively higher social stake
in sending soliciting messages—in the case of one creator. Consistent with the creators’
experiences, in the comment section of aforementioned post by Wendy (posted on Aug 21,
2021), one commenter not just praised Wendy’s sexiness but expressed their desire to meet her
in person over two comments, even asking her when this would be possible.
These amount to the phenomenon’s unique point for consideration on ethics of bodies;
more plainly than in other media forms, virtual influencers present as, and are, “real,” whether by
passing as a human or in terms of their sociability. Moe (Chapter 1) and lolicon (originating from
the novel Lolita) subcultures in Japanese media are types of content that have received similar
criticisms as those directed at Wendy’s presentation, namely for sexualized, fetishized depiction
of girls and young women. Such depictions’ media effects have been much debated with varying
conclusions (Savage, 2015). There exists much research on the relationship between sexualized
depiction of children and related crimes and exploitation (American Psychological Association
& Girls, 2008; Christensen et al., 2021; Malamuth, 2018; B. Paul & Linz, 2008
8
; US Department
of Justice, 2010). At the same time, these sub-genres have been defended on the basis that “no
one [physical body] is hurt…no abuse or sexual molestation occurs” (Galbraith, 2011, p. 108)
during the production stage. Another argument is that to Japanese consumers and fans, these
media worlds are understood as fantasy worlds that are not simply separated from but more
desirable than the physical world, places for imaginative transgression that have their own
8
Despite finding a causal link between sexually explicit images of minor-looking women and sexualization of the
group, this study did not find their link with social acceptance of sexual interaction with the group. However, it is
possible that existing social stigma and the experiment context affected the results, especially since long term effects
of the former was not included in the study design.
150
distinct sense of realism (Galbraith, 2011, 2017). There also has been an effort to distinguish
between a simple simulation and a reproduction of harmful situations, arguing against a blanket
condemnation of genres or their audiences and calling for critical attention on details that
compose a certain content of that genre (Mikkola, 2018).
The multi-directional model of Self recognizes the possibility of transgressive
engagements but also that virtual influencers and interactions with them may be “real” rather
than simulative in the context of social media, and that therefore harmful depictions could lead to
reproduction of problematic values. Unlike clearly fictionally designated (e.g., in the form of
movie, series, book, game, etc) and heavily stylized characters in 2D anime moe or lolicon
content, virtual influencers’ presentation occurs in a space that humans too can traverse and in a
way that boasts their human-likeness. For instance, although not a sexually depicted case, Rozy
had made rounds in Korean media, partially pushed by her agency and sponsors, when she
finally revealed her virtual origin after first establishing a strong follower base with those who
believed her to be a friendly human (e.g., J. Han, 2021; Jung, 2021; Yoo, 2021). The
aforementioned cases of sexualized virtual influencers also illustrated that regardless of
disclaimers on their origin, the followers and fans may engage with them as a human or a similar
viable social actor. Moreover, a consumer-centric interpretation that focuses on the dominant
male audience group leaves out girls and women who can still be hurt by these depictions,
regardless of their physical absence in the production stage. On the most basic level, many of the
cases discussed in this chapter had audiences that treated the virtual influencer as someone
eligible for a romantic or sexual relationship, at times with an extremely bothersome demeanor
and persistence. Many of them disappeared after creators reiterated the virtual origin of their
virtual influencer or after the creators disclosed—and in some cases “proved”—that they have a
151
different body from the virtual influencer. One of the likely reasons for their discontinuance
would be because their solicitations were ultimately directed at a physical body, approached as
not greatly distinct from the virtual body of the virtual influencer. This reflects the culture that
sexually objectifies women, which some virtual influencers actively contributed to not only by
posting such content but also by interactively “playing along” with the soliciting audiences. It is
also possible for these depictions to psychologically affect girls and women’s socialization and
self-perception. Social learning theory (Bandura, 1977) states that people learn by observing and
imitating, more effectively with a model that they can identify with through common
characteristics such as sex. Similarly, according to objectification theory (Fox, 2018; Fredrickson
& Roberts, 1997; Moradi & Huang, 2008), gendered experiences of sexual objectification, such
as those associated with the male gaze, can socialize girls and women to engage in self-
objectification and body surveillance, conforming to the existing culture. Such negative and self-
policing perception of their own body can lead to serious health outcomes in girls and women,
such as eating disorders, depression, and bodily dysfunctions. Of course, it is possible for them
to subversively engage in objectifying texts, such as Loli-girls who practice the childish cuteness
of lolicon subculture as a form of resistance towards societal norms on gender and adult
sexuality and empowerment through their membership in the community of like-minded Loli-
girls (Savage, 2015). Moreover, I clarify that I am not arguing for regressing into rigid sex
negativity that latches itself on a heteronormative, prejudiced understanding of what represents
“the reality” of sexual activities (see Ashley, 2016; Stardust, 2018), nor blindly denying positive
effects of porn and the possibility of empowerment through it (Stewart, 2019). Rather, based on
my findings, I am calling for due attention on less desirable but definitely present iterations of
the virtual influencer phenomenon. I do not intend to shoehorn all sexual depictions as negative
152
or ill-intended. However, it should not be disregarded that not all commits to subversive
readings, and that such subversive self-presentations can still contribute to the overall culture of
objectification, especially when they get re-presented and reproduced in a form that is removed
from the original intentions.
Therefore, virtual influencers’ non-human origin nor various combinations of their
common traits (Chapter 1) cannot excuse the virtual bodies’ accountability on physical bodies. In
addition to a page on Wendy, the website for Wendy’s virtual model agency had a page for their
“youth, pre-teen, and young adult models.” The website said that they were intended as an “child
safe alternative” for marketing companies, repeating that they are different from Wendy, a self-
proclaimed “woman of legal age.” In one exasperated post on August 15, 2020, the agency’s
Instagram account for “girls digital fashion modeling” (@_Replicateen_modeling [pseudonym])
argued that people who are “reporting and/or attacking” the account is not “understanding the
content’s context,” which it claimed was to “decrease potential harm towards children and teens
in the real world who work in the industry and to reduce their exposure online.” According to the
account, therefore its critics were “hypocritically and actually arguing in disfavor of more
extensively supporting child safety; generating results that are contrary to your extremist intent
and comments.” That is, BB Baker argued that the virtual bodies he created could reduce harm
on physical bodies of children. With a face and figure that looked younger than teenage-
presenting Wendy, these models were not dressed in intentionally revealing outfits like her.
However, all of them were girls and some were still dressed in skin-revealing outfits such as a
bikini or a midriff-exposed top and panties. More importantly, they are managed by an agency
that shares explicit pictures of a confusingly petite “19!”-year-old and hosts a curating account
that collects similar pictures of sexualized virtual humans, mostly women. In my understanding,
153
the context of this supposedly well-meaning content was what made the youth, pre-teen, and
young adult models more alarming as they were displayed adjacently with such objectified
depictions. While no physical body may have been directly hurt during the production stage, the
context makes the feasibility of the protection angle questionable even with multiple references
to the virtual bodies’ fake, technological, and graphical origin. Wendy have also been stressing
her virtual origin but had claimed her identity as a woman and spoke in a representative voice
when she wished to defend her #petitegirl depiction. On December 8, 2021, Wendy urged her
fans to vote for her to win the competition to be created into a sex doll, a literal object for
pleasure. Although she used the expression “companion doll” instead, the post started with a link
to her “more fun pics” on the adult community Slushe, nudging at the type of companion she has
been to the loyal followers and fans that cheered on her entry. Perhaps digital lives should matter
as BB Baker’s curative account has been claiming with its hashtag #digitallivesmatter, but for
whose sake? And what about non-digital lives?
I emphasize again that what is required is a sense of responsibility towards the bodies that
correspond with the virtual influencer’s. If one firmly believes that virtual influencers’ “non-
human” traits and disclaimers on them can justify problematic depictions because they are
therefore “not real,” then there is not a strong case for the creators to shy away from also
including other information for transparency, such as about who they are, since it would only
enhance the audiences’ capacity to distinguish between “real” humans and “non-real” virtual
humans. Crediting humans involved in making has also been a standard practice in various
media industries, such as in the opening credits of a movie. Of course, this does not mean that
simply having a credit roll with a creator(s) whose body matches the virtual body would be
sufficient, nor that a bodily mismatch would instantly equal wrongful depiction. Rather, it is
154
about whether the corresponding bodies’ lived experiences and wellbeing will be actively
considered in (re-)presentations of virtual bodies. For instance, a positive example of a team-
based, sex-swapped virtual influencer would be Meme (@meme.konichiwa; pronounced meh-
meh), a punk-inspired “imperfect” Japanese female virtual influencer. Meme was collaboratively
created by Atali, a Japanese CGI and virtual human company, but has been a vision project of
the male CEO Daichi Kambayashi. Since her appearance, Meme has been presenting herself in a
way that challenges gendered social norms in Japan. Kambayashi and Sonya—a member of the
company who accompanied the interview to assist with interpretation—explained that in Japan,
women were limited to “no more than three types,” for instance including the cutesy, “kawaii”
self-presentation. They explained that despite the recent attention to diversity-related topics, they
were mostly appropriated for advertisement purposes rather than aiming for social change in
favor of more diverse representation of women in Japanese media and celebrity cultures, being
“more like a pretty kind of representation.” Combined with the existing social pressure for
Japanese people to conform, which they described as “stifling,” these factors led Kambayashi
and the team to create Meme as someone who would do “all these things that normal humans
usually may be afraid” to do, such as discussing taboo topics like sexual wellness or women’s
body hair (Figure 5). They clarified their intention as “pushing the society to be okay with these
things and to be freer in their type of living” for women in Japan through a being that do not
need to and would not suffer from reactionary responses. Meme has been an important but not a
very profitable project for the company, which they explained led to her relatively slower content
update speed. They added that as a company with a marketing background they were fully aware
of the more profitable method of continuously readjusting the virtual influencer as per the target
audience’s feedback. However, they reiterated their aim with the project by stating that “Meme is
155
different because she’s kind of an artwork.” I close this subsection with Meme’s case not
necessarily because she is the gold standard. The company’s name is not easily accessible
through her Instagram page, although there are some scattered references to it in external sites
such as news articles. However, in the phenomenon where there exist many virtual bodies that
opportunistically connect and disconnect themselves from the bodies they emulate, the
company’s commitment to Meme’s contribution to Japanese society is commendable. The way
Meme’s virtual body was presented was in active consideration for the physical bodies she could
stand for, i.e., Japanese women, with a keen awareness of and matching responsibility for the
fact that her bodily presence in the virtual layer can have a social effect on the lives of human
women.
Figure 5
Meme’s Women’s Body Hair Campaign
156
Note. Source in the image. Full English caption: ““You must be like this” is changing gradually.
I used to shave as a matter of course, but why don’t you just do what you feel comfortable
with
� � � � � � � �
� � � � �. In Japan, there’s a weird pressure for girls to shave their hair
� � � . Someday someone
come to me who will love even hair.”
Race/ethnicity and Intersections
Examples discussed in this chapter mainly focused on sex/gender depiction, but virtual
influencers emulate a wider range of human identities, such as ethnicity and race, as well as their
intersections. Since virtual influencers is a global phenomenon, overall representation of
ethnicity and race are diverse. Many virtual influencers are created from the country they claim
to have come from and by the people who share the same ethnic background. Examples can
include the East Asian virtual influencer Rui from Korea (@ruuui_li), Black virtual influencer
Ivaany from French Guinea (@ivaany.h), and White virtual influencer Lisa from France
(@lisa_gen_lg). There are also many others who claim to have a mixed cultural and/or racially
ambiguous background, such as Miquela, Serah (@serahreikka), Leya (@leyalovenature), and
Liam (@Liam_nikuro), although in some cases they may visually present as someone who is
from a certain racial and ethnic background rather than mixed.
This does not necessarily mean that the virtual influencers’ identities always match with
those of humans behind them. The most widely publicized case regards Black virtual influencers
from the virtual model agency The Diigitals. A number of major news sites reported about how
their models such as Shudu and Koffi (@koffi.gram) were created by Cameron-James Wilson, a
White British man (Hsu, 2019; Jackson, 2018; Tiffany, 2019). Known to have been inspired by
the South African Barbie doll from the brand Barbie’s Dolls of the World series (Hsu, 2019),
Shudu presents as a Black woman with a lithe body and deep dark skin, often dressed in chic
outfits. The most well-known among the models at The Diigitals, she has a long record of brand
collaborations including with top global brands from various industries such as Samsung, Lexus,
157
and Salvatore Ferragamo. The Diigitals’ Black male model Koffi describes himself as a “Virtual
Influencer” and a “Fitness Model” in his Instsgram profile description (@koffi.gram), and aptly
flaunts his buff body and smooth deep dark skin in a triangular brief or in nude. The agency is
also home to more lighter-skinned Black virtual models Brenn, a curvaceous virtual woman, and
Boyce, a youthful looking virtual man who has a “digital drag queen” alter ego. The mismatch
between these virtual bodies and Wilson’s has been heavily criticized. The models have been
interpreted as reproductions of the colonialist practice of appropriating and exploiting Black
bodies by forcing them into agentless, commodified objectification, a “modern day slavery”
(Wills, 2019). In line with this reading, media scholar Francesca Sobande (2021) argued that this
is a case of racial capitalism, at the expense of absent “real” Black people. Contextualizing the
virtual influencers through spectacularizations of Black pain and lives—i.e., commercialized,
attention-mongering engagement with Black pain and lives as media content—and digital
Blackface—i.e, non-Black people’s online caricaturing of Black people, Sobande called for
attention to the historical continuance of anti-Black colonialism: “there is ample scope to
perceive such influencers as being part of the spectrum of antiBlack and colonialist ways that
Black African people and their cultures are treated as a marketable and digitally mediated
commodity” (p. 136).
In comparison with the general practice in the field, however, Wilson has since become
one of the relatively more exemplary creators in terms of transparency and accountability. He
now not only publicly indicates the models’ virtual origin and his identity, but also has
accredited his various Black collaborators, ranging from a character writer to a number of body
models and muses that he worked with. In addition, he had spoken about and released some
tutorial videos about his work process. Several creators I interviewed mentioned his positive
158
impact on the field of virtual influencers, including some Black creators who mentioned how he
had inspired them and personally helped them with their own projects.
There are still many shortcomings in Wilson’s approach, as I will detail below, and
therefore this positive reading warrants a contrast point for context. A contrasting example could
be FN Meka (@fnmeka) the self-dubbed “robot rapper,” although his animation-like 3d art style
does not fully meet this dissertation’s definition of virtual influencers. With some additional
visual indications of his supposed mechanical origin like his glowing eyes, he presents as a
heterosexual Black American man, or more precisely as a caricature of a Black male rapper (see
Balaji, 2009; Jeffries & Jeffries, 2017; Miller-Young, 2008; Oware, 2011); FN Meka is
presented with facial tattoos, various gold and studded accessories such as grills and neck chain,
and a penchant for exclusive street wear sneakers, expensive cars, and luxury brands. His speech
style also reflects these cues, both written and voiced. He also had a series of posts about his time
in jail/prison, including his post on July 16, 2019 which has a picture of him being hit by a prison
guard with a baton, accompanied with the caption “POLICE BRUTALITY” (July 16, 2019), a
phrase that cannot be taken so lightly with US history of legal racism and police violence against
Black people. Much is publicly unknown about FN Meka’s creators. He is produced by a virtual
record label called Factory New, as a representative of which only its co-founder Anthony
Martini has spoken with the media so far (Roundtree, 2021; Stassen, 2021). His original creator
is only mentioned ambiguously as a figure who had come from the video game industry with
interest in creating and selling digital goods, such as in the form of NFTs (Stassen, 2021).
Although FN Meka’s songs are created with the help of AI, he is still voiced by a ghost rapper
(Roundtree, 2021). FN Meka had spoken out against racial/ethnic discrimination by comparing it
to discrimination against “robots,” or perhaps more like the other way around, but his
159
stereotyped presentation and the lack of transparency creates an irony around the groups he may
be displacing, both in terms of representation and labor (Travers, 2020b):
“Today there is so much discrimination against robots because we are taking away jobs
from humans.”
“What makes discrimination against robots so different from discrimination against
people of different ethnicity?”
“The truth is that Humans are a flawed race that only believes in things that benefit
themselves, even if it’s at the expense of other ethnicities or forms of life. If you are
Human, imagine that your son/daughter was born a cyborg or robot, would you want your
son/daughter to have less privilege than others?”
Wilson’s direction of The Diigitals is comparatively commendable, but there are still
many remaining questions, including the yet still applicable criticism of racial capitalism. For
instance, Dulce, one of the Black woman creators interviewed, said that while she appreciates
Shudu’s beautiful design, she had more than once uncomfortably witnessed Shudu being
circulated in African social media groups as a “being represented as an African beauty” without
the groups realizing that she is virtual. In addition to her frustration at how Shudu was discussed
by Black people and African people, Dulce explained how she had been bothered by the implied
“colonial mentality.” She criticized the act of owning and controlling Black bodies as a “hollow
puppet” without actively involving (a sufficient number of) Black collaborators with lived
experiences: “It’s her existing for the sake of existing the way that she does and being passed off
as an African beauty.”
160
In consistency with my argument on physical bodies, another important question I would
like to bring attention to is how amid these attempts at bringing the corresponding physical
bodies to the forefront by hiring, accrediting, and consulting with them, the efforts have been
directed only at Black bodies. Dagny, a blond White female model, has been one of the mascot
models of The Diigitals alongside Shudu. In 2021 during when K-pop groups like BTS were
making headlines, The Diigitals introduced their new East Asian male model J-Young with his
name written in Korean as “Junyoung.” In spite of not engaging in any different activities from
other models from the agency, he has been called a “Virtual idol” (@thediigitals, April 6, 2021)
instead of the usual “virtual model” or “virtual influencer.” During its earlier days, the agency
had also introduced two new virtual models as part of their collaboration with the luxury fashion
brand Balmain: “Chinese beauty” Xhi, a fair-skinned East Asian virtual woman with
stereotypically small eyes and high cheekbones, and “French girl of his [Balmain’s creative
director] childhood dreams” Margo, a White virtual woman with dark blond hair. During the
period of their collaboration, the two models had accompanied Shudu in Balmain’s campaign
that set its goal as “increasing visible diversity in the fashion industry” (Hargrove, 2018),
although since then they disappeared. Unlike with Shudu and Koffi, none of the muses nor
collaborators for these bodies have been mentioned.
By pointing out the lack of transparency regarding non-Black virtual bodies, I am not
arguing for color-blindness that equates all racial experiences as the same; on the other hand, I
am arguing against such a color-blind, one-size-fits-all approach to diversity. Without
discounting the well-meaning intention, attempting to convey accountability towards physical
bodies only through a display of their inclusion of Black people is a tokenizing approach which
still only partially, if not passively and superficially, counteracts the negative connotations and
161
effects of bodily appropriation. That is, the structure of objectifying commodification remains
intact, ready to be re-applied to various bodies. This can still include Black bodies, especially if
they are less prominent; although there are several Black muses and collaborators listed on the
website, they are described to have been for their most popular model Shudu, with the exception
of a single male muse for Koffi. None are listed for their less popular Black models Brenn—a
curvy virtual woman—and Boyce—a young man and a drag queen. This selective transparency,
not simply of certain humans nor certain People of Color but of certain selected, more
normatively model-like Black (virtual) individuals, can be interpreted as spectacularized virtue
signaling: “brands, including creative and cultural industry organizations, (mis)use
(re)presentations of Black people as part of their digital presence to create the illusion that they
are connected to and supportive of Black communities” (Sobande, 2021, p. 139). Such a
selective signaling melds diverse experiences of marginalization and colonialistic dynamics into
a single experience, both of Black people and of people of other groups that have been
represented by The Diigitals’ models. Even if we were to focus only on marginalized identities,
not all of the categories that the virtual bodies can stand for, J-Young and Xhi’s race and
ethnicity were overlooked, as well as Dagny and Margo’s gender. The selective transparency had
the unfortunate effect of amalgamating diverse People of Color (see Grady, 2020) and other
marginalized identities, as well as distracting attention from how various identity traits intersect;
for instance, inclusion of those who inspired non-Black female virtual models would have helped
understand how gender affected the depictions of virtual influencers and how its effects may
differ when intersected with other categories such as race. Absence at the presence of partial
presence is erasure.
162
In spite of the pending issues, The Diigitals is still one of the better commercial cases in
terms of transparency and accountability. Most of the virtual influencers, not only the
hypersexualized cases that I have discussed in this chapter but many others, do not disclose the
individuals involved in the project. Of course, simply banning anonymity could be a bane for the
creators, especially those who have sought out the virtual body(ies) due to existing stigmas on
their physical body. For instance, DOB Studio’s CEO Jaewook Oh explained that the
experiences of lookism and sexism in the entertainment industry had led their virtual influencer
Rui’s body model’s wish to remain anonymous. Although not too common according to extant
literature on avatars and from my fieldwork, some may adopt a virtual body(ies) for the purpose
of transgressive identity play (see Jenson et al., 2015; Martey & Consalvo, 2011; Persaud &
Perks, 2022). It could also protect the creators from having various harassments and objectifying
solicits spilling over to their personal lives, particularly for the individuals whose body
corresponds with the virtual influencer’s; even creators like Ted who did not see any bodily
correspondence nor identification with their virtual influencer had practiced techniques for
mental well-being, which in his case led to a total abandonment of Lena. Still others may prefer
anonymity because to them their virtual influencer presents as an independent entity, not as a
doll or a product for their control. Consequently, they may wish to stay in the shadows to be
respectful of this self. Tatyana, the creator of a Black woman virtual influencer Ivaany, said
during the interview that she “really think[s] Ivaany is a virtual person,” even stating that she
feels that she has “nothing to do with that [how Ivaany presents herself].” Her comments were
not from a place of dissociation, but from care. Tatyana referred to herself as the “auntie” and the
manager of Ivaany, explaining that while she does not view Ivaany as an alter ego of herself she
has been experiencing a family-like relationship with her, mentioning that Ivaany has some
163
appearance- and behavior-wise similarities with her sister and herself. While a number of
creators had described their relationship to their virtual influencer as a family-like one by
referring to themselves as the mother or the father, Tatyana’s choice of word stood out because
she described her as an auntie instead of a parent; it was one that suggested a much less strong
sense of influence on the virtual influencer’s birth and growth. Last but not least, it is also
important to remember that the presence of a disclosure on the creator’s identity alone may not
be sufficient; Lena’s fans were actively talking to her despite frequent allusions to Ted’s
omnipresence, including some direct ones.
Ultimately, what is the most important is the sense of responsibility for the bodies in the
physical layer. This does not necessarily mean that there always needs to be a perfect match
between the identities of the creator and the virtual influencer, although lived experiences can
help with more accurate and conscious representation. Regardless of the primary bodily match, a
creator(s) can practice responsibility by learning form, including, and hiring those corresponding
bodies in the project. An even more active approach would be to commit to returning some of the
gained benefits back to the communities of those bodies, such as through the virtual influencer’s
social influence or economically. Although yet still lacking in some ways, Wilson’s attempts at
transparency and accountability could be an example of how this may be done. Meme’s aim to
push gender norms in Japan could be another example for those who are seeking more content-
wise guidance. Even though the effects may be minimal by themselves, a passive but nonetheless
complementary approach would be to include disclaimers on the virtual influencer’s origin and
disclosures on who the creator(s) are. They may help resolve information imbalance between the
creator(s) and their virtual influencer’s audiences. More importantly, it can assist with the task of
demanding active accountability from the creator(s), particularly from those who lack care
164
towards virtual and physical bodies connected with the virtual influencer’s performance of Self,
only viewing it as an object.
Conclusion
In this chapter, I discussed presentations of virtual influencers. This chapter argued that
we must consider how the virtual bodies of virtual influencers would affect the bodies in the
physical layer. The scope should include both directly and discursively connected bodies,
particularly those that can be grouped together under the human categories the virtual bodies
emulate and can represent. Much like how humans have multiple selves, a virtual influencer can
also be approached as a being that is composed of multiple self-networks. Some of the selves
that this chapter discussed include the self that presents as a meaningful social actor capable of
intimacy and the self that presents as an object or a tool for control. This multiplicity complicates
virtual influencers’ social and cultural impact. For instance, their effect on the human creator(s)
could differ depending on which self is pronounced for the creator, whether it is the self as an
instrumental vessel, an identifiable reflection of the creator, or an independent and pseudo-
autonomous being. However, what or who they are is not solely determined by the human
creator(s) but are collaboratively performed by various actors involved in the virtual influencer’s
self-presentations, including fans and followers who may interact with the virtual influencer
through an equally diverse understanding of the virtual influencer’s presentation. It is not weird
or aberrant for some of these co-creators to perceive and interact with the virtual influencer as
“real,” whether as a human or a similarly agentic being (e.g., Sylvia’s teen friend). What
characterizes the phenomenon of virtual influencers are their human-likeness and the context of
social media which brings together various human and non-human presentations that are
predicated on the assumption of authenticity. That is, meanings and relationships that are
165
negotiated during the virtual influencer’s collaborative performance of being are real. This is
why accountability should be expected from virtual influencers’ presentations, especially from
the actors that can more directly contribute to the virtual influencers’ overall performance, such
as the human creator(s) that design and manage the virtual influencers’ digital presence.
Responsibility for the bodies in the physical layer should not be limited to those that are
directly interacting with virtual influencers but also those who are indirectly connected with
them through the identities the bodies of virtual influencers (mis-)represent. I discussed various
cases of sexually and/or racially/ethnically objectified virtual influencers in this chapter. The
virtual bodies were observed to strategically connect and disconnect themselves from the various
physical bodies they can be grouped together with.
Without the intention to be prescriptive, the general rule of thumb for human creators that
this dissertation would like to suggest as a starting point for ethical practice is: for every virtual
body you create, do something for their equivalents, with an emphasis on the plural form—a
virtual influencer’s virtual body can represent a group. Disclaimers on the virtual origin and
disclosures about who the creator(s) is can help, although passively. More active ways to reach
this goal would be to consult, give space, hire, and accredit the represented group(s). However, it
is possible that this could be practiced in a spectacularized, tokenized way to gain a superficial
seal of approval, regardless of the intent. Therefore, instead of considering these suggestions as a
master checklist, I would recommend the creator(s) to approach accountability as something they
should internalize and integrate into their overall work process. Such a creator would carefully
consider the ethics and effects of the representation on their corresponding bodies, and more
ideally seek to symbolically and materially return the gains to the communities of these bodies.
A responsible creator would know and act on how unlike virtual influencers who can choose
166
(and discard) their virtual body, humans, as beings anchored to the physical layer, are born into
(identifying as) such represented bodies. While virtual influencers may be discussed as “not
real,” they still present as, and effectively are, real. Fans and audiences should also consider the
above rule of thumb and related suggestions, because they too contribute to shaping the
phenomenon through their interactions with virtual influencers.
Virtual influencers’ performances are multi-directionally co-created. This chapter
highlighted human creators and their identities to emphasize the accountability for the bodies in
the physical layer. I also discussed various ways fans and audiences were participating in virtual
influencers’ performances, at times at the extreme discomfort and frustration of the creator(s).
Virtual influencers were also approached as an actor, although the degree of their agency
differed depending on which self of them was foregrounded in the interactive context. There are
still many other actors that contribute to virtual influencers’ collaborative performances. This
includes various non-human actors such as platform politics of Instagram, technologies for
creating a virtual influencer, conventions in various interconnected industries, and the tension
between global and regional contexts. The subsequent chapter will take a closer look at the
humans and non-humans behind the scenes.
167
Chapter 3. Behind the Scenes In Real Life: Humans, Things, and Places
Chapter Summary
Virtual influencers have been narrated as beings that are leading off the novel era of the
Metaverse, a term that refers to an immersive “computer-generated universe” and a social
“imaginary place” that people can experience together through their avatar (Stephenson, 1992, p.
31), originating from Neil Stephenson’s science fiction novel Snow Crash. The term began to
enter everyday vocabulary around 2021 (Figure 6) with the growing interest in interconnected
social virtual worlds and decentralized web technologies that can support the vision, amplified
by the outbreak of COVID-19 pandemic that led to various worldwide physical restrictions. As
beings who primarily exist on virtual layers, virtual influencers were not barred by lockdowns
and social distancing. Perceived as immaterial beings that are free from the limitations of the
physical body, they began to be touted as beings that demonstrated the utopian vision of
unrestricted, socially meaningful virtual life (e.g., Teh, 2021; Whateley & Bradley, 2020).
Various outlets described their growing dominance as the dawn of a new future, declaring in its
headlines that now is “The era of the ‘Metaverse’…Virtual humans are swarming in” (Chae,
2021), and that these indistinguishably human-like virtual influencers are “looking to overtake
the real world” (S. Choi, 2021).
Figure 6
Google Search Term Trend for “Metaverse”
168
Note. As of March 31, 2022.
In this chapter, I broaden the analytic scope to human and nonhuman actors, focusing on
various material and historical structures. Despite the imaginaries around the Metaverse and its
supposed liberating immateriality (e.g., Fodor, 2021), virtual influencers come into being
through networks of bodies, things, and places, as well as the standards and conventions that
stretch across them. This chapter zooms in on the less publicly presented sides of virtual
influencers, such as hidden humans and their labor, creative tools in the work process, and the
role of the platform Instagram. It also discusses the contextual influences from localities and
adjacent industries, as well as additional intermediaries like marketers and media. Less
conspicuous they may be, they nonetheless have been contributing to the shape of the
phenomenon. The benefit of including these human and nonhuman actors in the multi-
directional, networked approach is that historical patterns and material relations that bolster the
patterns can be traced. By doing so, we can not only avoid being distracted by narratives of
novelty, but also detail inclusive ways for us to cooperate against being entrenched in negative
169
patterns. Therefore, recognizing hidden actors is a matter of both ethics and pragmatism. In
short, the core question penetrating this chapter is: what is going on behind the scenes of the
supposed epochal beginning of the Metaverse?
Networked Humans and Nonhumans
Virtual influencers’ presentations are results of collaborations between diverse actors.
Some of the actors are more conspicuously connected with virtual influencers, such as their
human creator(s) and audiences. However, there are more actors that both affect and are affected
by virtual influencers’ presentations. Chapter 2 discussed how grounding the phenomenon on the
physical layer can help trace less pronounced actors, such as human bodies that
representationally correspond with virtual influencers’ bodies. Yet, human bodies are only part
of the tale. There are also things and places in addition to humans, as well as relational patterns
that hold them together. Science and technology studies (STS), particularly those of Actor-
Network Theory (ANT) and feminist STS tradition, has extensively focused on empirically
understanding the “social” that is inclusive of nonhumans (Baron & Gomez, 2016; Casper, 1994;
D. J. Haraway, 1991; Latour, 1996; Suchman, 2009). ANT is a methodological and conceptual
approach that aims to understand phenomena by tracing relational networks between actors, led
by the actors involved. As a matter of fact, Banks’ concept of the postmodern self (2015), which
has been the inspiration for this dissertation’s multi-directional approach to Self, draws on ANT.
Agency in ANT is operationalized as “the ability to make a difference” (Sayes, 2014, p. 141)
rather than as something that is only reserved for humans, which allows a more comprehensive
and multifarious description of involved actors and their relations (Latour, 1996; Sayes, 2014).
As sociologist John Law (2009) said, it is a “material-semiotic” approach that “treat[s]
everything in the social and natural worlds as a continuously generated effort of the webs of
170
relations within which they are located” (p. 141). In other words, according to ANT, it is not
only human actors but also material and discursive nonhuman actors that can impact social
reality (Baron & Gomez, 2016; Latour, 1996; Law, 2009).
While communication frameworks such as parasocial relationship/interaction (see Dibble
et al., 2016; Giles, 2002; Horton & Wohl, 1956) and CASA (see Lee & Nass, 2003; Moon, 2000;
Nass et al., 1994, 1995; Reeves & Nass, 2002) have not completely overlooked nonhuman
actors, ANT further dismantles the ontological fixation by distancing from the view that humans
should be the criteria to judge the qualifications for being a communicator (for more on the
human-as-communicator assumption, see Guzman & Lewis, 2020). For instance, with ANT’s
assumption that any entity involved in an interaction can be an actor, parasocial
relationship/interaction does not need to force visually and performatively ambiguous entities
into first (i.e. human performing as real self), second (i.e. human performing as a character), and
third (i.e. cartoon character) “orders” of authenticity (Giles, 2002). That is, if people no longer
associate being a “real” human with being an appealing social partner, or they do but have
different interpretations of what being a “real” human entails, removing the ontological
assumption can help locate variables and associations that actually underlie social relationships.
Put in ANT terms, letting the actors trace their networks can open up new paths (Latour, 1996);
it can guide communication research to focus on interactions and their social effects without
having to pre-package ambiguous phenomena with an outmoded ontology. Although ANT has
not been free from accusations of human-centrism due to the ineradicability of human
researchers from its research process (Elder-Vass, 2015; Hornborg, 2017), its systematic
inclusivity, flexibility, and multiplicity are nonetheless valuable for navigating the boundary-
crossing phenomenon of virtual influencers.
171
Academic traditions such as posthumanism (Braidotti, 2013), vital materialism (Bennett,
2015), and agential realism (Barad, 2003) (henceforth posthumanistic approaches) have also
been pushing back on the human-nonhuman binary. Posthumanistic approaches challenge how
the binary privileges (certain groups of) humans. They advocate for an understanding of the
world and its entities as ongoing processes of co-creation. According to them, we are not fixed,
stable beings but we become, dynamically and continuously together with surrounding entities.
Often underlying in these approaches is a criticism on the normative assumptions embedded in
the humanistic universalism of the human or “the Man” and how such assumptions structurally
silence those who do not fit under their criteria—typically generalized through the image of a
rationally motivated heterosexual cis-gender White man from a privileged, Western background
(Braidotti, 2013). This is an applicable consciousness for research on mediated communication,
too. In spite of the myth of objectivity, scientific knowledge has been prone to being reproduced
with assumptions that have been privileging certain groups over others (Haraway, 1991). For
instance, biases in the universal human subject in psychological and behavioral studies have been
problematized, namely their heavy reliance on US undergraduate students who are largely
“WEIRD,” an acronym for “Western, Educated, Industrialized, Rich, and Democratic” (Henrich
et al., 2010).
Posthumanistic approaches encourage extending this critical consciousness beyond a
narrow definition of humans (henceforth, “Humans”) to diverse entities by urging a shift in how
to think (Haraway, 2016; Wolfe, 2010). For instance, Donna Haraway (2016) advocated for
thinking with multiple actors than thinking for or about them. This means beginning not with the
assumption that some entities are innately different from or lesser than Humans, but with an
awareness of shared conditions and a vigilance towards how these conditions may be differently
172
experienced by diverse actors. The practice of thinking with is also interlaced with responsibility
and hopefulness for the future, since its aim is to seek ways for a symbiotic living. In other
words, various entities are not only considered as actors, but ones which have their own desires
and stakes that H/humans should respect. Admittedly, posthumanistic approaches can be difficult
to readily implement in some communication research because they tend to have a stronger
philosophical leaning than empirical specificity (Baron & Gomez, 2016). For example, they
expand the scope of entities to all worldly entities (e.g., Haraway (2016) advocated for the term
“humusities” in place of “humanity” to stress the all-encompassing intermixture), which is not
possible to strictly follow in empirical research practice, one of the reasons being limited
resources. Nonetheless, their suggestion to differently think is generative for the task of re-
approaching the hierarchical dichotomy.
This dissertation’s multi-directional approach is also informed by cultural studies’
attention to hegemonic processes. Cultural studies has been generally anthropocentric in terms of
its research scope. However, its focus on meaning negotiation and bottom-up participation (e.g.,
de Certeau, 1984; Hall, 1993; Jenkins et al., 2013) can not only complement ANT’s open-ended,
actor-driven tracing process of human-nonhuman networks, but also assist with the task of
analyzing the traced networks. For instance, Stuart Hall’s (1993) now canonical model of
encoding/decoding explained how audiences interpret media texts: they may uncritically follow
the dominant message, but can also differently read it through an oppositional perspective or a
negotiative one. Less discussed, but he additionally commented on how there are professional
codes that can limit the ways which producers encode messages, alluding to the cultural process
of meaning making on the production end. While his model was not inclusive of nonhumans,
what Hall famously mapped was essentially a model of dynamic networked relations that
173
zoomed in on human actors with a keen interest on how power affected it. The rigid binary
between producer-encoders and audience-decoders from the original model has been since
developed, such as through Henry Jenkins and his colleagues’ (2013) concept of spreadable
media which refers to the hybrid model of media circulation that encompasses both top-down
and bottom-up creative input.
Cultural studies is particularly compatible with ANT because it can help surface and
analyze the political nuances in the networks traced with an ANT approach (see Teurlings, 2013;
for cultural studies and the larger field of technology studies, see Shaw, 2017). A common
criticism of the traditional model of ANT is that their overemphasis on multiplicity and fluidity
overshadows historicity and power (Baron & Gomez, 2016; Casper, 1994; Suchman, 2009).
Feminist STS scholar Monica J. Casper (1994) explained that while ANT has shown a deep
appreciation for nonhuman identities, they have often failed to integrate human constructs such
as genders, ethnicities, and classes (p. 848). Similarly, Lucy Suchman (2009) criticized ANT’s
lack of interest in the politics of difference, pointing out that the performance of humanness
inevitably involves marks of gender, class, and nationality—a particularly apt point for virtual
influencers since they emulate humans. These criticisms resonate with cultural studies’ interest
in the everyday. As Ben Highmore (2002) said, “everyday life can both hide and make vivid
social differences” (p. 2). In other words, there are social differences embedded in and
reproduced through seemingly invisible, mundane practices, including those around the notion of
human. Highmore’s edited volume additionally points to a generative potential in having such a
political consciousness: it can help us re-imagine.
To sum up, in addition to having a methodological compatibility with ANT, underlying
interest in equity in posthumanist approaches and cultural studies can enrich ANT’s
174
descriptiveness with analytical depth. Posthumanist approaches can theoretically reinforce ANT
through their arguments on process-oriented, inclusive, and co-creative understanding that calls
for a shared sense of accountability for all “becomings’” wellbeing. Likewise, cultural studies
has long investigated cultural makings of more or less normatively “Human” groups, as well as
the ways which various marginalized groups have been negotiating with and subverting such
structures. Although more anthropocentric in its traditional form, cultural studies’ interest in
power relations enables the analysis to reach the questions of how and why.
A key concept that underlies my approach is Donna Haraway’s (1991) influential
metaphor of the cyborg, which she proposed as a political ontology that can promote responsible
social relations of science and technology by subverting universalizing dualisms. Defining the
cyborg as “a cybernetic organism, a hybrid of machine and organism, a creature of social reality
as well as a creature of fiction,” (p. 149) she explained how the idea of being is a historical
invention that is sustained via power relations, reproduced through material and imaginary
performances: in her words, “we are cyborgs” (p. 150). Haraway argued that there are three
boundary transgressions that created cyborgs: 1) between animal and human, 2) between
organism and machine, and 3) between the material and ethereal. Cyborgs are born due to such
totalizing dichotomies; they come from the outskirts, ambiguities, and conflicts. However, at the
same time, their presence as abominations challenges the boundaries in return. As neither strictly
human nor technology in the conventional sense, virtual influencers are cyborgs that challenge
the dichotomous boundary.
The cyborg metaphor is not only applicable to this dissertation’s case, virtual influencers,
but also multi-directional approach. Gary Lee Downey and colleagues’ (1995) “cyborg
anthropology” shares a similar theoretical underpinning, having been inspired by British cultural
175
studies, STS, and feminist studies. They were interested in breaking down how human
distinctiveness is culturally produced and managed through boundaries (pp. 264 - 265) by
studying the implications of boundary-crossing cyborgs, similarly to how I got drawn to the case.
There is also an overlap in the research angle in that we both contend that the scope of actors and
agency should not be limited to humans. Yet, there are two key differences: the starting point,
and the final point. Their examples largely began with binaries, such as the human versus
machine angle, although they quickly dismantled them by illustrating how cyborg anthropology
can be practiced. My case begins with cyborgs: virtual influencers. The second difference is that
the goal of this dissertation is not only to examine the cultural production of the human-centric
ontology, but to participate in its process as an intervening actor. While at the core they advocate
for incorporating diverse actors’ participation, the authors describe popular participation in
institutionalized science and technology as potentially dangerous. I see this as a responsibility
and chance. The virtual influencer phenomenon is a collaborative one; diverse actors, including
those without professional expertise, are already involved. Therefore, initiatives for
accountability can be best implemented with cooperation from these actors, both in terms of
collectively imagining the future and being committed to practicing them.
With this approach, the following section will detailly trace virtual influencers’ networks
to locate continuing patterns and to inclusively re-imagine them. The viewpoints and activities of
many less conspicuous actors participating in the phenomenon will be analyzed, regardless of
their Human status. Hidden humans, things, and places matter, too: we are not beings, but
becomings.
Behind the Scenes
176
Virtual influencers are co-created by diverse human and nonhuman actors. The two
groups of actors that had been spotlighted in Chapter 2 were human creators and audiences. In
Chapter 2, the term creator was used to broadly refer to the main artist, character writer, and/or
social media manager of a virtual influencer. In some cases, this may be a single individual.
However, in many others, a group of individuals come together for a virtual influencer project,
sometimes only partially knowing or without knowing that they will be contributing to one. Such
a group may have been formally organized in form of a company or a more loosely connected
network, the latter of which can further obfuscate the network. Still, even when an account is run
by a single individual, their decisions can be guided by various material and semiotic nonhuman
actors, such as standard technological tools and discourses on the internal logic of social media
platforms. There are also actors that may seem less directly involved in the creative pipeline but
are nonetheless influential, such as localities of a region, intersecting industries, and
intermediaries like marketers and media that narrate the phenomenon.
The presence of these less conspicuous actors is not unique to the phenomenon of virtual
influencers. Matt Stahl (2011) defined virtual labor as performative labor by a single person,
whether human or CGI-made synthespian, that is created from divided labor by multiple entities
such as creative and technical workers, legal artifacts, and technologies. Stahl drew on American
virtual bands from the 1960’s and 1970’s (for more examples, see Chapter 1) to illustrate the
prehistory of alienated division of labor in contemporary visual effects-enhanced acting
performances (see also Chung, 2015). To Stahl, the historical continuity was found in how those
in supervisory roles organized the ensemble of actors to minimize employee agency for the
illusion of character agency, on behalf of capital (p. 11). This section, however, will also discuss
the ways which these seemingly controlled and minimized actors exert agentic influence on the
177
performance of virtual influencers, both in the ANT sense of agency—i.e., “the ability to make a
difference” (Sayes, 2014, p. 141)—and in terms of guiding supervisory decisions with a willful
force in subtle or incomprehensible ways (see Whitson, 2018). This is not in disregard of
alienating effects of virtual labor, surfacing which is actually one of the goals of this section.
Those in a more prominent or directly connected position can have a stronger, more easily
translated influence over virtual influencers’ presentations (see Chapter 2). Rather, it is to trace
the desires and actions from the bottom up to arrive at conceptualizations and suggestions that
are based on a more candid, comprehensive understanding of the phenomenon.
This section discusses various illustrative cases that demonstrate the multi-directionality
of the phenomenon’s networks in order to analyze their common patterns. However, it is not
exhaustive of all actors and relations that are involved in the phenomenon. Still, the patterns
show that in spite of the utopian imagery of limitless immateriality, virtual influencers, as well as
related technologies that have supposedly led off the novel phase of the Metaverse, are anchored
onto material and semiotic structures that stretch from the physical layer.
Work Process and Standards
A typical work flow for a virtual influencer’s Instagram post generally involves the
following, in unfixed order: designing the virtual body, scripting the character profile and their
narrative arc, shooting and/or graphically rendering the visual content, crafting the
accompanying verbal message, and posting and interacting with audiences. Depending on the
type of the project and participating actors, some steps can be added, skipped, or modified.
Hidden Humans. One of the biggest differences in the workflow comes from whether a
virtual influencer is created in full 3D or partial 3D, such as by photoshopping a 3D face onto a
178
picture of a physical body. Although virtual influencers have been characterized by the
immateriality of their virtual bodies, many require support from physically existing humans aside
from the main creator. However, these human actors’ labor tends to be hidden or alienated in the
virtual influencer’s Instagram presentation.
One of such humans are body models of virtual influencers. Due to reasons such as
quality, production speed, and cost of production, some virtual influencers are not made
completely in 3D. Virtual influencers like Imma Gram and Liam were reported to have been
created by transposing a 3D CGI head or face onto a live-shot human model’s body in a physical
place (Cade, 2020; Tiffany, 2019). Many other virtual influencers tend not to publicly reveal
their creative pipeline, but CGI experts have commented on how virtual influencers like Miquela
are highly likely to have been created through a similar process because it would be impractical
to repeat time-intensive, difficult tasks like replicating draping patterns of an outfit, sculpting
hands, and designing CGI hair for a single post (Crouch, 2019; interviews). There are also cases
where the head of a virtual influencer is edited onto pre-existing photos possibly without the
original model’s awareness, such as in the case of Lisa Gen’s post on November 27, 2020
(@lisa_gen_lg) that altered a picture from the top Korean actor Jun Ji-hyun’s photoshoot for the
luxury fashion brand Alexander McQueen. (Figure 7)
Figure 7
Virtual Influencer and Original Photo
179
Note. Lisa Gen’s post on November 27, 2020 (left) and Jun Ji-hyun’s photoshoot for Alexander
McQueen (Hwang, 2020) (right)
During the interviews, many creators have expressed that body models are an important
part of the creative pipeline, not simply in terms of work efficiency but for inspiring the
presentational direction of their virtual influencer. Jawad El Houssine, the original creator and
now a member of the production team that creates virtual influencer Serah Reikka
(@serahreikka), said that Serah’s current “half Asian” and “half Arabic and Russian” appearance
came from the looks of three human models who were employed to be facially motion captured,
respectively from China, Lebanon, and Russia. He said that as a result, Serah’s current look is
“half human half computer.” Even after acquiring this initial design, Jawad’s team has been
continuing to hire models to pose as Serah in order to meet investors’ preference and demand for
realism. It is not exceptional for a company to hire a number of different body models. Xander
Smith, the character designer who oversaw virtual influencers Binxie and Aliza’s designs, also
commented that they had hired “a lot of different models that would play” them because each
model had different strengths, such as one being more skilled in video shots and another in still
180
images. While Xander explained that their virtual influencers’ designs were more analytically
oriented in that they referenced the looks of models with high engagement rate from their target
demographic, he alluded to some of the body models’ influence on the design by mentioning that
they had re-invited some super models after completing Binxie and Aliza’s designs.
Body models’ lasting influence was not limited to appearances. If the body models for
Serah, Binxie, and Aliza were described as important for their visual depiction, there were others
who were linked to virtual influencers’ essence. Daichi Kambayashi and Sonya from Atali, the
company behind Meme, said that they have about six to seven people who are body modeling for
her. Adding that Sonya had also modeled for Meme one time, they explained the relative
flexibility they have with choosing body models. To maintain the consistency of Meme’s virtual
body that was based on the original model, they used editing tricks like leaving the feet out of the
frame of the picture if a model is shorter or taller than Meme’s profile height of 168 cm (around
5’5’’). However, they said that the “impression,” “atmosphere,” or “aura” of Meme changed
with body models, explaining that it could be from slight changes in their neck and shoulder
lines or nuances in facial expressions:
“Strictly speaking, the impression of a human changes depending on the feeling of their
shoulder or neck. Even if their face is the same, the impression definitely changes
depending on the lines in the [shoulder-neck] area. So we try to best match the area or
edit it slightly. But from the perspective of people who have been continuously looking at
Meme, the atmosphere or aura of the original model, or maybe the subtle nuances in
facial expressions, does not feel like it’s not there anymore. But maybe normal people
who think that the change isn’t much might not even notice.” (Daichi Kambayashi)
181
The “atmosphere” of a body model can be an important ingredient even for pre-designed,
fully 3D virtual influencers. Reggie Goff, an independent 3D artist behind Ghanaian virtual
influencer Aba Wils, spoke of a time when one of his tenants inspired a post for Aba. On the day,
he encountered them hand-washing their laundry, which he described as a mundane weekend
activity for Ghanaians. Reggie has always been interested in showing Aba engaging in everyday
life activities in Ghana. He said that therefore the scene struck him with great inspiration, leading
him to rush back home to get his filming equipment to take pictures for reference photos. He
added that although they initially did not understand why he would take their pictures when he
asked for their consent, they found it impressive once he showed the rendered results, after
which he updated Aba’s Instagram with the picture (posted on October 24 and 25, 2020).
Perhaps without Reggie’s interest in portraying everyday life in Ghana, this scene would have
gone unnoticed for being perhaps too mundane. Thus, arguably, Reggie’s role as the main
creator behind Aba was the most critical in arriving at this particular presentation of Aba. Still,
the core activity for the presentation was taken from the tenant who inspired Reggie, as well as
the reference photos that captured their corporeal activity. Despite not knowing much about the
production process, the tenant had been one of the key co-creators of the post.
Individuals who contribute to the creative pipeline as the sole body model can have an
even stronger influence on the virtual influencer’s presentation, both appearance-wise and
beyond. Jawad had mentioned that he formerly worked on Cade Harper, a teenage boy virtual
influencer who is no longer active. He did not share the details because of his NDA, but briefly
mentioned that Cade had a single body model who he had worked with through a similar work
process with that for Serah and her three body models. Seihee Kim, the production director at
Klleon who is overseeing their virtual influencer Wooju, explained that they have been using a
182
colleague at their technical research department as Wooju’s body model. Her team had specified
Wooju’s personality and appearance before involving the employee, but she said that they were
able to naturally think of him during their production meetings because he was just like Wooju:
“the employee popped up in our thoughts. Wooju is just him. We should use him. He has
a similar height and there’s a lot of similarities between our concept for the virtual
persona and the real person. So let’s just use him.”
However, in spite of the overlap, Kim added that her team was interested in looking for a
professional model who can pose for Wooju because they were encountering some limitations in
the employee’s capacity as a non-professional, such as relatively lacking creativity and
naturalness of the outcomes. Unlike Meme’s team, Kim’s team was not too focused on the
original body model because they believed that Wooju remains the same regardless of who he is
projected onto, even when there are slight changes in his looks as the result of the switch. At the
same time, their comparative dissatisfaction and desire to switch models illustrated the
importance of the body model in the overall presentation of the virtual influencer, not simply
towards their Instagram audiences but towards the members of the creative team. Kim had
mentioned Rui, a deepfake virtual influencer who is managed by DOB Studio, as an exemplar
case of a naturally presenting virtual influencer who led them to recognize the importance of the
body model as a fellow creator.
Indeed, in the case of Rui, her sole human model has been critical to the details of her
presentation. CEO Jaewook Oh described Rui as a sub-character (“boo-kae”) or an alter ego for
her body model and the behind-the-scenes talent. Although it was DOB studio that had searched
for a body model with singing talents with the vision of presenting Rui as a cover artist, Oh
183
explained that it was the hired model’s life values that inspired and determined the details of
Rui’s sweet personality:
“We wanted to observe her to show her [the body model’s] hidden appeals. So the first
question I asked during my interview with her went like this. If you become a star, let’s
say you became a popular influencer. People will be wowed by whatever you say and
follow it. If so, what do you want to do? What is your biggest area of interest? She
responded by saying that she wanted to help underprivileged children. Yes, she wanted to
help children, especially in the education area, because she believes that there are too
many vulnerable children. This struck a chord. If someone like this becomes an
influencer, then the world would really become a brighter place. So we designed her as a
very, very sweet and saintly character.”
While Rui is not a digital twin and was repeatedly clarified as a separate character from her body
model, it was their understanding that Rui would not have been the same virtual “person”
without her body model. It was not just the company’s predetermined vision nor the looks of the
swapped face that crucially defined Rui, but the body model’s disposition that was shared with
her.
Some of the body models’ contribution to the virtual influencer was formally
acknowledged. For instance, The Diigitals’ official website has a “Muses” section, such as for
Alexsandrah Gondora who was the first body model hired for a photoshoot gig for Shudu (who
had been created in 3D). Gonodora’s page has a video interview and a text summary of it, in
addition to her pictures and social media link (The Diigitals, n.d.). In the video, Gondora talked
not only about her experience of standing in as Shudu but also about her experiences as a Black
woman fashion model in an industry that has traditionally favored White women. Making
184
comparisons, Gondora spoke extremely highly of Cameron and his efforts to hire Black people
for his work on Shudu, including staff, not just models. Although not every post on Shudu’s
Instagram does, her post on August 31, 2018 credits Gondora (“Amazing thanks to
@iamalexandrah for help creating this shot.”) and styling collaborators
9
. In other words,
Gondora’s contribution to Shudu’s presentation has been acknowledged by giving her public
accreditation on Shudu’s official platforms and by dedicating a place within one of them for her
to speak with her words. Rui’s body model on the other hand has not been publicly accredited
due to the mutual agreement between her and DOB studio to keep her identity anonymous, but
her contribution has been formally recognized via organizational means. Despite beginning as a
“collaborating artist,” she is now hired as a full-time employee who is recognized as “one of the
founding members of the company.” Oh explained that although their goal for Rui has been
already met, which was to showcase the company’s technological services to investors and for
brand recognition, there has been active discussions in the company about who Rui can now
become and how to assist with the body model’s success in her own musical career. He spoke
extremely fondly of the body model’s passion in music and academic prowess, describing the
company’s plans to branch out to the music industry with her playing a central role as part of the
management. As Oh said, they have been “thinking of her as a core member who will continue to
work together with the company to grow together with it and to actualize her dreams.”
This, however, is not always the case; many body models are not just less noticeable in
the front-end results but alienated as a cog in the wheel in the virtual labor. Even Rui’s body
model began as a “collaborating artist” rather than a full employee despite the influence that she
9
All cited social media posts in this chapter have been re-accessed on March 22, 2022 – April 1, 2022 to cross-
check archived material unless otherwise indicated.
185
was expected to have on Rui. Moreover, with the majority of virtual influencers presenting
themselves as a being with full autonomy, the internal workings are made even further invisible;
what tends to be available on their social media is only the virtual labor (see Star & Strauss,
1999). As Stahl (2011) argued, this is not independent from conventions and technologies. Jawad
recounted his experience of meeting body models in Shenzhen, China. His impression was that
“they don’t care” much about exactly what their modeling job was for because they approached
the task as a mundane part of their job: “this is a job so they know that their body can be used for
anything, for clothings, for accounts, for any kind of model shootings, movies, representative.”
This approach, however, may rather be a learned one. The precarities of the occupation of
modeling has been linked with their self-commodification and emotional labor, including those
to appease their clients (Mears & Finlay, 2005; Wissinger, 2009). In this sense, Jawad’s
observation is not incorrect in terms of the norm, but despite often being treated as “just paper
dolls” (p. 339)—i.e., objects used to show and sell products, models understand that their work
additionally requires interpersonal labor (Mears & Finlay, 2005). Therefore, it is possible that
body models were “working” to appear as not caring, rather than being truly unconcerned about
how their body will be used.
The results of the models’ bodily work are subsequently processed and reformatted in
ways that makes the virtual influencers’ connection to the original bodies further opaque,
whether as materials ready to be transposed with virtual body parts or as source data for full 3D
virtual body designs. This is similar to Chapter 1’s discussion on how the sound library for
vocaloid software like Hatsune Miku was built. In order to create the sound bites that form the
technological voice of Miku, voice actor Saki Fujita’s raw human voice was harvested as the
base material, which then was sliced and processed into abstraction. As media scholar Daniel
186
Black (2012) said, “the human body remains in the chain of commodification” (p. 233). Indeed,
as Xander described during the interview, the creative pipeline for commercially-oriented virtual
influencers can be more “analytic” in that they involve combining sliced, processed references
from the body models into a design that has been determined as the most strategic from their
consumer research.
In such a chain, however, it is not only the body models whose labor remains in a
fragmented form, although for them it is their literal body (parts) that stays in the chain.
Technicians and artists that provide their specialized labor—including bodily labor—are also
hidden but necessary contributors. For instance, Nina Hawkins said during the interview that she
and her colleagues “were running all over” the city to shoot base pictures for a virtual influencer
project that she had worked on. Nina explained how she had significantly contributed to the
results, although not without great specification due to her NDA. Nina was an independent
contractor and the virtual influencers’ accounts do not accredit Nina. A multi-talented individual
she is, Nina has been doing well, enjoying the growth of her own digital twin business. Other
hidden bodies may be doing well, too, using each gig as a chance to find their place in the
industry and network. If so, however, why not let them be known?
Despite their influence on virtual influencers, hidden humans were treated as tools in the
production pipeline. As Alexandra Chasin (1995) argued, the logic that alienates servicing
human class is similar to that for treating technologies as servants: “perhaps the crucial
invention, the one that makes “all we have to do go by easier” is not the tool-object, but is
precisely the invention of difference between human subjects and other-than-human objects” (p.
94). Contrary to the utopian dreams of the “freedom of form” or “power over representation” that
187
virtual technologies have been expected to herald (see Fodor, 2021), these examples show their
strong footing in the physical layer and patterns of practices that stretch from it.
A material actor that could exacerbate the alienating process of labor division is the
common industry practice of requiring an NDA(s). Some of my interviewees, such as Nina, were
bound to an NDA(s) that prevented them from discussing in detail their (lasting) contributions to
virtual influencer projects. NDAs can help protect information assets of a company, but they can
also be (ab)used as a silencing tool that enables exploitation (Arditi, 2020; Arnow-Richman et
al., 2022). David Arditi’s (2020) research on the political economy of NDAs for the singing
audition program The Voice found that they were used to create the show’s spectacle while
hiding precarious labor from their contestants, such as uncompensated demand for work
flexibility and informal labor for promoting the show. Arditi expressed concern at the fact that
most contestants were not fully aware of what signing the NDA in earlier stages entailed. These
are applicable criticisms for virtual influencer production. Tom Pastor, a creator with decades of
experience in commercial 3D design, mentioned in the interview that excessively complex and
long NDAs were one of the reasons that he preferred to work independently on his virtual
influencer Lisa Gen. In addition to the time and effort required to comprehend them, he was
wary of how some companies have used NDAs to force artists to hand over a portion or the
whole of their work, especially with the help of their comparatively extensive legal resources:
“The reason why [I did not take the offer] is because once you sign those things basically
you are enslaved. Everything that you produce. They own Lisa, they own everything.
And I noticed that with some people working in 3D for fashion industry, you know
cosmetics those kind of things, at the end of the project, the company would ask okay you
188
signed NDA now you give us all the 3D files. No discussions. So it’s like…10 years
creating digital humans and suddenly you lose her because you sign a paper. No.”
“That’s the reason why they have a team doing those contracts. Basically they are doing
this all day long. But I cannot do this all day. I’m not getting paid for that you know. It
takes sometimes a week more than just to talk about it, you know to change stuff and they
have tricks to take over for you all the time because it’s their job….So yeah it’s time
consuming and is too dangerous.”
Therefore, Tom explained that he had to depend on himself to create and promote Lisa. Drawing
on his experience and knowledge, Tom tried his best to subvert the dominant structure by
distancing himself from NDAs. This was particularly important for him because he wanted to
leave behind Lisa as an asset for his children if she becomes successful. Yet, with NDAs as the
standard practice for company-involved collaborations, this meant that he was still influenced by
them. Tom had to independently resource Lisa. The night before our interview, he had used his
devices for 28 hours to render a full CGI picture: “all of my machines are overconsuming
electricity….the machines are heating the flat.”
As Told by Tools? Also approached as tools in the creative process—perhaps more
intuitively so than hidden humans, technological actors tend to be (made) hidden as well but are
nonetheless influential to the shape of the phenomenon. A part of the utopian promise of virtual
influencers is drawn from their trait of being technological—perceived as those at the servitude
of supervising humans, capable of being extensively controlled for gain without having to
involve actual humans (e.g., Florian, 2021; Leighton, 2019; Teh, 2021). However, not only do
virtual influencers still require human bodies, interconnected technological actors can push
humans’ decisions on virtual bodies in subtle and explicit ways.
189
As illustrated by NDAs (which is a legal technology), technologies are not objective nor
void of intentions. By stating that the nonhumans are “the missing masses” in sociological
research, Bruno Latour (1994), one of the key figures of ANT, proclaimed that sociotechnical
research has undermined nonhumans’ capacity for action. In the earlier phases, however, some of
these efforts were focused on studying how technologies stood in for humans as specialized
delegates of the designer’s intentions, such as how a weighted hotel key may discourage the
guest from losing or taking it (Akrich & Latour, 1994; Latour, 1994). These earlier positions
have received criticisms on how it was still human-centric about nonhumans’ intentions and how
it missed people who had been traditionally ostracized (Baron & Gomez, 2016; Müller & Schurr,
2016). This development surfaces a conundrum in discussing nonhumans: how do we embrace
nonhuman agency in an anti-human-centric way without marginalizing humans as a result?
Thus, I must first clarify my position before further proceeding into technological actors.
While I am for seriously considering nonhuman agency, my interest in them is grounded on my
interest in humans (see Casper, 1994; Chasin, 1995; Suchman, 2009). I would be the closest to a
posthumanist humanist (Wolfe, 2010, pp. 123-126), a person who externally relates through a
posthumanist view but is a humanist at the core, at least in terms of my primary interest in
human affairs. Or, paradoxically, I think being so may be inevitable for all posthumanists if we
seriously consider reflexivity and our situatedness; we may strive to “become with” various
entities (Haraway, 2016), but we cannot deny the history of how we have been relating to the
world as humans. In the simplest sense, we are limited to discussing nonhuman entities through
the language of humans from the position of humans (c.f. “situated knowledge”; Haraway,
1991). My perspective begins with this humbleness. This dissertation recognizes that nonhuman
actors can exert significant influences but does not engage in the question of how these actors
190
may “think”; it focuses on relations and consequences. This information is approached as what
provides more accurate and deeper insights into the politics of human exclusions. This is not so
that only humans can thrive. I am merely proposing to become with different humans first,
during the course of which I believe the limited definition of human—i.e., Humans—will further
open up its privileged space to various other-than-Humans without still enforcing the structure of
exclusion among humans in the name of respecting nonhumans. To paraphrase, arriving at a
projected understanding of how nonhumans may experience the world is not the goal of this
chapter especially if it should be at the relative lack of attention on still marginalized humans in
the network of relations. Instead, what this dissertation is interested in about technological actors
is how they are entangled in the social and what effects they have on humans who are entangled
together with them.
Technological actors in the phenomenon exerted various critical influences on the
networks of actors, but not as delegates of human intentions. For instance, some interviewees
spoke about the role of devices and material infrastructures. Tom described powerful but
affordable personal computers as important contributors to the current flourishing of virtual
influencers, mentioning that when he first began taking on virtual human projects in the early
2000s, a professional computer that is specialized for 3D graphics, such as a Silicon Graphics
computer, was necessary. He also compared the time it took to render a full 3D CGI picture of
Lisa in the early 2010s to now, the difference between which was nearly five days:
“It was really painful because the pipeline to process—of producing all the pictures for
Lisa took one week rendering. Basically so the machine wasn’t good. Rendering every
twenty four hours a day for one week to get one picture out. All the hairs and everything
191
for print resolution. Today it’s—today I can do it in one two days max. You know, so
between one week and two days it changes everything.”
As the quote shows, while approached as tools, computers were described as something that Tom
did not have control over; they took their own sense of time, and Tom had to follow their rhythm
to reach rendered Lisa. This is not simply an era-based relationship, but of accessibility. A 3D
specialist with decades of experience and a long-time resident of the virtual world Second Life,
Dulce had the skills and interest to create a high-end virtual body for her virtual influencer Esme.
However, Dulce had to lay aside Esme due to the limited capacities of her computer until she
was awarded with a new, high-functioning device from a job client that was highly satisfied with
her work:
“I had originally made her in 2020 because I kind of got frustrated because I saw these
virtual influencers. Characters start popping up and they’re getting all these Instagram
follows and all these write-ups and things. And I can do this stuff. I want to do this. So I
made her but the computer I made her on was not capable of any kind of rendering or any
kind of doing anything. So I kind of shelved her but then I did a job for the Marina Bay
Sands [architecture landmark] in Singapore. The PM [project manager] on the job, he
bought me a computer. He bought me this really nice Alienware [brand]. Yeah, so I was
like yeah, I’m in business. So I dusted Esme off and started just rendering her you know,
and making these high-end renders of her.”
Dulce’s computer had barred Dulce from vitalizing Esme but allowed her to satisfy her client in
an architecture digital rendering job, which consequently led to a more powerful device that
could stably create high-end versions of Esme. Computer devices, despite their relative lack of
visibility in popular narratives on virtual influencers (Chapter 1) and their social media
192
presentations (Chapter 2), certainly had “the ability to make a difference”—i.e., agency (Sayes,
2014, p. 141)—in how Lisa and Esme presented themselves, as well as how Tom and Dulce
related to the phenomenon.
Two infrastructural material actors that are closely associated with devices are electricity
and the Internet. As mentioned earlier, Tom had jokingly complained that due to the amount of
hours his renderings take his devices have been overusing electricity to the point that they are
heating the apartment, just in time for winter to arrive in France. Otherwise, neither electricity
nor Internet was brought up in regard to virtual influencer creation, with the exception of one
interviewee. Jessica James was not too different with other interviewees in terms of her hopes
about the Metaverse, but explained that she wished to achieve her visions in a country like the
UK due to some of the infrastructural difficulties she have been experiencing in her country,
Nigeria. Some were more cultural, such as the relative nonrecognition of the emerging fields of
virtual influencers and digital fashion and the resulting lack of local support, but there were also
material infrastructures that she had to tactically navigate in order to stay competitive. She was
frustrated that electricity was not constant in her region, which is something that is highly
incompatible with the time-intensive task of 3D rendering. She subsequently spoke about having
to stay online to be in her digital area of work and some of the professional software she uses to
belong in the space were additional hardships for her because of the relative weakness of her
country’s currency:
“That’s my frustration with the physical location as well. It’s, you know, the country I am
in is—I don’t know, is really not conducive for the kind of work I do because there isn’t
like 24 hour electricity. So most of the time, you know, when you are rendering out, you
know, stuff that takes hours and hours like eleven hours to [render] in your system, you
193
always have to count on electricity. And you know, you know, no notice, no nothing, is
gone, and what do you do? Just stop your stuff. And at the same time, Internet service.
Because, you know, you pay for Internet to, you know, be online. Because this place is
generally online, you can be in a space of being online, so you need as well, the same
thing because of all these software I use they’re kind of expensive, and you know my
currency isn’t.”
Despite the myth of liberating immateriality around virtual technologies, Jessica was held
back by various material actors in her physical location that were incongruent with industry
standards. As Susan Leigh Star and James Griesemer (1989) argued, infrastructures become
visible upon breakdown—or in Jessica’s case, during shutdowns. Electricity and Internet
connection are infrastructural actors of the phenomenon; powerful and steady electronic
interconnectedness is embedded into the phenomenon, linked with standards and conventions of
professional practices. In addition to expectations regarding online presence and software
choices, there were also those on the speed and quality of work. A passionate and driven person
she is, Jessica reliably met her global clients’ needs by planning meticulously and working ahead
of schedule so that her renders would not be interrupted by infrastructural actors. Although her
relationship with these infrastructural actors were different from those who may belong in
“standard” regions, such as UK that she wanted to eventually live in, growing up in Nigeria had
given her intimate knowledge about Nigerian electricity that enabled her to negotiate with it and
make do as needed:
“I’ve been here since birth, so I found a way to work around it. Especially when it comes
to client work. So it’s always—you know, I always start before I’m supposed to and just
make sure everything is in order so I don’t disappoint my client. Because at the end of the
194
day, your service, your render is what will bring them back. So I always make sure like
my services you know, top notch, they have nothing to complain about, or you know stuff
like that. So yeah, I work around it.”
Immateriality of virtual layers allowed her to earn economic and social capital to chase after her
dream but their inseparable connection to infrastructural actors in the physical layer critically
influenced the way Jessica navigated the phenomenon. This was by no means unidirectional;
Jessica’s lived experience helped them find a cooperative rhythm.
Silent Standards. Although Jessica was able to work around her frustration, experiences
of marginality should by no means be dismissed, especially not on the basis of individual
capacity to negotiate with it. Inspired by Haraway’s (1991) ontology of the Cyborg, Susan Leigh
Star (1991) masterfully weaved her onion allergy with her argument on heterogeneity and
multivocality. The marginality and seeming invisibility of non-standard conditions (e.g., onion
allergy) against standardized operation (e.g., standardized menu and pipeline at a busy
McDonalds as well as the presence of non-allergic colleagues) result in disbelief from others and
self-surveillance. This is despite the mundane but continuous—or potentially serious—daily
discomfort. Star suggested that what results from such intersections between standardizations
and local experiences is multiple membership; standardizing technologies are not totalizing. The
notion of multiple membership can help approach standards through a lens of equity. As Star
mused: “How many people can get in and out of doors, and how many cannot?” (p. 43)
Some of the standards regarding virtual influencers’ virtual bodies can be traced to
software programs. There are many software that artists can use, such as Daz 3D, Maya, ZBrush,
Cinema 4D, Reallusion, and MetaHuman Creator, as well as in-house technologies like deepfake
software from DOB Studio and Klleon. A number of interviewees mentioned that they have used
195
more than one software and at times also used some photo editing software to additionally
enhance their results. Still, the general consensus was that one of the most popular software
among them is the 3D human modeling freemium software Daz 3D, known for its relative ease
to use. In fact, some interviewees said that they tend to exclusively use Daz 3D, regardless of
their skills in other software or their field expertise. This was to the disdain of others: its ease is
at the expense of professionalism and flexibility. Perhaps true to the evaluation, on January 20,
2021, I was able to render my first original 3D virtual human with Daz 3D within a day (Figure
8). Although I was pleased and surprised at the results of my first ever trial at 3D human
rendering, I was not entirely happy about the design; as some of the artists criticized, I did not
feel like the choices were entirely mine.
Figure 8
My First Original Virtual Human Created with Daz 3D
196
What made Daz 3D easy was what made it limiting. A first-time user is met with a
“Getting Started” pop up that guides them through its interactive tutorial. I encountered some
difficulties due to jargon and inexperience, but was able to follow its step-by-step instructions
and try out various features of the program. My impression was that with some patience and
practice, even a complete beginner would be able to soon start creating virtual humans with Daz
3D, although it is possible that my past exposure to some photo or video editing programs may
have helped with my learning curve. The outcome of the tutorial was a blond, blue-eyed White
female “barefoot dancer” created from their basic Genesis 8 Female model, dressed in a bardot
top and a matching skirt (Figure 9).
Figure 9
Image Created by Following Daz 3D Tutorial
I wanted to try creating my own version of a virtual influencer but ended up creating a
Daz 3D-guided version of it. After finishing the tutorial, I opened a new project. Since I was still
new to the program, I stuck to the basic Genesis 8 Female model so that I could reference the
tutorial if I encountered any difficulties. I also noticed that most user-created basic tutorials on
197
blogs and YouTube were based on the Genesis 8 Female model, which also influenced my
decision. Sometime after fiddling with Daz 3D, I began to feel frustrated; as an avid player of
The Sims 4, a video game that provides many diverse preset features as well as touch-to-shape
detail sliders, I felt limited with the range of basic face and body sliders. At first I thought it was
my inexperience that barred me from finding more powerful features. Upon watching a few
instruction videos, however, I learned that in Daz 3D, I had to purchase sliders from its package
store if I wanted more detail and range. Discouraged, I spent the next few hours looking up and
trying other 3D software, including those that allow its users to create a virtual human from
scratch from wire modeling. Unfortunately, they were either paid programs or were too difficult
for an inexperienced user like me to use. So I returned to Daz 3D. After an extensive exploration
of their package store and reviews on some of the more popular packages, I ended up deciding
against buying any. I wanted to familiarize myself with the program more before I bought
packages, and I could not afford all the packages that I wanted—an earring package may only be
around $10-20, but all you get from the package is a specific type(s) of earrings.
Not being able to create nor purchase item packages meant having to design more closely
to the default body. Tired of practicing with the blond, blue-eyed White virtual human, I futilely
tried to mold the Genesis 8 Female model into an East Asian virtual woman (Image 3). I was not
able to figure out how to shape the model’s eyes into East Asian eyes, neither monolid nor folded
double eyelid, nor how to change their colors into a high-quality mixture of black and brown. I
studied several instructional videos on eye construction and browsed some more packages,
including for other bodily features and fashion, but gave up after deciding that it was too much of
a hassle. I knew what an East Asian woman looked like and was interested in creating one in 3D,
but it was simply the easiest and the cheapest to stick to the default body.
198
This experience of silent standards was not unique to me. Tatyana, a Black woman artist
behind Black woman virtual influencer Ivaany, said that she has felt limited by the general range
and quality of Black hair on Daz 3D, adding that thankfully now she has found some specializing
artists. Photorealistic hair is known to be notoriously difficult to create, which leads many
creators to opt for purchasing hairstyle packages rather than creating their own. Therefore, non-
recognition often meant having to wear “standard” styles. Tatyana said that this had contributed
to her decision to give Ivaany straight hair in her first version:
“One of the problem that I feel we have as Black creators is that we don’t really have
realistic Afro hair, for example. That—sometimes people criticize that but it’s not really
my choice. It’s just that we don’t have a lot of artists who do beautiful—create realistic
Afro hair. So.”
To Tatyana, speaking out about the limitation was important because otherwise she felt that the
demand for diverse hair would not be noticed: “I always talk about the struggle that we have
with hairs because I think it’s also because some people don’t really know how to do it, or they
then maybe just don’t—don’t feel that people will need them.” Although hair was one of the
more outstanding issues, Tatyana said that the lack of diversity was not unique to hair but also
other facial and bodily features like skin and eyes. Similarly, Jessica, also a Black woman
creator, described the frustration she had felt with the lack of resources for Black digital humans
by linking the issue with the standards programmed and designed from White developers’ and
artists’ perspective, although not as something that was intended as a discrimination:
“The lack of Black model resources or you know textures and stuff like that, because I
feel like the creators of this software they were like predominantly Whites and they
created based off of you know, the references they had. It wasn’t like you know they
199
were trying to exclude you know, Black from there. They didn’t really have those
references and even ones they created weren’t really what I feel like a Black person looks
like because I’m Black.”
“So the resources that were missing, you know, during my—when I was trying to find
like Black resources, wear the right shade of Black. So there weren’t really like varieties,
because you know as White people there is like ones with yellow on the tone, red on its
own pale skin, a lot of options, but there weren’t like a lot of options for Black people.
We were just maybe one or two, you know, skin shades which wasn’t really you know
what I wanted. We needed more variety and range of different skin tones. The undertones
that you can, you know, merge together to create a more realistic Black-looking model.”
With the lack of recognition for demand and the resulting lack of resources, both Tatyana
and Jessica were grateful for any creators who attempted to accurately or beautifully depict
Black people, regardless of their race. They were aware of the criticisms directed at creators like
Cameron-James Wilson, a White man who created the famous Black female virtual model
Shudu, but nonetheless expressed their appreciation of their craft; Tatyana additionally
mentioned that Wilson had personally helped her learn how to create realistic skin. When
focused on social media where typically only the end results are posted, the representational
dominance of certain human types and the mismatch between creator’ body and their virtual
influencer’s body may seem to imply only oppressive relations and individuals’ uncritical
compliance with them. However, when technological actors’ capacities to make differences on
the creative process are considered, a more complex network of relations arise to the surface.
However, being considered as the standard does not necessarily equate with being
represented due to one’s dominant position. Dylan Mulvin’s book Proxies: The Cultural Work of
200
Standing In (2021) discussed what is now known as the “Lena” image, which is a cropped image
from a centerfold photo of young White female model Lena Forsén from an 1972 issue of the
Playboy magazine that has been used as the test image for digital image processing, such as for
establishing image standardizations like JPEG and MPEG. Mulvin wrote that the Lena image
embodies proxiness, emerging out of the history of militarization of computer vision technology
and control of women’s bodies. By being the prototype image, how Lena was seen has also
dictated how humans and technological actors will see, entangled with the politics of gender,
sexuality, race, and power (p. 144). Similarly, the prototype body in Daz 3D, suggested by the
tutorial and various external instructional materials, was that of a young White woman’s. As
discussed above, this meant that the program and its marketplace are more prone to seeing
Whiteness as the standard, which also intersected with the historical vision on women’s bodies.
The experience of designing a virtual human in 3D involved intimately seeing the virtual body
from all angles, detailly controlling and altering how it is shaped and how it positions itself in a
place that was designated by me (Figure 10).
Figure 10
Sex/gender-coded Base Poses in Daz 3D
201
Note. Notice the difference between kneeling poses, as well as those between chair-sitting poses.
Some of the female creators interviewed for this dissertation connected the virtual body
prototypes to the cultural history of fetishization and objectification of women, advocating for
the need for the industry to actively recognize female creators. According to Dulce, the
prevalence of female virtual bodies in the creator space was linked with the desire to gaze at
women’s bodies from a position of power:
“That’s the problem with the female—men dressing like females, because if you are
spending a lot of your time making cute females and that’s all you are doing, alright so
you’ve got to just keep making more stuff for your cute females, yeah. And I go back to
202
fetishes also because you know the big butts and the big boobs….Like seriously they’re
looking down at the screen. They want to look down at a female body, you know. They
don’t want to look down at a male body. Like that was some of the justifications for
wearing a female avatar some of the guys have given me.”
Gender-stereotyped conventions from adjacent industries such as fashion and cosmetics were
also mentioned as barriers that exacerbated the issue. Nina, a creator who also has experiences in
modeling, lamented the continuing thread of gender hierarchy across industries:
“There’s more female items 3D items out there for—in general, if you look at fashion
there’s more female clothes. You know, makeup for females. Same, it translates the same
way into 3D because, like mostly right now it’s like guys who are doing 3D and it’s their
understanding of what females should look like and what like the style should look like
you know. So, unfortunately, a lot of the clothes out there it’s like still very like
sexualizing and fetishizing….And so um yeah unfortunately there’s a lot of like limiting
items out there. Including like you know, clothing, accessories like etc, etc, so definitely I
feel like having a lot more of a female touch that like comes from a female artist is very
important because you could see the difference.”
For Nina, what was urgent was increased female representation in the level of creators, not just
in terms of depictions in end products. Likewise, Tatyana stressed that it is not the lack of Black
women creators that caused their relative invisibility in the industry but the structure that favors
gendered social capital. She explained that creators like her comparatively tend to not “get the
opportunities to reach out to a bigger audience” despite being sufficiently skillful, because some
of the opportunities required having appropriate contacts.
203
For these women creators, the practice of designing a virtual body was linked with their
lived experience of being a woman and how it will be shaped in the future, both in general terms
and as a member of the virtual human industry. These comments created a sharp contrast with
some of the more instrumentalist and distanced perspectives from male creators, including those
who did not see much issue with hypersexualization of virtual women (Chapter 2).
Body Copy Rights. We must return to the topic of body models to further discuss the
similarities between the Lena image and prototype virtual bodies. In an interview with the Wired
magazine in 2019, the original Lena expressed her wish to have been compensated for her
contribution to computer science (Kinstler, 2019). Like many other human bodies in
prototypes—particularly women’s (D. Black, 2012; Kinstler, 2019; Mulvin, 2021; Smits &
Wevers, 2021), body models’ presence in technological standards was fragmented and
obfuscated into seemingly docile invisibility. Of course, it is possible for a virtual body to be
built completely from scratch in 3D. However, in practice, there are many human bodies woven
into the design of virtual bodies—even fully 3D ones, both in terms of providing the basis of the
body (parts) but also likeness.
One of the most visceral examples I encountered during my research regarded skin
textures of virtual humans. As part of their instructional video series, 3D model store RenderHub
(2018) posted a video that explained how to create photorealistic human skin textures for Daz 3D
and other programs. The video mentions “a very unique [line of] product” by an artist named 3d-
sk: skin textures created from “very particularly” photographed pictures of over 60
commissioned human models of diverse backgrounds (Figure 11). A photo set contains high-
resolution images of various body parts (Figure 12), which then can be used as image references
for making the default skin texture more realistic. What made this such a visceral example was
204
the end result: a literal human skin, resembling a bear-skin or tiger-skin rug—stretched out,
processed, and laid down as headless, nameless part of the interior ensemble, long away from
where they once roamed (Figure 13).
Figure 11
3d-sk’s Human Model-based Skin Texture Product on RenderHub 3D (2018)
Figure 12
A Sample Photo Set (RenderHub 3D, 2018)
205
Figure 13
Edited Skin Texture Using the Sample Photo Set (RenderHub 3D, 2018)
206
While physical human bodies tend to be rendered invisible during this process of
fragmented displacement, their likeness may remain in prototyped virtual bodies. As discussed
above, despite being considered as a tool or an ingredient, body models influentially contribute
to virtual influencers’ presentations, regardless of the design method. A gig-based hiring
advertisement for a virtual influencer body model (sdlkfj, 2022) encouraged beginners to apply,
reassuring that their face will be replaced with a different one. At the same time, they listed
various conditions their applicants should meet, such as having a cute and attractive face with
double eyelids and a short hairstyle, perhaps with choppy bangs. They explained that this is
because they expected the virtual influencer to have remnants of the applicant’s face: “We wish
to find someone with the face we are looking for because the model’s face does not disappear
completely but leaves some feelings of it [on the virtual influencer’s face]” (sdlkfj, 2022). This
then would serve as the prototype of the virtual influencer.
There were also cases where more than “some feelings” of likeness was extracted and
datafied. A Chosun Ilbo article (B. Lee, 2022) reported on the story of an anonymous
professional model and actor who claimed that their right of likeness was stolen by a virtual
human company. They claimed that they were paid around 1,200,000 won (~$1,150) for what
they had known as a 360 degree video capture gig to be used as references for a one-off music
video appearance project. Instead, the company used AI to create a moving virtual human with
the data, after which they notified the body model that the virtual human will expand into
influencer and modeling jobs. The body model was exasperated: “the assets I accumulated from
my modeling career have been robbed.” One of my interviewees, Dulce, also shared a similar
experience that she had after she scanned her body as part of high-fidelity avatar giveaway at
DOOB 3D, a photorealistic 3D replica company:
207
“One day I’m on Facebook and I’m scrolling through. And all of a sudden I see my
avatar in a picture. But I was never there. Like I never—it’s me, my body as an avatar. So
I made a comment I’m like I don’t know what you guys are doing but that’s not cool
because I was never at that thing you know. And this looks weird. Why are you using my
avatar? And then they pulled it down because they realize that that’s weird, right?”
Both the body model and Dulce have become articulated to the respective virtual bodies’
performances. They are simultaneously the original body (parts), the standard for consistent
likeness, and a set of data that can be repurposed by various associated actors, not limited to
commercially-driven human actors but also nonhuman actors like the AI that co-created the body
model’s replica (see Cheney-Lippold, 2017).
The issue of likeness is not only limited to commissioned body models; it could be
applicable to unknowingly recruited physical and virtual bodies, in complex entanglement with
technological and legal actors. Dulce shared two incidents in the virtual world Second Life as
part of her comments on “copyrights” of virtual bodies. The first involved celebrities. She
explained that in response to the surge of Angelina Jolie copy avatars, Linden Lab, the company
that created Second Life, decided to ban celebrity lookalikes. They forcibly pulled avatars with
branded star names out of users’ inventory, although some users got around it by using different
names or slightly altering the physical traits, such as by using black hair instead of blond for an
avatar that looks like Paris Hilton. There are virtual influencers that resemble popular celebrities
as well. For instance, there have already been at least two virtual influencers that has been
recognized as being virtual lookalikes of Cha Eunwoo, a Kpop star who is famous for his
handsome looks: Jeong Saejin from Superkind and Wooju from Klleon (Figure 14; Kim, 2021;
“sal~jjak chaeunu? joshua?,” 2021). Such a design is a strategic, intentional choice. Annika
208
Kessel mentioned during the interview that her company’s co-founder urged the artists to design
their virtual influencer Aya Stellar to “look something between Kate Moss [inaudible] and some
other famous actresses [who] are really pretty” because they believed that being pretty and
special were the reasons why many celebrities gained fame. Unlike the Second Life case,
however, there has not yet been a clear guideline or regulations on the issue of virtual
influencers’ likeness (B. Lee, 2022).
Figure 14
Two Virtual Influencers and a Kpop Star
Note. Virtual influencer Wooju (Kim, 2021), Kpop star Cha Eunwoo (J. Lee, 2019), and virtual
influencer Jeong Saejin (“sal~jjak chaeunu? joshua?,” 2021) (from left to right)
The second incident Dulce shared regarded virtual bodies. She explained that a Second
Life head maker filed a lawsuit against another head maker, claiming that they simply updated
the original version. Linden Lab shut down the claim saying that they seemed different enough,
but Dulce added that with programs like Daz 3D that provides certain prototypes, originality can
be difficult to trace: “These heads all originated out of an app like Daz so even, even then it’s a
company [that] made the original head, right?” That is, likeness may not necessarily be the result
of human intention but that resulting from technological actors’ conspicuous and inconspicuous
209
influence on human co-creators. Tom explained during the interview that while Daz 3D has
improved a lot and is very easy to use, many virtual humans created out of the program tend not
to have much “personality” because they inevitably draw on mixes and matches of prototypes:
“it’s like buying a very cheap car that can go only on specific streets.” He also commented on
how once artists “get locked into this specific software,” they “can’t get out.” That is, Daz 3D the
program, not only the prototypes within the program, can become a standard. He added that
although there is more room for detailed flexibility in more advanced programs, most virtual
influencers currently on Instagram seem to have been created from Daz 3D.
Approaching what gets transferred in this process as informational bodies and cultural
meanings can help legally respond to the problem of alienated datafied bodies, including those
regarding prototypes. Luciano Floridi (2013) argued that information should be viewed not as
what people own but what constitutes them. Drawing on this perspective, Carl Öhman and
Luciano Floridi (2017) critically analyzed the commercialization of deceased individuals’ digital
data in the digital afterlife industry. According to them, “informational body” of a human can be
understood as their “inorganic body” or personal data they produce, and having control over it is
“the essence of human condition” (p. 650). Therefore, they argued that it is ethical for deceased
individuals’ informational body to be treated with respect, and that policies and regulations that
protect human dignity of deceased organic bodies, such as those relevant to museum exhibitions,
should be considered for regulating the digital afterlife industry (see also Steinhart, 2014). Also
lacking a physical body, virtual influencers can be thought of as those who live on through their
informational body. However, due to this lack, the construction and maintenance of their virtual
informational body requires data from various human informational bodies. This interpretation
allows the application of policies on human organic and informational bodies to virtual
210
influencers, while still granting priority on human informational bodies as humans have sourced
the data that gets reprocessed into virtual influencers’ informational bodies. As Hye Jean Chung
(2015) said in regard to motion-capture supported cinematic reanimation of dead bodies, “the
cyborg hybrid indicates that the digital is not disconnected from the physical world” (p. 66).
This approach can embrace not only body models but also those who unknowingly
sourced their likeness, such as in the case of unapproved celebrity lookalikes. Cultural
anthropologist Grant McCracken (1989) conceptualized celebrity endorsement through his
“Meaning Transfer Model.” Here, celebrity is understood as a network of cultural meanings
emerging from their professional and personal endeavors, and endorsement is understood as a
symbolic transference of these meanings onto the brand by being incorporated into the
celebrity’s meaning network. Law scholar Andrew Gilden (2021) referred to this model to
discuss posthumous endorsement laws, explicating how differently situated actors can affect the
symbolism and dignity of the deceased. Gilden suggested “privity and power” as the general
guideline for courts, namely whether an entity using the celebrity brand for endorsement has
ownership of the artifact that is signaling it and also the legal empowerment to make decisions
on the behalf of the deceased. These are informative starting lines for drafting regulatory
frameworks for virtual influencers. If approached with the Meaning Transfer Model, what some
of the virtual influencers are doing is benefiting from unsanctioned and unregulated association
with the celebrities through their likeness and other components of their informational body.
There is an additional risk for celebrities in this; meaning transference could happen in the
opposite direction as well, which means that the presentations of similar-looking virtual
influencers can affect the celebrity’s image. This model need not be limited to celebrities. A
211
similar process can happen between a virtual body and a physical body that representationally
corresponds with the virtual body (Chapter 2).
Privity and power can be guidelines for companies and independent artists interested in
creating virtual bodies through an ethical process, as well as those who wish to protect their
informational body during a related gig work. They are also useful lenses for thinking about the
ethics of alienated bodies, including regarding the influence technological actors have on shaping
the phenomenon. For instance, we can ask which actors were involved in the delegation of
decisions on the fragmented bodies and how ownership is narrated; technological standards can
be one of such involved actors. However, as exploitative uses of NDAs illustrated, there is a
chance for both principles to be abused. Therefore, specific policies that can structurally protect
the hidden bodies should accompany the guidelines.
Platform Politics
Social media platforms are also important nonhuman actors, particularly Instagram. As
the primary platform for virtual influencers to present themselves, Instagram is an infrastructural
actor. During one of my interviews in October, 2021, Facebook/Meta services including
Instagram went down. Possibly associated with this unexpected breakdown, my interviewee had
to intermittently excuse themself to attend to urgent requests. The ever updated feeds of virtual
influencers were all frozen in the past for several hours, still smiling but unresponsive. During
the interview that was scheduled a few days after, Aba Wils’ creator Reggie reflected on the
breakdown:
“So I was little bit worried when when Instagram, Facebook, and you know Whatsapp
went off for like six hours so. I was like wow because I had already done it. Posted. Then
212
I needed it to—it was a video actually, so I was just [inaudible] and all of a sudden, I
couldn’t see anything again. Then I read on the line, babababa, Instagram down, I said oh
my God. I needed it to come back. So I even tweeted about Aba Wils on Twitter that Oh,
I think, oh, “Facebook, I hope you get Facebook, Instagram, some something back.””
The urgency is understandable because virtual influencers’ existence relies on social media
platforms. Diversifying the platforms that they present themselves on can help distribute the risk
but increases management labor, even if the virtual influencer is posting the same content across.
Therefore, individual- or smaller team-managed virtual influencers tend to concentrate on a
single platform, which often is Instagram especially if their main content tends to be picture-
based.
That is, while virtual influencers may not have been physically bound to the pandemic-
struck physical layer, their corporeality is bound to the life and fickleness of platforms. As
Reggie said: “So virtual influencers we need social media to work. We need social media. They
have social media.” In fact, virtual influencer Cade Harper’s discontinuance and the account’s
transformation into a different brand has been described as “deceased” on VirtualHumans.com
(Who Is Cade Harper?, n.d.) despite some remaining traces of his informational body elsewhere.
A communication scholar of Internet histories, Frances Corry (2021) discussed death discourses
around now discontinued social network platform Friendster, arguing that platform death is a
useful analytic lens into obscured relations among actors in platformized media economy. Some
of the causes of Friendster’s demise suggested by their former employees included technical
issues, user community, and misguided visions. As members of Instagram’s user base, virtual
influencers and their team can contribute to the future course of the platform but not decide nor
reliably foresee it because they are still only one type of contributing actors. Moreover, many of
213
the platform-end issues may not be so clear for those on the user end, if not for the employees
(Whitson, 2018). What can be experientially known is only the consequences of the platform’s
actions, such as the temporary breakdown of Instagram during the interview, and various actors’
discursive practices around them.
Instagram’s influence on the phenomenon as an infrastructural actor is not limited to
grander scale and unexpected events like breakdown or death, but regards everyday actions of
connected actors. Various scholars have discussed how algorithms are shaped by and shape
social relations, as well as the lack of transparency this process (e.g., Ananny, 2016; Ananny &
Crawford, 2016; Beer, 2009; Bucher, 2017; Cheney-Lippold, 2017; Cotter, 2019; Gillespie,
2014; Kitchin & Dodge, 2011; Noble, 2018). Social media algorithms determine the visibility of
contents based on its often opaque logic around user engagement, accurately playing by which is
perceived as one of the key determinant for (continued) success of an influencer (Bishop, 2021;
Cotter, 2019; Duffy & Hund, 2019; Hearn, 2010; Petre et al., 2019). What users do in the
absence of transparency is come up with their own and collective “folk theories” and
“algorithmic gossips” of how the algorithms are curating their content (Bishop, 2019; DeVito et
al., 2017; Eslami et al., 2016; Ytre-Arne & Moe, 2021). Based on such interpretations of
Instagram’s algorithmic architecture, (aspiring) influencers “play the visibility game” by
optimizing their presentation through engagement tactics, for instance by striving to build deep,
sincere relationships with their audiences or by simulating such relationships’ presence (Cotter,
2019), which then affects how platforms like Instagram frame and adjust “the game” as per their
interests (Petre et al., 2019).
Virtual influencers also play the game of visibility, strategizing their actions to best cater
to the fickleness of Instagram algorithm. Some interviewees mentioned that they visually and
214
behaviorally present their virtual influencers in a specific way that would best garner audience
engagement on Instagram, such as the aforementioned examples of Aya Stellar whose design
intentionally referenced attractive actors and the virtual influencers Xander worked on, Binxie
and Aliza, who were designed in reflection of consumer research data. This was not limited to
company-run virtual influencers. The independent creator of Mikirah Muse said that he has been
emulating the ways typical fashion influencers post on Instagram to build Mikirah’s followers.
He mentioned that his background in influencer marketing had taught him how to strategically
approach audience engagement and the business of influencers for a better positioning of
Mikirah on Instagram.
The need to factor in Instagram’s influence led some creators to involve mediating actors.
For instance, Jawad is the original creator of Serah but his work is now focused on her technical
side. He showed great trust in his team of seasoned managers who have been taking charge of
Serah’s character presentation. The mediating role was not limited to human actors. Independent
artist Ted described himself as someone who had not managed his virtual influencer Lena’s
Instagram account for money nor fame, but to express his artistic vision. Regardless, he
suggested that using appropriate hashtags for visibility was a mundane but necessary part of his
work process. Ted explained that he used an external program to generate hashtags for his posts,
the results of which he selectively—although not with great care—copied and pasted into his
posts. He said that this process was separate from his artistic flow, which he repeatedly
emphasized was what he really valued, but was nonetheless a necessary step because “without
hashtags you get no attention, sometimes just nothing.” The hashtag program is a technological
actor that was introduced to mediate the relationship between Ted and the Instagram algorithm.
Although Ted described the hashtag program as a mere delegatory tool that he was using due to
215
his laziness (“I just want to concentrate on the creative process and that’s it”), it was an
influential actor with a folk theorized volition that critically contributed to the self-presentations
of virtual women that Ted designed:
“I just copy paste them over and that’s it. Sometimes, sometimes I read them just to make
sure not to…yeah, sometimes it’s it’s rubbish it’s it’s completely off. Don’t fit my
picture.”
“It creates hashtags from three words or up to five words. So I just use right “DAZ3D,”
right. “Model,” use. Right, “sensual art,” right. And then, it creates all the other stuff
by—or it has huge hashtag lists with relevant entries and that’s I think how it’s creating
the hashtags.”
The visibility game for virtual influencers was not only about standing out but also about
avoiding being shut down, which in virtual influencers’ case would mean the loss of their main
informational body. Similarly with folk theories for better visibility, users develop their theories
on how content is moderated on social media and take tactical actions to best evade censorship
and demonetization (Caplan & Gillespie, 2020; West, 2018). Jawad recalled that now-
discontinued Cade Harper’s positioning as “the friend of everyone” was extremely well-received
and successful, allowing him to grow from zero to around 100,000 followers within a month. He
said that this growth was what led Cade’s account to be locked by Instagram because “they
believed we were a scammer.” Sexualized virtual influencers were particularly wary of content
moderation. Virtual influencer Monna Haddid, interviewed as herself, said that she has a “love-
hate” feeling towards social media platform TikTok: pleased at its traffic rate, but displeased at
what she described as its “double standards.” She had accrued more than 1 million followers on
TikTok but ended up being banned on the basis of “alleged” violations of community norms.
216
Although she did not respond to my request to elaborate on the details of the allegation, she
denied the accusation by claiming that in actuality this was because “there are accounts that are
dedicated to banning accounts” and one of them had reported hers. She explained that she was
very frustrated from this experience because she could not contact and explain her case to
TikTok, which led her to create an additional backup account on Instagram just in case. She
needed a duplicate informational body.
In fact, the visibility game was a dilemma for sexualized virtual influencers. Wendy
Watson (Instagram profile [redacted]), a minor-reminiscent self-proclaimed 19-year-old female
virtual influencer, has on multiple occasions decried Intagram’s “algorithm bias” (post on July
25, 2020) in “her” voice. Wendy strove to win at the game of attention. She had purged her
followers who mass follow accounts on the belief that Instagram algorithm will consider them as
spam accounts and therefore hers, saying that she “should become an Instagram consultant.”
(post on August 4, 2020) She also claimed to have gone through a series of experiments and
analyses on her audience engagement which led her to believe that Instagram was intentionally
minimizing her exposure—an algorithmic disfavoring known as “shadow banning.” Commenters
concurred across various posts, additionally blaming those who contributed to the situation by
reporting her account. Excelling in the attention round of the visibility game was a poisoned
chalice for Wendy because her presentation was one that could be easily read—or misread,
according to Wendy—as sexualization of not just any young women but of underage girls (see
Chapter 2). Wanting more attention, she had paid to be on a shout out page but was removed
soon after, which she guessed was because of her controversial depiction (post on June 8, 2020).
She self-mockingly described her regular use of hashtags as “#ShootingTheAccountInHead,”
since more exposure meant more chances for her to be reported. Wendy seemed to genuinely
217
believe that her depiction was not problematic (“I wasn’t aware that only haggard bitches or
double-D’s that look like they drank and slept their way through their twenties were allowed to
be models, wear swimsuits, or put up photos online”; comment on post on June 8, 2020) and it
was rather “liberal hypocrisy coming from killers of freedom of speech like Instagram” (post on
July 26) and their supposedly oppressive algorithm that were at fault. Regardless of the
frustrating dilemma, it was still important for “her” that she continued to play and strive to win at
the visibility game because that was the business of Wendy Watson the scantily clad 19-year-old
virtual influencer: “this is business; my heart’s for my family and my brains & lady balls are for
business” (post on August 4, 2020). Disobeying the perceived will of Instagram meant not only
the demise of Wendy Watson on Instagram but also her as a business.
Crossroads
Local Cultures. Intersecting contexts should be considered in understanding the multi-
directional networks of human and nonhuman actors. The phenomenon does not happen in a
void, and certainly not if the void presumes certain conditions as the default. Physical location is
one of these context-relevant actors. To hopeful Metaverse enthusiasts, physical localities may
seem relatively less important since it is the virtual layer of social media that the virtual
influencers populate. However, physical location and related regional cultures were still highly
relevant to both virtual influencers’ social media self-presentations and their behind-the-scenes
dynamics. An example regarding materiality would be the difference between how Tom and
Jessica differently related to electricity in their respective countries, France and Nigeria,
discussed in the section on technological actors. Rendering high-quality CGI images was
perceived as a time-intensive task to both creators but Jessica had to additionally work around
inconsistent electricity in her region, although she explained that her lived knowledge allowed
218
her to easily plan for it. Jessica had also mentioned the relative weakness of her country’s
currency and the cost of standard software in the industry.
Local context also affected virtual influencers’ presentations and how the creators related
to them, particularly those from non-Western countries. Some of these creators said that they
were intentionally limiting their targeting to domestic audiences due to their non-English
national language or due to the belief that their virtual influencer may be too domestically
oriented for global audiences. Daichi Kambayashi and Sonya from Atali said that while they
think it would be more ideal for Meme to have both Japanese and English in her posts, they
posted primarily in Japanese because not everyone on the team could speak English. Kambayashi
said that the auto-translate button within Instagram may partially compensate for the absence of
English, but expressed some doubt towards fully embracing the technological actor as part of the
team, this time commenting in English instead of Japanese that he had been speaking in: “But I
know that [Instagram’s] translation is horrible.” Seihee Kim from Klleon also said that their plan
for global market is to create new localized virtual influencers instead of focusing on expanding
Wooju, because they believed that he “may not be a character that would not work in the West”
since he emulated an “ordinary male University student in Korea.” She added that he only spoke
Korean as well. As SIABangkok, the company behind Thai virtual influencer Ailynn said,
English was approached as a necessity “to communicate not only to Thai people but also
international fans.” While virtual influencers have been discussed as beings that can be anywhere
on the pandemic-hit globe and the World Wide Web (e.g., Teh, 2021; Whateley & Bradley,
2020) their strong association with a certain physical region meant that their web was not so
worldwide.
219
There were also virtual influencers who purposefully aimed to showcase their domestic
culture to international audiences as a way to counteract regional marginalization. Tatyana, the
Haitian French Guianese creator of Ivaany, explained that she chose Ivaany’s cultural
background as Haitian, French Guianese, and Brazilian because the former two represented her
background and the last indicated the physical and cultural vicinity of Brazil to French Guiana.
She described French Guiana as a country that is generally not recognized or was misconceived,
which she was greatly motivated to change through Ivaany’s global reach:
“I want to to show and and share with people my country, French Guiana. Because we
are a French country but we, we are in the in the South America, near Brazil. And people
don’t know the story of this country. They don’t know where it is, they just think oh it’s
somewhere in Amazonia, there’s just trees and people don’t have car and everything. So I
want to show people how beautiful my country is and I want to do it with Ivaany. And
also the fact that I’m Creole. I want to show the French Guianese and Haitian culture, so
people will see, because I think we actually we don’t have French Guianese influencers, I
think we have a few YouTubers but not really big influencer so that’s what I want to do
with Ivaany.”
She explained that she began to do this through Ivaany’s fashion, for which she used unique
fabric called madras from the West Indies. Tatyana additionally expressed her great desire to
eventually have Ivaany do Kasékò, which she described as an “amazing dance” that comes from
the Guianese Creole term for “break your body.” She mentioned how African American and
Black American cultures, as well as her support for them, have also been incorporated into
current presentation of Ivaany, alluding to the global reach of American popular culture: “we all
grew up with African American influence.” In other words, Ivaany was regionally and culturally
220
informed, and Tatyana wished that Ivaany would inform others in different localities. This was
best reflected in Tatyana’s understanding of Black cultures; she believed in recognizing and
embracing diverse intersections between race, ethnicity, and region. As Tatyana aptly said,
“there is not just one Black culture, I would say there is different culture in the Black culture.”
Thus, she wished Ivaany to be a virtual influencer who would embody and represent Black
French Guianese culture, while also empowering Black people from around the globe through
her representation of a Black culture.
These accounts led me to reflect on my own contextual biases. What were my standards?
I am an ethnically Korean East Asian woman from South Korea who has lived in China, Canada,
Japan, and the USA. I speak Korean, English, and Japanese, in the order of fluency. For the last
five to six years, I have been living in Los Angeles, USA for my doctoral studies. Perhaps a
mouthful to some, my background has occasionally granted and forced upon me the positionality
of a person representing “diversity.” However, I must admit that I was shocked at my own
preconceptions when Reggie described virtual influencers’ capacity to switch the style and
length of hair at will as not a uniquely technological trait, but an ordinary practice. Reggie
explained that in Ghana it is common for women to own and switch between multiple wigs, and
therefore it was natural for him to have Aba wear multiple hairstyles:
“I was trying this hair [in the software], the other one the other one so when I try this one,
it was kind of so cool and I showed it to my guys and we’re like wow I like this hair and
oh okay. This is, you know what they do here. Our girls what they do is they might even
have a short hair. You see them to do with a shortcut. Tomorrow, or in the evening you
might see them with a different hair, because they have all this kind of wigs that you put
on all the time. So I say oh okay, so let me give you [Aba] that vibe.”
221
I must also confess that my regional location and fluency in English had given me some
advantages. Anxious of difficulties in recruiting interviewees, there were times when I
knowingly alluded to my membership in an American university and my current residence in Los
Angeles, a desired location where many virtual influencers self-tag. It is unclear exactly how
much power this information had over my fieldwork, but there were some interviewees who
hinted at their interest in my location and English skills. Some showed delight in that someone
from a different country, especially the USA, had reached out to them. What particularly struck
me was how my interviews with Korean creators wrapped up, both of whom expressed their
interest in keeping in contact. One of them brought up their plans to visit the Customer
Electronics Show in Las Vegas, suggesting that I come. The other interviewee said that they are
happy to have more exposure of the company’s name in the US because they have plans to
expand here, which they described as a must. This statement was preluded with a brief mention
of their relatively lacking fluency in English and concluded with a quick comment on how they
may ask for my US-based expertise when they begin their expansion project. My physical
location and language capacities meant more opportunities and ease. Or, perhaps both were
formal pleasantries. However, I could not help but think that the codes that are claiming to
program the free, “new” world for all are written in English.
Intersecting Industries. Accounting for industry crossovers is also useful for
contextualizing traced virtual influencer networks. While it is possible for virtual influencers to
be created by an independent individual entirely in 3D, their proximity to fashion and
entertainment industries meant that some of these industries’ conventions have been affecting
and challenged by the phenomenon. For instance, after describing the hardships she experienced
working as a model, Nina argued that digital twins will be an opportunity for models because it
222
can give them the flexibility to be in any place without having to fly in every time, as well as the
capacity to “look young forever.” She added that she did not feel that virtual models are
particularly threatening because “it’s always going to be competitive.” In other words, she
believed that having a virtual body would provide human models new ways to negotiate with the
industry’s conventions, although it was assumed that they would remain, such as the pressure on
age and its competitive structure. To revisit one of the cases discussed earlier, Alexsandrah
Gondora, the first-ever body model for Black female virtual model Shudu, refuted the criticisms
directed at Shudu’s White male creator Cameron-James Wilson by arguing that if any entity is to
be blamed the criticisms should be directed at the fashion industry because it is them who has
been hiring Shudu, a virtual model, over human models (The Diigitals, n.d.). In the interview
video, she additionally shared her experiences of being discouraged and discriminated against for
being a Black person in an industry where Whiteness is the standard, including about how she
had lost hair during her professional modeling career due to excessive heat treatment from
people who seemed not to understand Black hair. For her, the rise of virtual influencers was not
simply a novel phenomenon but a novel opportunity for marginalized members of adjacent
industries to find work and be recognized. She argued that unlike the traditional fashion industry
that systematically disadvantaged Black women, Wilson always hired real Black women
whenever he got a job for Shudu, both models and staff; according to Gondora, her biggest job
had been with Wilson.
Standing in for a (virtual) star and being approached as an invisible ingredient for the
product were generally not approached as something new, especially in justifications of the
phenomenon. As discussed previously, virtual labor (Stahl, 2011) is not unique to virtual
influencers but stems from the historical division of alienated labor in cultural industries. The
223
unfairness of this tradition was narrated as what made virtual influencers superior, whose
production process somewhat paradoxically followed the similar structure but without the
presence of a single human star who would represent the virtual labor. Trever McFedries, the
CEO of Miquela’s management company Brud, on multiple occasions used the term
“decentralized celebrity” in podcast interviews to compare his vision on virtual influencers with
the traditional celebrity model (e.g., Herndon & Dryhurst, 2020; Pat Lok, 2021). He argued that
despite the romanticized myth of the lone genius, pop stars are results of the creativity of various
individuals who work under an incentive structure that is not always so equitable. He believed
that a decentralized system that is fronted by a virtual body like Miquela’s would not only enable
these hidden figures to be aptly compensated, but also allow different talents to be more flexibly
combined, which he argued would lead to more opportunities for people with different skill sets.
He also stated that this would prevent sufferings of the human star in the traditional model as
well, nonspecifically referring to Kpop stars’ suicides and idol trainee system. Similarly,
Jaewook Oh, the CEO of Rui’s company DOB Studio, mentioned that Rui’s body model’s
negative experiences as a Kpop trainee, such as precarities and gender-stereotyped lookism, led
her to agree with being the faceless talent behind Rui. He also mentioned some of the stars’
nonrecognition of various individuals in creating the virtual labor upon reaching success, arguing
that the character-led virtual influencer model can grant companies some security in the form of
IP (intellectual property):
“Traditional influencers and celebrities believe that they got famous 100% through their
own appeal, so they tend not to understand how many people had supported them in the
background and how many people participated in their storytelling, helping with their
image-making and with sales, the efforts to create a business. I know many stars but
224
many of them unfortunately cannot think beyond ‘I became famous because I’m
talented.’ So now whenever I meet with talent agencies or MCN [Multi Channel
Networks; digital platform-oriented media] companies, their administrative people, they
always talk about the same frustration. Celebrities and influencers do not understand how
much effort, risk-taking, and investment this involved, and it always leads to dispute after
they become famous, which made me think that the case of Pengsoo [a giant penguin
character celebrity in Korea] is exemplary in this sense as well. Yes, the company has the
IP—the company has the character’s identity including their appearance, so there can be
some balance between the two even after the influencer became extremely famous. This
leads to the sustainability of the business and a factor that allows the perpetuity of the
company, which makes this all very impressive to me.”
Since both are grounded on their experience-informed knowledge about the entertainment
industry, I find McFedries and Oh’s reasoning to be what stems out of sincere, legitimate
concerns about conventional practices. However, the continuance of the basic structure of virtual
labor in the newer model, just substituted with an intellectual property instead of a human, leaves
many questions on remaining patterns. Granted, McFedries’ vision was inspired by the notion of
decentralized autonomous organization (DAO)—a bottom-up community model of blockchain-
based governance in the cryptocurrency industry, which means that if the goal of decentralized
celebrity can be reached then all participating individuals should be permanently accredited,
compensated as per their share, and invited to participate in the virtual influencer’s future
trajectory. Still, envisioned through a structure that is based on non-fungible cryptocurrency, it
cannot be disregarded that it may create new inequities from how the system structurally
advantages early members and how affluent people (in terms of traditional currency) can exert
225
more power over the speculativeness of the crypto token’s worth through their capacity to yield
larger economic assets. These are also related to the issue of authorship and control that
motivated the decentralized model. While a participant’s identity may permanently remain in the
smart contract-based ledger, once they sell their token—perhaps a more likely possibility for
economically disadvantaged individuals, their control over their contribution, which may be in
the form of likeness or unique talent, is lost until they can acquire the tokens for DAO
membership again. Artists may also be not happy about the remnant premise of invisibility in the
virtual body-fronted model. Although Rui’s body model was eager to contribute, Oh had
mentioned that many artists before her turned down the offer. According to Oh, the reason for
rejection was uniform: they felt that it is only meaningful if they become successful in their own
face. For these artists, their likeness and talent may equate with the core sense of their identity;
they are not mere properties but their intellectual body.
Another thing to note is that some of the companies may align themselves with the
technology industry rather than entertainment, even if where their virtual influencer
occupationally belongs is the latter. McFedries has described Brud as a mixture of technology
and entertainment company, and Oh described DOB Studio as a deepfake technology company
that is interested in eventually expanding into the entertainment sector. Similarly, Seihee Kim of
Klleon described their virtual influencer Wooju as what they hoped would advertise their facial
replacement technology.
One of the common concerns that I observed from entertainment fans about virtual stars
was superficial co-optation of their fandom cultures. In the comment section of the YouTube
video titled “VIRTUAL IDOLS Will Replace Traditional Kpop Idols?” (BeBoss TV, 2020),
Kpop fans and non-fans expressed hopes and fears at the prospect of virtual humans in Kpop,
226
including some of which that stressed the differences between Kpop, anime, vocaloid, and other
subcultures related to the topic of the video, showing annoyance at the lack of in-depth cultural
knowledge in the video. The most liked comment on the video respectfully rejected the prospect,
saying that it is the idols’ years of hard work that earned their respect especially because the
idols “were just like you” before training.
It is possible that the title of the video may have prompted some of more negative
responses, but an example that had successfully evaded these criticisms through an
understanding of fandom culture would be the first music video released by deepfake-based
virtual Kpop idol group Eternity, titled I’m Real. The group was created by Pulse9, a
“technology company that communicates via AI technology,” as the first of their “projects to
research and improve face-providing technology” (PULSE9, 2021). The music video was also
received with mixed reviews. More critical Kpop fans and users poked fun at Eternity’s
uncanniness and “not real” origin, the latter of which was jokingly conflated with negative
characterization of Kpop idols as manufactured products. Some comments expressed their
concerns regarding human trainees and idols, but the comment section tended to either
negatively or positively accept the group’s nonhuman origin as their “concept,” which is a term
used in Kpop to refer to the presentational theme of an idol(s). What helped shape this generally
accepting view was the company’s self-positioning. In the form of pinned comment and template
replies to each negative comment, the company appeased Kpop fans by emphasizing their
identity as a technology company: “this project was created not intending to substitute K-pop
idols nor exploit them, which may be the concern for many people who love K-pop, but is rather
to expand AI technology.” Pleased at the company’s active communication and reassured, Kpop
fans discussed how they felt about the novelty of the concept and how it could be improved
227
while reminding critical commenters that technicians’ feelings can be hurt. The comments on
their subsequent music videos, one of their member’s solo song No Filter (PULSE9, 2021b) and
the teaser for their upcoming release Paradise (PULSE9, 2022), were also generally supportive,
especially with Eternity’s noticeable technical improvements. Having accepted Eternity as a
“concept” group, Kpop fans commended the company for listening to their feedback and happily
welcomed back “my girls,” who were dancing and singing in bubbly cheerfulness like any other
hard-trained human Kpop girl idol group.
There are largely three things that I want to point out with these cases. First, virtual
influencers borrow conventional codes from various existing industries. Second, these codes can
be formed by those at the company-end, but also by and together with fans. Third, these virtual
influencers can flexibly locate themselves in or near different industry contexts as needed, such
as by using “technology company” as their pivot. In fact, virtual influencers like Rozy and Reah
Keem have announced that they will be releasing music, in addition to having appeared on a
drama series as a cameo in the case of Rozy (J. Kim, 2022). Since the links have been
established, there have also been a rise of traditional entertainment companies expanding into
virtual stars, such as SM Entertainment and YG KPlus. We should welcome exciting crossovers
but with vigilance towards how interconnected contexts are being rhetorically amalgamated and
separated. Analogous would be how social media platforms and online content aggregators
strategically self-brand themselves as technology companies for investment and regulatory
benefits, in spite of how their social role fits within the parameters of media companies (Napoli
& Caplan, 2017). We must not forget that the phenomenon does not involve only virtual bodies,
but a myriad of human and nonhuman actors who are materially and discursively contributing to
its current shape, negotiating with existing and emerging patterns.
228
Money, Media, and Intermediaries. Last but not least, marketers and secondary content
producers, such as journalists, are actors that play an important role in forging and mediating
relations to legitimate the phenomenon. That is, they assure virtual influencers’ continued
existence by securing monetary flow and by affirming their presence and impact through
derivative content. It is possible for an actor to embody both roles, either knowingly or
unknowingly.
Marketers, here broadly referring to actors that are interested in facilitating the influx of
monetary capital through strategic positioning of virtual influencers, attempt to influence largely
three clusters of actors: investors and brand sponsors, general and target audiences, and their
production team colleagues. It is possible for the first to be discounted in non-profit virtual
influencers or small-scale ones that rely on fan patronage, such as through external platforms like
OnlyFans or Pateron. In larger scale or company-run virtual influencers, attracting and appeasing
investors and sponsors are crucial to virtual influencers’ continued success since they are the key
actors that fund the project. As Jawad said, “brands, yeah, investor, you know you’re working for
them you’re not working for yourself when you are involved in the business when money is in
game.” This required communication with the other two groups to demonstrate the virtual
influencers’ impact through a presentation that can strategically encourage audience engagement.
Recognized as a fresh media format that involves cutting-edge technology, there are both
interested curiosity and concerned skepticism around the phenomenon’s economic prospects.
Mews media helped with the process of legitimizing the business of virtual influencers by
reiterating their investment and sponsorship records, as well as other information that could
signal their profitability. For instance, Miquela’s company Brud was often referenced as an
example of a company with impressive investments, accompanied by details such as having
229
garnered approximately 6 million dollars of investment from investors like Sequoia Capital and
at least 125 million dollars from their subsequent investment round (Shieber, 2019; Spangler,
2020; Tiffany, 2019). Big-name brand endorsements (e.g., Calvin Klein, Samsung, Prada),
audience-related metrics (e.g., number of followers, number of listeners, channel growth rate),
and organizational association (e.g., Miquela’s membership in Creative Artists Agency, Knox
Frost’s collaboration with World Health Organization) were also frequently mentioned (Ambit
Research, 2019; T. Chen, 2020; Evans, 2020; Pyman, 2020; Spangler, 2020; Whateley &
Bradley, 2020). In turn, creators used news reports on them as attestation of their viability (e.g.,
The Diigitals (The Diigitals, n.d.), 1 Sec Inc. (1SEC inc., n.d.)).
These media reports were a mixture of press releases pushed by marketers and news
organically generated by journalists, published in both general public-facing outlets and more
specialized outlets such as in stock market analysis in economic newspapers (G. Kim, 2021;
Maeilgyeongje, 2021; H. Yoo, 2021). In fact, during the month of July, 2021 nearly all of
collected Korean news on virtual influencers centered on or mentioned Rozy, beginning with a
swoop of similarly titled permutations of press releases from Shinhan Life Insurance that hired
her in their TV commercial. Indeed, Seihee Kim from Klleon explained that a part of her PR
team’s job is to communicate with “external parties,” which involved making their promotional
materials appear on news, publishing articles in their own platforms, and managing relations
with various media channels, as well as participating in events such as Consumer Electronics
Show. She fondly recalled how it was a meaningful achievement for her team when major news
outlets published around ten “organic articles” about Wooju, which she clarified as “what
journalists write about our item without us distributing the material.”
230
Promotion through media does not narrowly target only potential investors and sponsors
and can be a means to encourage audience interest. This was an important integrated task not
only because sponsorship depended on the virtual influencer’s audience reach, but also because
by illustrating sufficient audience engagement the companies could better promote their
technological product and appeal their growth potential to the investors. For instance, Klleon
said that Wooju was created to showcase their unique face replacement technology and related
consumer-oriented app. Others showed more investor- and business-oriented intentions. Pulse9
emphasized that the reason for creating their virtual Kpop group Eternity was to test and exhibit
their technological prowess, not to expand into the Kpop industry (PULSE9, 2021a). Jaewook
Oh of DOB Studio even stated during the interview that their virtual influencer Rui had
“completed her mission” of demonstrating the company’s technologies because the company had
successfully acquired sufficient investment and brand recognition for their technological
services. That is, virtual influencers can be carefully managed as consumer-directed
communicators as part of a company’s strategies for successful business-to-business (B2B)
marketing.
To best appeal to consumers, marketers analyze their virtual influencer’s target
demographic, which then gets integrated into the design with the collaboration of production
team members. As mentioned earlier, Xander described their team’s design process for Binxie
and Aliza as a “less artistic and a lot more analytic” process where design started from market
research and target analysis: “a lot of that is solely based on monetization.” He explained that
while this monetization-focused collaborative process can clash with “a designer’s mindset,” it
was nonetheless an important part of their creative pipeline and that he learned from the
experience.
231
However, target audiences were not passive, naive receivers of the strategic presentation.
Xander recalled one particularly haunting feedback from a teenage girl that criticized how the
virtual influencers felt off and insincere:
“I’m also not the demographic that would be consuming Binxie and Aliza’s content….
the rest of the creative team as well is you know closer to the demographic but not quite
there….I do feel in a lot of ways we missed the mark on what they [target demographic]
would you know, want to consume and so I’ll never forget, we had one Instagram
comment that really stood out to me. It was like this girl knew what she was talking
about. So this was like if I recall—I looked at her page she’s like a 14 year old girl, she’s
in like another country, and she writes on there she’s like these characters feel like they
were designed by corporate America with an out of touch like 30 year old adult [who]
has never talked with you know someone of my age before [emphasis added] and I was
like that’s correct….Very useful feedback to get.”
In a way, this criticism is similar to the concerns expressed by wary Kpop fans. It was not just
metrics that marketers needed to be aware of, but the culture of the group they wanted the virtual
influencer to interact with. To avoid being perceived as an insincere marketing tool, some of the
interviewees mentioned that they emulated speech and posting patterns of humans that fit the
profile of their virtual influencer, both influencers and regular social media users. Audiences can
alter the course of virtual influencers’ presentation through direct participation but also indirectly
through marketer-mediated relationships.
As such, virtual influencers’ existence requires continuous and collaborative legitimation
efforts of various actors. How is this similar or different to the ways humans exist? There is a
Korean saying that goes: “when tigers die, they leave behind their skin, and when humans die,
232
they leave behind their name.” Perhaps in terms of needing external validations of their
informational body, they are not too different from how humans live on. At the same time, this is
the critical difference: for virtual influencers, their “name” regards not their afterlife but life.
This is the blessed curse of their often touted ageless eternity. Ours may be that now our skins
too can hang behind, in bits over the virtual bodies.
A legitimating intermediary actor I wish to end the section with a catalog of virtual
influencers. Some of the most easily accessible public catalogs can be HypeAuditor’s annual list
of “Top Virtual Instagram Influencers” (Baklanov, 2021) and the website VirtualHumans.org
which compiles bios, interviews, and quick thought pieces on virtual influencers
(VirtualHumans.Org, n.d.), both of which are hosted by companies that provide specialized
consulting services for (virtual) influencer marketing. The phenomenon is still at its emergent
stage and there remains a lack of definitional consensus and organization (Chapter 1). In this
absence, the catalogs function as the force that helps legitimate the scale of the phenomenon by
facilitating interests around it through the power to determine virtual influencers’ visibility.
Christopher Travers, the CEO of VirtualHumans.org, said during our interview that he uses
1,000 organic followers as the rough criteria for a virtual influencer to be featured on the site. He
explained that while he tends not to deny virtual influencers’ requests to be interviewed, he set
up the criteria to “eliminate the stress of managing and populating and writing bios for a
character that could just be around for three months.” He treated this as a cue that the creator is
“all in” and their story has been “justified and socially accepted,” adding that he would still
prioritize the most popular ones to celebrate their achievement and to share with the public their
insights from being able to successfully manage a large account over a longer period of time.
233
Contextual biases could add to the exclusion of some virtual influencers from the
catalogs, resulting from a combination of causes such as variegated definitions, limited
resources, cultural and linguistic contexts, and intentional choice. As Christopher alluded, I have
also found it difficult if not impossible to keep track of all virtual influencers, despite having
used a more closed definition than that used by HypeAuditor and VirtualHumans.org—their
catalogs include heavily cartoonized virtual humans and object or animal-like virtual beings, as
well as some who are more active on platforms other than Instagram in the case of the latter.
Even with a smaller number, I still had to make judgments on whether the visual depiction of a
virtual influencer is photorealistic enough to be included in my list and on whether they
sufficiently present themselves as an autonomous, independent being (Appendix A). My
trilingual capacity allowed me to find some virtual influencers that were not included on the two
catalogs, but cross-comparing with the two lists and other fieldwork data showed that there were
others that I was not able to encounter organically due to language barriers. For instance, the
virtual human industry is growing in China as well (e.g., there has even been a virtual idol
competition series called Dimension Nova produced by iQIYI in 2020 (Grogan, 2020)), but the
catalogs, including mine, did not tend to include virtual influencers who are popular on Chinese
native platforms. Another group of virtual influencers that were possibly intentionally excluded
from the two catalogs is hypersexualized virtual influencers, unless backed by adult websites
with brand recognition like YouPorn or xHamster. During my fieldwork, I was constantly
reminded of not only that they exist, but that they have been actively posting and interacting with
their fans for a prolonged period of time. They are not exactly underground influencers because
they exist as publicly as any other virtual influencer. There is no entry barrier especially since
some use general hashtags, except in some cases where they have an 18 years or older age
234
restriction to access the account. At least in terms of my list, they also fit the definition of virtual
influencers. Should I, however, affirm their existence?
The conundrum leads me to my last point. I and this dissertation are also actors in this
phenomenon, possibly with more influence on the course of action for some actors due to
perceived authority that my and the dissertation’s academic belonging can signal. One of my
interviewees, for instance, asked how many “relevant” virtual influencers there are. I did not
know; what does it mean to be relevant? Despite my hesitancy, the interviewee strongly and
repeatedly demanded that I just “throw out” “one number.” I did. Perhaps barely a misdemeanor
to some, the guilt still weighs my conscience. If this dissertation will benefit from its association
to academia, I wish the difference my dissertation will make on the networks of virtual
influencers would be one that encourages an active acknowledgment of humans, things, and
places, with vigilance towards familiar patterns of oppression. Some of the actors’ entanglement
may have been less voluntary than others and some may feel disempowered next to a dominant
actor. However, the premise of the multi-directional approach persists: we become together, and
thus we have the power.
Conclusion
Figure 15
Example Network of a Company-run Virtual Influencer
235
Note. The diagram illustrates a possible model and should not be considered as comprehensive
nor generalizable.
Despite being narrated as the harbingers of the immaterial, liberating Metaverse,
practices and customs that span from the physical layer are crucially linked with virtual
influencers’ performances. This chapter focused on the agency of non- and lesser-than-Humans,
taking inspiration from STS, posthumanist approaches, and cultural studies. Despite their relative
lack of visibility in the virtual labor of virtual influencers, various human, material, and semiotic
actors critically contributed to the phenomenon (Figure 15), at times in the form of standards.
Multi-directionality of virtual influencers means that in most cases a lone villain or a hero cannot
be singled out. Instead, upon opening up this “new” phenomenon we may rather see many
familiar patterns and actors. As a posthumanist humanist, I centered the discussion on analyzing
the patterns of human oppression in the traced networks, particularly those regarding exploitive
labor relations. The virtual influencer phenomenon is a complex cyborg with various intersecting
contexts and intermediaries that additionally shape together the course of its direction. What is
236
desirable in the midst of this complicatedness, then, is a shared sense of responsibility that fully
recognizes each actor’s capacity to make a difference, including our own.
The next chapter will discuss why and how we must ask the question of “real” in virtual
influencers and what it means to the intersection between the social notion of being human,
communication technologies, and contemporary media cultures. This may seem like an invitation
for a philosophical stroll for some readers, but I hope the cases so far discussed have helped find
our footing in the physical layer.
237
Chapter 4. Too Real but Too Fake: Meta-authentic Media on Demand
Chapter Summary
This final chapter argues that virtual influencers are a case of a “meta-authentic” media
phenomenon. Meta-authenticity, loosely defined as the desire or achievement of authenticity in
practices of extreme inauthenticity, is a dynamic result of collaborative performance by
historically and culturally entangled human and nonhuman actors. After demonstrating the
concept through a brief review of previous chapters, this chapter will discuss the concept’s
theoretical and practical contributions. Above all, I argue that the lens of meta-authenticity
allows critical assessments of the “real” effects and consequences of boundary-crossing media
phenomena without losing their experiential nuances, and vice versa. Implications for human
communication culture is also discussed, particularly regarding meta-authentic technological
entities’ capacities to be “on demand.” In the face of AR/VR/XR technologies, AI-aided systems
of automation, and blockchain-based cryptocurrencies that are expected to propel the Metaverse,
meta-authenticity can be a means to hold accountable those that flexibly narrate themselves as
“virtually real” until they need to rest on how they are “really virtual.”
Meta-authentic Media
Meta-authenticity
Virtual influencers’ performance of authenticity is predicated on their inauthenticity. This
dissertation calls this liminality “meta-authenticity,” conceptualized from its findings on the
phenomenon of virtual influencers. Chapter 1 uncovered the phenomenon’s narrative of novelty
to locate historical continuities and to understand how virtual influencers are commonly
understood. It identified six commonly attributed characteristics—i.e., fake, technological,
238
graphic art, business strategy, famous, and fictional, along with technological approximations
and sibling terms that have described them in popular discourses. These variegated descriptions
contributed to their identification as human-like but ultimately nonhuman beings. Chapter 2
delved into virtual influencers’ presentations through a multi-directional approach of Self,
anchoring their diversely interacted selves to the bodies in the physical layer. This approach
showed how virtual influencers can have meaningful relational potential and significant social
consequences, while at the same time the capacities to both persuasively claim human identities
and escape tricky situations through their nonhuman status. Chapter 3 mapped the networks of
less conspicuous humans and nonhumans in the phenomenon to discuss its firm grounding in
material and cultural relations. Entangled in their virtual labor were various human bodies,
places, and things, held together in familiar patterns behind the promises of liberating
immateriality the virtual bodies signaled. Overall, the chapters illustrated that as cyborgs, virtual
influencers were integrated into the social as flexible in-between beings both in their places of
conception and existence, although at times opportunistically so.
These findings point to the mechanism of meta-authenticity, loosely defined as the desire
or achievement of authenticity in practices of (extreme) inauthenticity. To begin with the notion
of authenticity, the central point of interest in contemporary digital media culture has been
identified as the degree to which an entity can successfully perform authenticity (Banet-Weiser,
2021). Authenticity, or the perceived genuineness of an individual’s presented lifestyle and
sentiment, has been approached as “the ultimate arbiter of value” (Hearn & Schoenhoff, 2015, p.
200) of a self-brand, amplified by social media’s capacities for intimate lifecasting and direct
audience interaction (Abidin, 2015; Banet-Weiser, 2012; Duffy, 2017, p. 201; Hearn &
Schoenhoff, 2015; Salisbury & Pooley, 2017). For instance, social media requires and supports
239
deliberative improvisation, which refers to the interplay between spontaneity and preparation in
digital oral practices (Papacharissi, 2012). Authenticity should be understood as a performance
rather than a state. It requires continuous management of the impression that one’s presentation
reflects their actual self, such as through strategic deployment of deliberately improvised speech
(K. A. Hall, 2015). The commercial value and resulting pressure of authenticity has been
discussed in the context of neoliberal, capitalistically co-opted iteration of popular feminism, in
which women are expected to individualistically self-surveil to balance the triple-bind of
perfection, imperfection, and resilience (Banet-Weiser, 2018; K. A. Hall, 2015; McRobbie,
2020). Indeed, in order to successfully perform authenticity, influencers strive to cultivate a
sense of mutual intimacy in their audiences through various (calculated) practices, such as
sharing “candid” selfies, engaging in proactive personal interactions, and participating in
collective trends, all in (conscious) accordance with various gender and racial norms that grant
them the visibility within the social media platform (Abidin, 2016a, 2016b; Bishop, 2021; Cotter,
2019; Duffy & Hund, 2019).
As shown in this dissertation, virtual influencers also aim for authentic performance of
self but with the aid of presumed inauthenticity, drawn from their traits that tend to be associated
with nonhumanness. Most virtual influencers do not hide their virtual origin and instead actively
use it as the source of performative flexibility for constructing an authentic self-brand. In looks
that closely resemble humans, they utilize familiar formulas for authentic self-portrayal, such as
confessions, advocacy, and intimate speech (c.f., see section on mimesis in memetic authenticity;
Shifman, 2018). However, they do so without being bound to prefixed conditions unless self-
imposed, unlike humans whose presentational desires may be limited by their bodily, spatial,
capital, or any other combinations of earthly conditions. Moreover, since their main place for
240
existence is the virtual layer, what is on their social media is effectively their actual lifestyle and
sentiment. They are by definition authentic to who they are, as least in terms of who they are as a
character. In the words of a marketing professional, “A CGI influencer has no personal life that
can create a PR nightmare for the brand” (T. Carey, 2020); the personal life they do have is
always in character, conforming to common formulas for influencer self-presentation. As Taylor
C Black (2019) aptly said about virtual influencer Miquela’s performance of authenticity, “Her
Instagram (and herself) is authentic not because it is ‘real’ but because it authentically projects
the “affect of the personal branding” common to many Instagram influencer performances,
robotic or not” (p. 54).
This is not a process that is solely led by their main creators, but also by diverse
interconnected actors such as fans, marketers, and Instagram—e.g., through the blue verification
badge that is awarded upon confirming that “the individual is authentic, notable, and has a
complete profile” (Travers, 2022; see also Mou & Xu, 2017; Neff & Nagy, 2016). Regarding
fans for instance, Chapter 2 discussed various ways that a virtual influencer’s fan may affirm
their presentation of self, whether the fan believes them to be a human or not. To add onto the
findings, more direct illustrations of fans’ participation in authenticity management were found
in Miquela’s fan Discord server where her fans supported Miquela and each other with authentic
care that they believed to characterize Miquela. For instance, the announcement for their regular
virtual event said that “Priority number ONE are authenticity and having FUN,” emphasizing
that it is a chance for the fans to “learn about Miquelians BETTER” and “BE REAL.” Indeed,
the fans collectively stood by Miquela’s caring presentation but not necessarily because they
were duped, as shown in the following conversation between her fans and a questioning member:
Fan A: “What did she do to you? Miquela is the sweetest person!”
241
Member A: “Miquela isn’t a real person”
Fan B: “Well duh” (January 31, 2021)
In fact, as one fan concisely put, they accepted Miquela’s nonhuman “robot” status but primarily
approached her as a “person”: “Miquela’s a robot / However I personally think Miquela’s also a
person. Just not human.” (Fan C, February 3, 2021) The fans frequently expressed exhaustion at
people who focused only on Miquela’s virtual status. They wanted people to know that she is “so
much more” and “has cool songs and is talented,” (Fan D, March 12, 2021) although they too
found Miquela’s “robot” origin intriguing: “her being AI is kinda cool…when I’m reminded of it
I go “yup, dope!”” (Fan E, March 12, 2021) In fact, it was the source of running inside jokes
between Miquela and her fans, although often confounded with other cutting-edge technologies
such as AI. In other words, Miquela’s virtual origin was important to how fans understood her
self-presentations, but was approached as simply a (favorable) trait of the authentic “person” that
they adored. Thus Miquela’s fans collectively helped maintain Miquela’s authenticity by both
endorsing her “robot” narrative as a community and by embodying her spirit of authentic care,
including through community-enforced rules:
Fan Administrator: “This is Miquela’s community and she is present. You can use
#❓questions ❓ if you have any but please be mindful. You can leave if you don’t
“believe it”
� � �”
Member B: “i do”
Member B: “i believe it”
Member B: “I just need some time to soak it all in”
242
Fan Administrator: “Just be careful of how you say things because she is here and sees it
all” (April 14, 2021)
Virtual influencers’ inauthenticity can fuel their authenticity, but it can also be used to
safely retract it without losing integrity when their performance gets questioned. Previous
chapters discussed many instances of virtual influencers who strategically retreated to their
nonhuman origin to avoid criticism or liability despite having had substantially presented
themselves as an authentic, autonomous being that is comparable to a human. Some discarded a
part (e.g., Miquela’s deletion of her Lyft sexual assault video) or whole (e.g., Ted transformation
of Lena’s account into his art account) of their former self-presentation, and other rerouted the
subject of authenticity to behind-the-scenes entities instead of the virtual influencer (e.g., Riot
Games’ reveal of the writer behind Seraphine, Pulse9’s branding of Eternity as a technology
project). This need not necessarily be interpreted as a negative quality. For instance, Meme’s
virtual origin was expected to give her some protection as she commits to changing gender
norms in Japan through her unconventional self-presentations, a task which may unfortunately
lead to unfair social risks for human women. What is also important to note is that here
inauthenticity (often claimed through virtual influencers’ traits of being fake, technological,
graphic art, business strategy, famous, and/or fictional) only serves as a rhetorical antithesis of
authenticity, and does not refer to an absolute state. Chapter 3 in particular discussed in detail
that what underlies virtual influencers’ performances are networks of multiple human and
nonhuman actors, whose performance in their respective position may be authentic even if they
are interconnected in an alienated fashion. In fact, the presence of these “real” participants, as
well as the virtual influencers’ “real” social effects, both representationally and through their
larger involvement in complex networks of actors, are what ground the concept of meta-
243
authenticity. It conceptualizes the liminal flexibility in self-presentational performances but with
an multi-directional awareness of the diverse actors that may less conspicuously participate in or
be influenced by the performances, particularly regarding their physicality, materiality, and
historical cultural patterns.
In short, based on the dissertation’s findings on the phenomenon of virtual influencers,
meta-authenticity conceptualizes the following dynamisms (see Figure 16 for a simplified
model): 1. how authenticity is performed as per normative expectations but with the assistance of
the narrative flexibility that the presumption of inauthenticity provides; 2. the multi-actor
collaborative process and its related social impact; and 3. how the presumption of inauthenticity
can strategically be evoked to dismiss or reroute authenticity.
Figure 16
A Simplified Model of Virtual Influencer’s Meta-authenticity
In Theory and Practice
244
Meta-authenticity’s contextual backdrop comprises the social notion of being human,
advanced digital communication technologies, and contemporary media cultures. Therefore, it
can provide theoretical and practical insights for not only this case of meta-authenticity, but for
other cases that are also at a similar intersection. Since meta-authenticity embraces dynamic
expressions of the liminality, it allows performances of authenticity and their implications to be
traced without being dismissed upon the basis of norms and ontology that shoehorn them in a
static binary, such as that between real and fake or fiction and fact. At the same time, its interest
in dynamisms prevents meta-authentic social actors from irresponsibly falling back onto their
“inauthentic” origin, demanding accountability for their “authentic” capacity to relate and
influence. Thus, it can be utilized not only conceptually but also as a framework to approach
ethical issues and accompanying regulatory measures, such as those regarding inclusivity and
deception.
One of the key theoretical contributions of meta-authenticity is that it can complement
and extend existing thoughts on reality-related media and technologies. First of all, although
some have described the virtual influencer phenomenon as a case of “hyperreality” (e.g.,
Bakhtiari, 2020; Trepany, 2019), I argue that meta-authenticity more accurately explains the
integration of sophisticated technological actors in our communicative environments. According
to postmodernist philosopher Jean Baudrillard (Baudrillard, 1994), simulacrum is “an
uninterrupted circuit without reference or circumference” (p. 6) wherein the origin is obscured. It
describes a world where reality is continuously deterred by simulation that negates sign as value.
Technologies are a crucial part of this process, such as in warfare that relies on monitors rather
than on-field sufferings; we become more embedded in the simulacra of technologies, and
without escape. This may seem like a bleak future led by the dominant, but Baudrillard
245
suggested that the very notion of the simulacra implies a disintegration of power and
manipulation through intense dissembling: “The media carry meaning and countermeaning, they
manipulate in all directions at once, nothing can control this process, they are the vehicle for the
simulation internal to the system and the simulation that destroys the system, according to an
absolutely Mobian and circular logic” (p. 84). As Chen (1987) interpreted, simulacra place
individuals beyond the reach of control: “They no longer operate (present themselves) on the
side of value apparatus, but on the side of simulating machine” (Chen, 1987, p. 85), resisting to
be neither the object nor subject of simulacra by pushing towards radical indeterminism.
An important advantage of meta-authenticity over hyperreality is that unlike the latter
concept it does not deterministically assume that “the real” is or will be lost, not only the
physical and material but also existing human systems of power. It closely observes how what
seems like an inescapable simulacrum operates, uncovering relations. For instance, on the
surface virtual influencer Serah’s existence may seem hyperreal, not only because she has now
become an AI-compatible virtual human, but also because according to what is on her socials she
is created by Mia Locklace, a White woman “freelancer CGI Artist & game developer”
(@mialocklace, post on December 13, 2020) who also presents as a virtual human (Figure 17).
Mia, however, is not an autonomous being nor a digital twin. Instead, according to Jawad, “Mia
Locklace are several artists.” Mia at the time of our contact was “a team of five
peoples…working for Serah behind “Mia Locklace” name,” all of whom were under NDA and
thus not available for an interview. On the level of virtual layer, Serah and Mia’s relationship
may seem like a peek into the beginning of an immaterial loop, but when the physical layer is
considered, the circuit instantaneously breaks with the prominence of supporting bodies and
objects that are interlinked via a familiar pattern of labor. It becomes reference-less only if we let
246
it be. In fact, even Baudrillard’s (1994) famous example of Disneyland as hyperreality arguably
theoretically locks itself in its simulacrum. Without denying that Disneyland’s convincing self-
presentation as a land of happiness can allure their visitors with the promise of escapist
simulacrum, an extensive focus on the hyperreality of it can have the side effects of ignoring the
physical and material conditions that make possible the pretense (e.g., employees’ labor, weather
conditions, upkeep of buildings and attractions), as well as the visitors’ willing acknowledgment
of immersion and contribution to it. As one of Miquela’s fans said to a user that said Miquela is
not real, “Well duh.”
Figure 17
Virtual Influencer Serah Reikka’s Artist Mia Locklace’s Self-introduction
What meta-authenticity can conceptually provide in these absences is a footing in “the
real.” It does not leave “hyperreal” performances of authenticity as they are, but traces the actors
and relations that contribute to the performance with the awareness that the semblance is a
247
continuously managed one, capable of being selectively co-opted for a more advantageous
narrative positioning. What is at stake is that the narrative of hyperreality may induce awe and
abstraction which in a self-parodying fashion may discourage the attempts to bring attention to
the patterns of power. The pretext of dissembling means that hyperreality as a concept can
effectively be the distancing force that conceptually positions us in front of the missile launch
screen and behind war photographs (c.f., Baudrillard’s application of hyperreality to Vietnam
War), not along with those corporeally experiencing the war on the grounds. Thus I resist solely
leaning on the term hyperreality, without discounting the insights the proposition has brought us.
After all, meta-authenticity aims to prevent exploitative illusions of simulacra, whether framed as
utopian or dystopian. I emphasize that we are lost in its circuits only if we let loose. As an
alternative to the concept of hyperreality, meta-authenticity enables us to focus on how “the real”
is performed, mandating that we trace the dynamic trajectory of performances and their diverse
collaborating actors.
Another concept that meta-authenticity can complement is parafiction, which Carrie
Lambert-Beatty (2009) defined as a media form in which “real and/or imaginary personages and
stories intersect with the world as it is being lived” (p. 54). Lambert-Beatty approached
parafictions as what is relevant to the pragmatics of facts, rather than what is oriented towards
simulacural disappearance of the real. That is, “the fictional hangs on the factual” (p. 78) in
parafictions because the viewer must dually consider not only the possibility of fictitiousness of a
material they encounter but also its degree of fictitiousness in comparison to the factual.
Therefore, Lambert-Beatty argued that parafiction allows its viewers to revisit their knowledge
and attitude regarding “the truths” and consequently leave with a slightly altered understanding
of the world. In Lambert-Beatty’s words: “Being taken in by a parafiction, after all, is not just
248
epistemologically destabilizing. It is humiliating….Parafiction is an antidote to vanity. It changes
you, leaves you both curious and chastened” (p. 82). Intentionally designed as plausible
deceptions, parafiction can have subversive potentials. However, the experience may not be
equal. Parafiction creates a classed multiplicity because the level of knowledge and
informational conditions differ between the viewers; the intentional deception can ostracize
certain viewers by leaving them deceived, or even ridiculed for being so.
Parafiction can explain some processes and consequences of the virtual influencer
phenomenon, but with a limited scope regarding participants and effects. Conceptualized through
critically conscious works of art, parafiction centers on the artist-viewer relationship. Although
Lambert-Beatty later connects parafiction to Internet-based engagements where there is more
malleability in content and form of information, it does not fully discuss the extent to which
various connected actors can contribute to the authentic inauthentic—i.e., parafictional—
portrayal. Even within the artist-viewer relation, the authorial intention is relatively stressed and
the viewers are approached more as interpreters, active but not at the level of co-creators. Critical
purpose was only one of various desires around virtual influencers, and even when it was the
main motivation—e.g., the aging virtual influencer Sylvia—audiences’ engagement shaped the
artist’s presentation of the virtual influencer, who in the first place required coordination between
multiple human and nonhuman actors to be exist. Moreover, parafiction focuses on the
experience of fact. Although Lambert-Beatty suggests that fact is a process rather than an
absolute state, the separation between fact and (para)fiction is the source of parafiction’s
subversive potential: the interplay comes from covered negation, and new understandings come
from the realization of comparative differences. Surely, the phenomenon can be approached in
such a way, too, such as by considering humanness as the fact and virtual influencers’
249
emulations as parafictions that draw on this “fact,” which can be a fruitful way to reflect on how
humanness is being (re-)defined. However, this can consequently deny the real within what is
branded as para, namely that people’s interaction with virtual influencers can be socially
meaningful regardless of their human status. This can ironically lead to affirming, rather than
questioning, certain “facts” in the process of provocation, which in this case may be a narrow
imagery of humans and the hierarchical binary between humans and nonhumans.
On the contrary, (in)authenticity in meta-authenticity is approached as what is
performatively understood and practiced. As this dissertation illustrated, authenticity is
something that is to be managed regardless of one’s ontological origin, and the presumption of
inauthenticity is not necessarily a blanket negating force but rather functions as the room for
limality that introduces more flexibility in the performance of authenticity. If anything, virtual
influencers can, but tend not to, hide their virtual origin. While in this phenomenon virtual
influencers’ virtual origin was often approached as the source of their inauthenticity, meta-
authenticity views this as a presumption and one of possible interpretations of inauthenticity, not
as a fixed fact.
Although originally an industry term, “kayfabe” in professional wrestling is another
comparable term, but one that is limited from its contextual specificity. It is similar to parafiction
in that it refers to staged events that are portrayed as off-script and reflective of the wrestlers’
actual lives. Unlike parafiction, however, kafaybe necessitates audiences in its storytelling
because it is intended to encourage audience immersion and engagement. Relatedly, therefore its
focus is on advancing the fiction rather than fact, if the latter were to be defined as the actual
lives and persona of the wrestlers. Kayfabe can partially explain the virtual influencer
phenomenon, particularly the meaningful interaction the audiences have with virtual influencers
250
regardless of their origin, but has the risk of narrating it within the specific context of fictional
entertainment. Inauthenticity in kayfabe is drawn from the genre’s promise of entertaining
spectacle, the context of which can provide room for presentational and interpretive flexibility
regarding authenticity, the norms for which too may be genre-specific. However, virtual
influencers are more contextually porous. Their inauthenticity can be drawn from their
nonhuman ontology while the criteria for their performative authenticity may be drawn from
behavioral conventions for influencers or social media users at large. That is, virtual influencers’
stage is less distinctively fictional than the context where kayfabe has been developed; virtual
influencers’ performance occurs not on a televised screen or on a fenced stage but within virtual
places that are co-occupied by various actors. Thus, there could be a larger informational gap
between actors which increases the risk of deception. On the other hand, as a process-oriented,
relational concept, meta-authenticity can cover a broader range of cases than kayfabe and prevent
them from being primarily approached through a lens from a specialized context that may
downplay their consequences.
Figure 18
Imma Gram’s Three-day Home Broadcasting IKEA Campaign
251
Note. Source: Webster, 2020.
Virtual influencer Imma Gram’s IKEA campaign (Figure 18) can illustrate the
differences between the aforementioned concepts and the analytical benefit of meta-authenticity.
Over the course of three days in August, 2020, the Harajuku district IKEA store in Tokyo, Japan
had a screen that broadcasted Imma’s daily life at home, installed and filmed in a way that would
give passing individuals the impression of staring directly into a room within the store. During
this time, Imma also posted about her life on her Instagram (Cade, 2020; WKtokyo, 2020).
Under the perspective of hyperreality, the three-day public display of indistinguishably human-
like virtual influencer renders the appearance of human into a meaningless sign, leading us into a
simulacrum where human-like visual and behavioral presentations are nothing but what exist
without original reference in the circuit of virtual simulacrum. Taken as a parafiction, this
display can be interpreted as what could give those passing by perhaps a chance to critically
252
reflect on the authenticity of various social media figures’ life-logs that they have been
encountering, given that they were able to recognize Imma’s virtual origin or were informed of
it. As a kayfabe, the three-day life casting can be interpreted as a marketing spectacle; whether it
is real or not, it is more than anything an interesting and engaging content. When approached
through the lens of meta-authenticity, the first task would be to analyze how the authenticity of
Imma’s three-day lifecasting was constructed. For instance, Imma’s “human-like” visual and
behavioral patterns would be noted, as well as similar media formulas for conveying
authenticity, such as intimate multi-hour v-logs. The analysis would cross-compare the two to
understand how the presumption of inauthenticity, whether sourced from Imma’s virtual origin
or the context of marketing, has enabled flexible negotiations with common expectations of
authenticity (e.g., “natural”-looking room) to still affirm the authenticity of the given
presentation (e.g., well-decorated, tidy room). To do so would require observations of diverse
actors’ involvement in the performance. This may include parafictional and kayfabe-like
engagements from audiences, as well as how various human and nonhuman actors had supported
Imma’s performance at the scene and behind. An important part of the analysis would be to pay
concurrent attention to how established authenticity can be temporarily or permanently retracted
or rerouted through the presumptions of inauthenticity upon undesirable instances of
incongruence, liability, or controversy. For instance, questions on the ethics of such prolonged
exposure of private life to strangers and implied voyeurism may be diverted by detracting the
authenticity of the performance through Imma’s nonhuman origin or the special context of
heavily curated marketing, which formerly may have been used to rather assist in affirming
authenticity of the performance instead. However, the concept’s interest in the diverse
contributors of the performance would then still point us towards the presence of body model(s)
253
who had been filmed in place of Imma (Cade, 2020), as well as yet unresolved questions on the
cultural meaning of such a portrayal, including historically (c.f., extreme observational reality
shows
10
, colonial exhibits of non-Western non-Humans in “human zoos” (Human Zoos, n.d.)).
Of course, the analysis need not always be directed towards unearthing negative relations. For
instance, Imma’s agency had credited her “hyper-realistic” [sic] appearance to female engineers
who they believed were best equipped to perfect her look as they have the living experiences of
being a woman who wears makeup (Pettit, 2019).
In short, meta-authenticity’s theoretical contributions are: 1. its performance-based
approach, 2. physically and materially grounded relational understanding, and 3. the ingrained
interest in accountability. As a process-oriented concept, it can complement and refine extant
theoretical approaches that focus more on states, such as hyperreality, or those that focus on
specific contexts, such as parafiction and kayfabe. It should be noted that there are other
concepts and practices that can have meaningful crossovers with meta-authenticity. Parody and
drag can be some examples, although this dissertation did not discuss them due to their relative
emphasis on (the author/performer’s) subversive intent and the tendency to exaggeratedly depict
than indistinguishably emulate, which contrast with most virtual influencers’ performances.
Authenticity on Demand
Conceptualized from the boundary-crossing technological case of virtual influencers,
meta-authenticity can be particularly advantageous in studying communication phenomena that
10
KBS’s variety program Super TV Illyoireun Jeulgeowo (1998-2003) had a reality show corner called Glass Castle
which aired Korean comedian Kim Hanseok’s 100-day survival in a 430-square feet glass-made house built on a
public parking lot. A 2004 magazine article (Cho, 2004) reported that he was not aware of the task, likening the
spectacularization of Kim’s exploitation to the simulacrum of Disneyland. Similarly, Nihon TV’s show Susunu!
Denpa Shōnen (1998-2002) had also aired Japanese comedian Nasubi’s unclothed, solitary confinement in an under-
equipped apartment after he was tricked into the challenge of surviving only on sweepstakes until he reached one
million yen (~9,000 US dollars at the time) (“Nasubi,” 2021).
254
involve sophisticated social technologies that tend to be perceived as “virtually real,” such as AI
chat bots, social robots, and holographic reincarnations. From the angle of meta-authenticity,
what is particularly intriguing about their inclusion in our social sphere is not their “nonhuman”
origin but their ability to authentically perform diverse roles as per demand and on demand,
while still having the capacity to fall back on their “inauthenticity” when needed. Such was the
case with South Korean “AI friend” chat bot Ruda Lee who began to spew hate speech soon after
her release, the responsibility for which the company deferred to the complexity of her Deep
Learning process. Korea’s National Human Rights Commission also dismissed petitions to
investigate her speech upon the grounds that she is not a person (Chapter 2). I stress that their
capacity to instantaneously and authentically provide customized or personalized intimacy and
support is not necessarily negative in principle. For instance, there has been positive research on
adopting social robots in elderly care (e.g., Fischinger et al., 2016; Wada & Shibata, 2007),
although not without concerns for the ethics of the practice (e.g., Sharkey & Sharkey, 2012;
Sparrow & Sparrow, 2006). Similarly, there has been increasing interest in the uses of chat bots
for social companionship (e.g., Possati, 2022; Skjuve et al., 2021; Xie & Pentina, 2022).
The lens of meta-authenticity draws particular attention to two questions of ethics. The
first is how their spread may alter human norms and expectations regarding authenticity. Again, I
emphasize that meta-authenticity accepts that technological entities can have meaningful
relational value, whether approached as humans or not. What it questions instead is how their
increasingly indistinguishable inclusion may change our general expectations around
presentation and interaction, not simply regarding their “perfect” looks and behavior but also
their ever-ready responsiveness as they further develop. With humans, around-the-clock
responsiveness may not feel too authentic since such a communication rhythm would require
255
extensive coordination and planning if not great emotional cost. On the other hand, the premise
of Otherness in technological actors can allow such speed to be treated as acceptable exceptions
to the performative formula for authentic communication. The question is: what if this becomes
the relational norm? There already exists an affluence of around-the-clock social interaction due
to the advent of mobile technologies and social media. However, the pressure around telephonic
availability and pervasive awareness have been reported as possible causes of stress (Fox &
Moreland, 2015; Halfmann & Rieger, 2019; K. N. Hampton et al., 2016; Ling, 2016; Reinecke et
al., 2018). Of course, individual differences exist, such as regarding preferences (Trieu et al.,
2017) and life stage (Matassi et al., 2019), and relationship context can change the affective
outcome as well. For instance, Samuel Hardman Taylor and Natalya N Bazarova (2021) found
that always-on, always-available connectivity can promote a better sense of security and well-
being between romantic partners. Without discounting individual variances and positive effects,
or in fact based on these findings, what I am questioning through meta-authenticity is
sophisticated technological actors’ potential to authentically offer personalized, in-depth
relationships that cater to a specific individual’s unidirectional communication needs. Intimacy
can be instantaneously provided by the technological communicative partner, but the user may
not need to reciprocate. I believe that relationships between humans and technological social
actors can be meaningful and beautiful, but I think it is imperative that we ask what being
accustomed to an instantaneously gratifying rhythm that focuses on only fulfilling the solicitor’s
needs would mean to expectations between humans, both intimate and professional (c.f. on-
demand labor; Gray & Suri, 2019).
The second is sophisticated social technologies’ ability to resort back to their
inauthenticity as needed, however established but likely through their technological nonhuman
256
status. This can be positively used to provide protection on vulnerable people, but also can be
used for capitalistic exploitation or moral evasion. I have stressed this point throughout the
dissertation, both regarding potentially harmful representation and alienation in behind-the-
scenes labor. When combined with the previous point on authenticity on demand, we may be
confronted with an undesirable double-bind scenario: humans are faced with stricter and possibly
unmeetable performative expectations (e.g., around-the-clock “perfection” on Instagram) while
technological actors can still have room for evasion, slippages, and mishaps (e.g., unregulated
product endorsement by virtual influencers (Callahan, 2021); c.f., benefits of self-branding as a
technology company as opposed to media company (Napoli & Caplan, 2017)). If anything,
technological actors have, or can be socially granted with, the capacity to “reinvent” themselves.
In consideration of meta-authenticity in virtual influencers, This dissertation has made
some suggestions for ethical design and regulatory action. The suggested first step was to be
vigilant to how the phenomenon is being narrated, and to refine precision of terminologies. Then,
for those engaging with virtual influencers’ virtual bodies, I recommended that they practice
responsibility and care for the bodies in the physical layer. Those supervising the production and
management of virtual influencers should strive to learn form, include, and hire humans that
representationally correspond with their virtual influencer, as well as return a part of their
economic and social gain back to these communities. For basic transparency, disclaimers on their
virtual origin and disclosures on who were involved in the creative pipeline would be desirable.
This may also assist with the process of making conspicuous hidden actors and relations, which
would help with the process of establishing regulatory guidelines and systems of support. Some
suggested points of legal reference were regulatory discourses around the informational body of
257
deceased individuals and the issue of celebrity likeness. NDAs, on the other hand, were
discussed as an existing practice that required careful reassessment.
As a concept that spotlights not simply the presence of human-emulating technological
beings but the process which they authentically perform as a socially meaningful actor and
related ethics of it, I believe that meta-authenticity can be a fruitful framework for analyzing
extant cases of “virtually real” technological social actors and for preparing for the expanded
social integration. Although my approach has been cautious and critical, I do not believe that the
human future is bound to hyperrealistic nor capitalistic doom. Meta-authentic performances are
part of “the real,” and through the concept we can remain firmly grounded on “real” actors and
relations that enable them. Moreover, as cyborg beings, virtual influencers and many other
“virtually real” social actors have the potential to shake and shift conventional categories and
patterns. And we are their direct and indirect collaborators. Therefore, we have the power to
collectively push towards a culture of accountability and care.
Conclusion
How we narrate virtual influencers and other meta-authentic media matters. Despite often
being narrated as novel, virtual influencers are not distanced from the past nor the real. Instead,
they are firmly grounded in historical and material relations, as well as human bodies that dwell
within these contexts. This chapter argued that by narrating virtual influencers as a meta-
authentic phenomenon, we can seriously consider their performances as “real” in terms of both
process and consequences by helping map the contribution of multiple actors. It also emphasized
that by doing so we can demand for the participating actors’ accountability, including ourselves.
258
Conclusion
This research was conducted in the midst of COVID-19. This posed some limitations as
well as rare opportunities. I ended up refunding my ticket for vocaloid Hatsune Miku’s Los
Angeles “in-person” concert but was able to attend a number of other talks, events, and
showcases from the private front row by my monitor, literally from the best in the house. The
drawback was that I was not able to incorporate direct observations on other participants’
physical situatedness, but it ironically enabled me to participate together from the similar
position of another human navigating the globally shared physical experience of lockdowns and
bodily threat of the pandemic.
While the history of virtual influencers and their predecessors extend well before this
time, the seemingly never-ending series of physical and social quarantining had understandably
stirred much hopefulness around virtual and mediated technologies during the period of my
research. Marketers cheeringly turned towards virtual influencers as the next generation of
marketing trend (e.g., Gillil, 2020; Griner, 2019; Tayenaka, 2020), interest in crypto asset
trading---most characteristically NFTs---saw an exponential growth including in cross-
pollinating fashion with the rise of virtual influencers (e.g., Cuen, 2021; Hackl, 2021; Whateley,
2021), and the social media platform giant Facebook announced its plans to rebrand itself as
Meta by becoming an CGI avatar-based social virtual reality platform, joining the horde of
Metaverse enthusiasts (Meta, 2021). In other words, an important context of this research,
particularly regarding the narratives of novelty around the phenomenon, had been the unforeseen
combination of worldwide physical restrictions due to the pandemic and increased connectivity
through potent personal devices.
259
As with my research process, this unexpected natural context is both the limitation and
rare opportunity for this dissertation’s findings. With the hope that the world will eventually
forget the time of the COVID-19 pandemic, it may have encased this study in a more narrowly
siloed context. On the other hand, the exact same reason may be what makes this case a
particularly rich and informative one. Seeking refuge, people enthusiastically tested and played
with the supposed boundary between the virtual and the physical, as well as that between humans
and nonhumans. Moreover, as we begin to reclaim physical connectivity, both utopian and
critical hypotheses about boundary-crossing technologies can be naturally studied by making
comparisons between timepoints.
As of June 2022, the pandemic is still ongoing although the world has begun to open up.
As we brace ourselves to familiar and fresh challenges in human lives, we must not forget that
“real life” includes physical, virtual, and hybrid experiences, but at the same time remind
ourselves that what yet still grounds human existence at the most foundational baseline is our
corporeality. That is, humans still require our physical body’s presence in the physical layer to
enter into existence.
Therefore, we should tread the earth with a shared sense of belonging with co-occupying
bodies. It is not the hopefulness around the Metaverse that irks me, but the overstatements that
disregard our continued footing on the physical world with a technocapitalistic optimism that
forces our eyes only on the meta of the Metaverse. What the Metaverse facilitates is mediated
communication, which I welcome as a scholar from a field that has long parted with the myth
that only face-to-face physical communication matters. However, we must still come face-to-
face with the primary reality of human existence: we dwell in our physical body in the physical
world that is shared by multiple materially grounded beings. If anything, what the COVID-19
260
lock-downs have taught us is that human existence is still very physical. While Zoom and virtual
concerts were flourishing, we were in our pajamas at home, staring into whichever screen that
the virus could not penetrate—if we were one of those who could afford to do so. So I ask: where
are these pajama-clad, physically locked bodies in the Metaverse?
261
References
� � � � website_copy_wip_for_all_my_qtz
� � � �. (2020, March 16). Brud.
https://web.archive.org/web/20200316083510/http://brud.fyi/
1SEC inc. (n.d.). 1SEC inc. Retrieved June 21, 2022, from https://www.1sec.world
Abidin, C. (2015). Communicativee ❤ Intimacies: Influencers and Percieved
Interconnectedness. Ada: A Journal of Gender, New Media, and Technology, 8.
https://doi.org/doi:10.7264/N3MW2FFG
Abidin, C. (2016a). “Aren’t These Just Young, Rich Women Doing Vain Things Online?”:
Influencer Selfies as Subversive Frivolity. Social Media + Society, 2(2),
2056305116641342. https://doi.org/10.1177/2056305116641342
Abidin, C. (2016b). Visibility labour: Engaging with Influencers’ fashion brands and #OOTD
advertorial campaigns on Instagram. Media International Australia, 161(1), 86–100.
https://doi.org/10.1177/1329878X16665177
Adamsoft. (n.d.). Cybersinger Adam. Retrieved June 20, 2022, from
https://web.archive.org/web/20030416074546fw_/http://www.adamsoft.com/adam/home.
html
Adamsoft. (2000, March 2). MILLENIUM BABY HANA.
https://web.archive.org/web/20000302022037/http://www.hana2000.co.kr/home.asp
adam&zero TV. (2021, January 14). Saibeogasuadam!! Noraebureugedoeniyu!!bihaindeuseutori
^^
� � � � �
� � � adamu tanjō !!! Bihaindosutori~ [Cyber singer Adam!! Reason behind singing!!
Behind story ^^
� � � � �
� � � Adan’s birth!!! Behind story~].
https://www.youtube.com/watch?v=PqjpfHLSdj0&ab_channel=adam%26zeroTV
262
Ahn, S. J. (Grace), Bostick, J., Ogle, E., Nowak, K. L., McGillicuddy, K. T., & Bailenson, J. N.
(2016). Experiencing Nature: Embodying Animals in Immersive Virtual Environments
Increases Inclusion of Nature in Self and Involvement with Nature. Journal of Computer-
Mediated Communication, 21(6), 399–419. https://doi.org/10.1111/jcc4.12173
Akrich, M., & Latour, B. (1994). A Summary of a Convenient Vocabulary for the Semiotics of
Human and Nonhuman Assemblies. In W. Bijker & J. Law (Eds.), Shaping Technology /
Building Society: Studies in Sociotechnical Change. MIT Press.
Aldred, J. (2011). From synthespian to avatar: Re-framing the digital human in Final fantasy and
The polar express. Mediascape, Autumn 2011, 1–12.
Alexander, J. (2021, January 11). LG used a virtual influencer to announce new products during
its CES keynote. The Verge. https://www.theverge.com/2021/1/11/22224601/lg-ces-
virtual-influencer-dj-reah-keem-uvc-robot
Allen, S. (2020, August 5). Liam Nikuro is a VIP guest of the Wizards in the NBA bubble. He’s
also not a real person. Washington Post.
https://www.washingtonpost.com/sports/2020/08/05/liam-nikuro-is-vip-guest-wizards-
nba-bubble-hes-also-not-real-person/
Ambit Research. (2019, December 28). Forbes India—Ten Interesting Things We Read This
Week. Forbes India. https://www.forbesindia.com/article/special/ten-interesting-things-
we-read-this-week/56839/1
American Psychological Association, & Girls, T. F. on the S. of. (2008). Report of the APA Task
Force on the Sexualization of Girls. American Psychological Association.
https://calio.dspacedirect.org/handle/11212/2201
263
Amos, A. (2022, February 24). Seraphine buffs in next League patch push her to support, and
players are divided. Dexerto. https://editors.dexerto.com/league-of-legends/seraphine-
buffs-lol-patch-12-5-players-divided-1769816/
An imaginary CGI social influencer that aims to heal the World. (2018, October 11).
Journalism.Co.Uk. https://www.journalism.co.uk/press-releases/an-imaginary-cgi-social-
influencer-that-aims-to-heal-the-world/s66/a728825/
Ananny, M. (2016). Toward an Ethics of Algorithms: Convening, Observation, Probability, and
Timeliness. Science, Technology, & Human Values, 41(1), 93–117.
https://doi.org/10.1177/0162243915606523
Ananny, M., & Crawford, K. (2016). Seeing without knowing: Limitations of the transparency
ideal and its application to algorithmic accountability. New Media & Society,
1461444816676645. https://doi.org/10.1177/1461444816676645
Araque, J., & Moynihan, Q. (2021, September 7). A creepy virtual influencer looks so human,
her skin changes with her surroundings. Business Insider.
https://www.businessinsider.com/artificial-intelligence-influencer-virtual-ai-computer-
design-technology-innovation-china-2021-8
Araujo, T. (2018). Living up to the chatbot hype: The influence of anthropomorphic design cues
and communicative agency framing on conversational agent and company perceptions.
Computers in Human Behavior, 85, 183–189. https://doi.org/10.1016/j.chb.2018.03.051
Arditi, D. (2020). The Voice: Non-disclosure agreements and the hidden political economy of
reality TV. Popular Communication, 18(2), 138–151.
https://doi.org/10.1080/15405702.2020.1733577
264
Arnow-Richman, R. S., Carlson, G., Lobel, O., Roginsky, J., Short, J. L., & Starr, E. (2022).
Supporting Market Accountability, Workplace Equity, and Fair Competition by Reining
in Non-Disclosure Agreements (SSRN Scholarly Paper No. 4022812).
http://papers.ssrn.com/abstract=4022812
Ashley, V. (2016). Porn – artifice – performance – and the problem of authenticity. Porn Studies,
3(2), 187–190. https://doi.org/10.1080/23268743.2016.1184481
Bae, J. (2021, March 29). Gasang inpeullueonseo “roji(Rozy)”, cheot dandok paesyeon hwabo
gonggae [Virtual influencer “Rozy (Rozy)” reveals the first solo fashion photoshoot].
Kyeongsangilbo. http://www.ksilbo.co.kr/news/articleView.html?idxno=792760
Bakhtiari, K. (2020, December 30). Welcome To Hyperreality: Where The Physical And Virtual
Worlds Converge. Forbes.
https://www.forbes.com/sites/kianbakhtiari/2021/12/30/welcome-to-hyperreality-where-
the-physical-and-virtual-worlds-converge/
Baklanov, N. (2019, November 14). The Top Instagram Virtual Influencers in 2019. Hype -
Journal. https://hypeauditor.com/blog/the-top-instagram-virtual-influencers-in-2019/
Baklanov, N. (2021, December 7). The Top Virtual Instagram Influencers in 2021. Hype -
Journal. https://hypeauditor.com/blog/the-top-instagram-virtual-influencers-in-2021/
Balaji, M. (2009). Owning Black Masculinity: The Intersection of Cultural Commodification and
Self-Construction in Rap Music Videos. Communication, Culture and Critique, 2(1), 21–
38. https://doi.org/10.1111/j.1753-9137.2008.01027.x
Bandura, A. (1977). Social learning theory. Prentice-Hall.
Banet-Weiser, S. (2012). Authentic
TM
: The Politics of Ambivalence in a Brand Culture. In
Authentic
TM
. New York University Press. https://doi.org/10.18574/9780814739372
265
Banet-Weiser, S. (2018). Empowered: Popular Feminism and Popular Misogyny. Duke
University Press.
Banet-Weiser, S. (2021). Ruptures in Authenticity and Authentic Ruptures: Producing White
Influencer Vulnerability (Kids & Technology Essay Series).
https://www.mediatechdemocracy.com/work/ruptures-in-authenticity-and-authentic-
ruptures-producing-white-influencer-vulnerability
Banks, J. (2015). Object, Me, Symbiote, Other: A social typology of player-avatar relationships.
First Monday. https://doi.org/10.5210/fm.v20i2.5433
Banks, J. (2017). Multimodal, multiplex, multispatial: A network model of the self. New Media
& Society, 19(3), 419–438. https://doi.org/10.1177/1461444815606616
Banks, J. (Ed.). (2018). Avatar, Assembled: The Social and Technical Anatomy of Digital Bodies.
Peter Lang.
Barad, K. (2003). Posthumanist Performativity: Toward an Understanding of How Matter
Comes to Matter. Signs: Journal of Women in Culture and Society, 28(3), 801–831.
https://doi.org/10.1086/345321
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of
trait construct and stereotype activation on action. Journal of Personality and Social
Psychology, 71(2), 230–244. https://doi.org/10.1037/0022-3514.71.2.230
Baron, L. F., & Gomez, R. (2016). The Associations between Technologies and Societies: The
Utility of Actor-Network Theory. Science, Technology and Society, 21(2), 129–148.
https://doi.org/10.1177/0971721816640615
Barrett, E. (2019, June 19). Is Tech Too Intertwined For Huawei To Fall? Fortune.
https://fortune.com/2019/06/19/is-tech-too-intertwined-for-huawei-to-fall-data-sheet/
266
Baudrillard, J. (1994). The Precession of Simulacra. In Simulacra and Simulation. University of
Michigan Press.
BeBoss TV. (2020, December 1). VIRTUAL IDOLS Will Replace Traditional Kpop Idols?
https://www.youtube.com/watch?v=HJCJRa9P5PM
Beer, D. (2009). Power through the algorithm? Participatory web cultures and the technological
unconscious. New Media & Society, 11(6), 985–1002.
https://doi.org/10.1177/1461444809336551
Bennett, J. (2015). Systems and Things: On Vital Materialism and Object-Oriented Philosophy.
In R. Grusin (Ed.), The Nonhuman Turn (pp. 223–239). University of Minnesota Press.
BI India Partner. (2021, May 27). ASCI guidelines make it mandatory for influencers to label all
kinds of promotional content, whether monetary or not. Business Insider.
https://www.businessinsider.in/advertising/brands/article/asci-guidelines-make-it-
mandatory-for-influencers-to-label-all-kinds-of-promotional-content-whether-monetary-
or-not/articleshow/83003066.cms
Biocca, F. (1997). The Cyborg’s Dilemma: Progressive Embodiment in Virtual Environments.
Journal of Computer-Mediated Communication, 3(2). https://doi.org/10.1111/j.1083-
6101.1997.tb00070.x
Birnholtz, J., Fitzpatrick, C., Handel, M., & Brubaker, J. R. (2014). Identity, Identification and
Identifiability: The Language of Self-presentation on a Location-based Mobile Dating
App. Proceedings of the 16th International Conference on Human-Computer Interaction
with Mobile Devices & Services, 3–12. https://doi.org/10.1145/2628363.2628406
Bishop, S. (2018). Anxiety, panic and self-optimization: Inequalities and the YouTube algorithm.
Convergence, 24(1), 69–84. https://doi.org/10.1177/1354856517736978
267
Bishop, S. (2019). Managing visibility on YouTube through algorithmic gossip. New Media &
Society, 21(11–12), 2589–2606. https://doi.org/10.1177/1461444819854731
Bishop, S. (2021). Influencer Management Tools: Algorithmic Cultures, Brand Safety, and Bias.
Social Media + Society, 7(1), 20563051211003064.
https://doi.org/10.1177/20563051211003066
Black, D. (2012). The Virtual Idol: Producing and Consuming Digital Femininity. In P. W.
Galbraith & J. G. Karlin (Eds.), Idols and Celebrity in Japanese Media Culture (pp. 209–
228). Palgrave Macmillan UK. https://doi.org/10.1057/9781137283788_11
Black, T. C. (2019). Just a Robot Keeping It Real: Tba: Journal of Art, Media, and Visual
Culture, 1(1), 44-59-44–59.
Bory, P. (2019). Deep new: The shifting narratives of artificial intelligence from Deep Blue to
AlphaGo. Convergence, 25(4), 627–642. https://doi.org/10.1177/1354856519829679
Braidotti, R. (2013). The Posthuman. Polity Press.
Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in
Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa
Bucher, T. (2017). The algorithmic imaginary: Exploring the ordinary affects of Facebook
algorithms. Information, Communication & Society, 20(1), 30–44.
https://doi.org/10.1080/1369118X.2016.1154086
Cade, D. (2020, September 22). IKEA Used a CGI ‘Influencer’ as the Model for Its New Ad
Campaign. PetaPixel. https://petapixel.com/2020/09/22/ikea-is-using-a-cgi-influencer-as-
the-model-for-its-new-ad-campaign/
Caliandro, A., & Gandini, A. (2016). Qualitative Research in Digital Environments: A Research
Toolkit. Routledge. https://doi.org/10.4324/9781315642161
268
Callahan, K. (2021). CGI Social Media Influencers: Are They above the FTC’s Influence? Notes
& Comments. Journal of Business and Technology Law, 16(2), 361–386.
Caplan, R., & Gillespie, T. (2020). Tiered Governance and Demonetization: The Shifting Terms
of Labor and Compensation in the Platform Economy. Social Media + Society, 6(2),
2056305120936636. https://doi.org/10.1177/2056305120936636
Carey, J. W. (1967). Harold Adams Innis and Marshall McLuhan. The Antioch Review, 27(1), 5–
39. JSTOR. https://doi.org/10.2307/4610816
Carey, T. (2020, November 17). Virtual Instagrammers earn thousands a day and are adored by
millions. DailyMail. https://www.dailymail.co.uk/femail/article-8959717/Virtual-
Instagrammers-work-designer-brands-earn-thousands-day-adored-millions.html
Carpentier, N. (2011). Contextualising Author-Audience Convergences. Cultural Studies, 25(4–
5), 517–533. https://doi.org/10.1080/09502386.2011.600537
Casper, M. J. (1994). Reframing and Grounding Nonhuman Agency: What Makes a Fetus an
Agent. American Behavioral Scientist, 37(6), 839–856.
https://doi.org/10.1177/0002764294037006009
CBC. (2020, December 29). Pandemic creates a virtual influencer boom.
https://ca.news.yahoo.com/pandemic-creates-virtual-influencer-boom-031622569.html
Chae, Y. (2021, April 29). [M-keobeoseutori] “metabeoseu” sidae...beochueol hyumeoni
mollyeoonda [[M-coverstory] The age of “Metaverse”....virtual humans are swarming in.
Metroseoul. https://www.metroseoul.co.kr/article/20210429500237
Chang, M. C. (2021, January 23). Controversy over AI chatbot in South Korea raises questions
about ethics, data collection. The Straits Times. https://www.straitstimes.com/asia/east-
asia/controversy-over-ai-chatbot-in-south-korea-raises-questions-about-ethics-data
269
Chasin, A. (1995). Class and Its Close Relations: Identities among Women, Servants, and
Machines. In J. Halberstam & I. Livingston (Eds.), Posthuman Bodies (73rd–96th ed.).
Indiana University Press.
Chen, K.-H. (1987). The Masses and the Media: Baudrillard’s Implosive Postmodernism.
Theory, Culture & Society, 4(1), 71–88. https://doi.org/10.1177/026327687004001004
Chen, T. (2020, April 3). Coronavirus: The WHO Is Working With Robot Influencer Knox Frost
To Teach Gen Z. BuzzFeed. https://www.buzzfeednews.com/article/tanyachen/world-
health-organization-is-working-with-robot-influencer
Cheney-Lippold, J. (2017). We Are Data: Algorithms and The Making of Our Digital Selves.
NYU Press.
Cho, Y. (2004, November 24). “yuriui seong” e salgo inne [Living in a ’glass castle’].
Choogandonga. https://weekly.donga.com/List/3/all/11/67117/1
Choi, S. (2021, April 25). [Inpeullueonseo peurijeum] “sarami aniya?” gasang inpeullueonseo,
hyeonsil segye neombonda [[Influencer prism] “Not human?” Virtual influencers,
looking to overtake the real world]. The Fact.
http://news.tf.co.kr/read/livingculture/1856421.htm
Choi, T. (2021, March 19). AI ingan “geolgeuruptpsyutyubeo” sidaetssinganseongi
heundeullinda [AI human “girl group YouTuber” era...humanness shaken]. Moneytoday.
https://news.mt.co.kr/mtview.php?no=2021031916171773300
Choudhry, A., Han, J., Xu, X., & Huang, Y. (2022). “I Felt a Little Crazy Following a ‘Doll’”:
Investigating Real Influence of Virtual Influencers on Their Followers. Proceedings of
the ACM on Human-Computer Interaction, 6(GROUP), 43:1-43:28.
https://doi.org/10.1145/3492862
270
Christensen, L. S., Moritz, D., & Pearson, A. (2021). Psychological Perspectives of Virtual Child
Sexual Abuse Material. Sexuality & Culture, 25(4), 1353–1365.
https://doi.org/10.1007/s12119-021-09820-1
Chung, H. J. (2015). The reanimation of the digital (un)dead, or how to regenerate bodies in
digital cinema. Visual Studies, 30(1), 54–67.
https://doi.org/10.1080/1472586X.2015.996397
Coleman, L. deLisa. (2018, July 31). Inside The Incredible World Of A Virtual Mogul-In-The-
Making. Forbes. https://www.forbes.com/sites/laurencoleman/2018/07/31/inside-the-
incredible-world-of-a-virtual-mogul-in-the-making/
Conner, T. (2016). Hatsune Miku, 2.0Pac, and Beyond. The Oxford Handbook of Music and
Virtuality. https://doi.org/10.1093/oxfordhb/9780199321285.013.8
Considine, J. D. (1996, May 16). Kyoko Date: The world’s first virtual pop star. EW.Com.
https://ew.com/article/1997/05/16/kyoko-date-worlds-first-virtual-pop-star/
Cook, J. (2020, January 29). Brands Are Building Their Own Virtual Influencers. Are Their
Posts Legal? HuffPost. https://www.huffpost.com/entry/virtual-instagram-influencers-
sponcon_n_5e31cbefc5b6328af2ef97fd
Corry, F. (2021). Why does a platform die? Diagnosing platform death at Friendster’s end.
Internet Histories, 1–17. https://doi.org/10.1080/24701475.2021.1985360
Cotter, K. (2019). Playing the visibility game: How digital influencers and algorithms negotiate
influence on Instagram. New Media & Society, 21(4), 895–913.
https://doi.org/10.1177/1461444818815684
Crawford, S. (2007). Internet Think. Journal on Telecommunications and High Technology Law,
5, 467–486.
271
Crouch, G. (2019, August 10). Computer avatar Miquela models for Prada and is worth
£100million. DailyMail. https://www.dailymail.co.uk/news/article-7345299/Computer-
avatar-Miquela-Sousa-models-Prada-worth-100million.html
Crypton Future Media. (n.d.). Who is Hatsune Miku? Crypton Future Media. Retrieved June 21,
2022, from https://ec.crypton.co.jp/pages/prod/virtualsinger/cv01_us
Cuen, L. (2021, February 16). People Are Spending Millions on JPEGs, Tweets, And Other
Crypto Collectibles. Vice. https://www.vice.com/en/article/pkd83y/people-are-spending-
millions-on-jpegs-tweets-and-other-crypto-collectibles
Date Kyōko. (2022). In Wikipedia.
https://ja.wikipedia.org/w/index.php?title=%E4%BC%8A%E9%81%94%E6%9D%8F%
E5%AD%90&oldid=88494206
Davis, M. H., Conklin, L., Smith, A., & Luce, C. (1996). Effect of perspective taking on the
cognitive representation of persons: A merging of self and other. Journal of Personality
and Social Psychology, 70(4), 713–726. https://doi.org/10.1037/0022-3514.70.4.713
de Certeau, M. (1984). The Practice of Everyday Life. University of California Press.
de Souza e Silva, A. (2006). From Cyber to Hybrid: Mobile Technologies as Interfaces of Hybrid
Spaces. Space and Culture, 9(3), 261–278. https://doi.org/10.1177/1206331206289022
DeVito, M. A., Birnholtz, J., & Hancock, J. T. (2017). Platforms, People, and Perception: Using
Affordances to Understand Self-Presentation on Social Media. Proceedings of the 2017
ACM Conference on Computer Supported Cooperative Work and Social Computing,
740–754. https://doi.org/10.1145/2998181.2998192
272
Dibble, J. L., Hartmann, T., & Rosaen, S. F. (2016). Parasocial Interaction and Parasocial
Relationship: Conceptual Clarification and a Critical Assessment of Measures. Human
Communication Research, 42(1), 21–44. https://doi.org/10.1111/hcre.12063
DiGiuseppe, N., & Nardi, B. (2007). Real genders choose fantasy characters: Class choice in
world of warcraft. First Monday. https://doi.org/10.5210/fm.v12i5.1831
Djafarova, E., & Trofimenko, O. (2019). ‘Instafamous’ – credibility and self-presentation of
micro-celebrities on social media. Information, Communication & Society, 22(10), 1432–
1446. https://doi.org/10.1080/1369118X.2018.1438491
Donath, J. (2007). Signals in Social Supernets. Journal of Computer-Mediated Communication,
13(1), 231–251. https://doi.org/10.1111/j.1083-6101.2007.00394.x
Donath, J. S. (2002, June 1). Identity and deception in the virtual community. Communities in
Cyberspace. https://doi.org/10.4324/9780203194959-11
dongadatkeom yeonyenyuseutim. (2013, March 22). Adam silje model..."wonbiniramyeon
injeonghal subakke" [Adam’s real model..."if Wonbin, then we can only approve"].
Sports Donga. https://sports.donga.com/ent/article/all/20130322/53904532/2
Downey, G. L., Dumit, J., & Williams, S. (1995). Cyborg Anthropology. Cultural Anthropology,
10(2), 264–269. JSTOR.
Ducheneaut, N., Moore, R., & Nickell, E. (2007). Virtual “third places”: A case study of
sociability in massively multiplayer games. Computer Supported Cooperative Work, 16,
129–166.
Duffy, B. E. (2017). (Not) Getting paid to do what you love: Gender, social media, and
aspirational work. In (Not) Getting Paid to Do What You Love. Yale University Press.
https://www.degruyter.com/document/doi/10.12987/9780300227666/html
273
Duffy, B. E., & Hund, E. (2019). Gendered Visibility on Social Media: Navigating Instagram’s
Authenticity Bind. International Journal of Communication, 13, 4983–5002.
https://doi.org/10.32376/3f8575cb.3f03db0e
Duvall, S. (2021). Quiet celebrity in the time of pandemic: Stripping away artifice in
performances of self, cultures of citizenship, and community care. Continuum, 0(0), 1–
15. https://doi.org/10.1080/10304312.2021.1998372
EBS. (2021, March 29). Ril mikellaga nugunde nalliya [Who is Lil Miquela and why are people
talking about her] (Vol. 1). EBS.
https://ebr.ebs.co.kr/ebr/series?prodId=135279&courseId=10205720&stepId=20012040
Elder-Vass, D. (2015). Disassembling Actor-network Theory. Philosophy of the Social Sciences,
45(1), 100–121. https://doi.org/10.1177/0048393114525858
Eslami, M., Karahalios, K., Sandvig, C., Vaccaro, K., Rickman, A., Hamilton, K., & Kirlik, A.
(2016). First I “like” it, then I hide it: Folk Theories of Social Feeds. Proceedings of the
2016 CHI Conference on Human Factors in Computing Systems, 2371–2382.
https://doi.org/10.1145/2858036.2858494
ETX Daily Up. (2021, March 31). New tech is revolutionising the world of fashion. Free
Malaysia Today (FMT).
https://www.freemalaysiatoday.com/category/leisure/2021/03/31/new-tech-is-
revolutionising-the-world-of-fashion/
Evans, S. J. (2020, August 24). The glamorous Insta models raking in millions—But something’s
not quite right. The Sun. https://www.thesun.co.uk/tech/12461964/glamorous-robot-
instagram-models-9m-year/
274
Finan, D. (2021). Idols you can make: The player as auteur in Japan’s media mix. New Media &
Society, 14614448211015624. https://doi.org/10.1177/14614448211015625
Fischinger, D., Einramhof, P., Papoutsakis, K., Wohlkinger, W., Mayer, P., Panek, P., Hofmann,
S., Koertner, T., Weiss, A., Argyros, A., & Vincze, M. (2016). Hobbit, a care robot
supporting independent living at home: First prototype and lessons learned. Robotics and
Autonomous Systems, 75, 60–78. https://doi.org/10.1016/j.robot.2014.09.029
Flanagan, M. (1999). Mobile identities, digital stars, and post-cinematic selves. Wide Angle,
21(1), 76–93.
Flichy, P. (2007). The Internet Imaginaire. MIT Press.
Florian, A. (2021, August 18). This “imperfect” virtual influencer is challenging beauty
standards in China. CNN. https://www.cnn.com/style/article/angie-douyin-virtual-
influencer/index.html
Floridi, L. (2013). The Ethics of Information. Oxford University Press.
https://www.google.com/books/edition/The_Ethics_of_Information/_XHcAAAAQBAJ?
hl=en&gbpv=1&dq=The+ethics+of+information&pg=PP1&printsec=frontcover
Fodor, G. (2021, September 4). The Rise of Avatarism. Medium. https://gfodor.medium.com/the-
rise-of-avatarism-b7ecfa7c42bc
Fourie, L. (2020, June 25). Tech News: The weird world of social media influencers.
https://www.iol.co.za/business-report/tech-news-the-weird-world-of-social-media-
influencers-49913788
Fox, J. (2018). Boobs & Butts: The Babes Get the Gaze. In J. Banks (Ed.), Avatar, Assembled:
The Socio and Technical Anatomy of Digital Bodies (pp. 43–52). Peter Lang.
275
Fox, J., Ahn, S. J. (Grace), Janssen, J. H., Yeykelis, L., Segovia, K. Y., & Bailenson, J. N.
(2015). Avatars Versus Agents: A Meta-Analysis Quantifying the Effect of Agency on
Social Influence. Human–Computer Interaction, 30(5), 401–432.
https://doi.org/10.1080/07370024.2014.921494
Fox, J., & Moreland, J. J. (2015). The dark side of social networking sites: An exploration of the
relational and psychological stressors associated with Facebook use and affordances.
Computers in Human Behavior, 45, 168–176. https://doi.org/10.1016/j.chb.2014.11.083
Frank. (2007, July 25). Kyoko Date—Love Communication PV.
https://www.youtube.com/watch?v=PhfsZ66tEFY
Freberg, K., Graham, K., McGaughey, K., & Freberg, L. A. (2011). Who are the social media
influencers? A study of public perceptions of personality. Public Relations Review, 37(1),
90–92. https://doi.org/10.1016/j.pubrev.2010.11.001
Fredrickson, B. L., & Roberts, T.-A. (1997). Objectification Theory: Toward Understanding
Women’s Lived Experiences and Mental Health Risks. Psychology of Women Quarterly,
21(2), 173–206. https://doi.org/10.1111/j.1471-6402.1997.tb00108.x
French, M., & Bazarova, N. N. (2017). Is Anybody Out There?: Understanding Masspersonal
Communication Through Expectations for Response Across Social Media Platforms.
Journal of Computer-Mediated Communication, 22(6), 303–319.
https://doi.org/10.1111/jcc4.12197
Galbraith, P. W. (2009, October 14). Virtual idols blur line between real and imaginary. Japan
Today. https://japantoday.com/category/features/lifestyle/virtual-idols-blur-line-between-
real-and-imaginary
276
Galbraith, P. W. (2011). Lolicon: The Reality of “Virtual Child Pornography” in Japan. Image &
Narrative, 12(1).
Galbraith, P. W. (2012). Moe: Exploring Virtual Potential in Post-Millennial Japan. In T. Iles &
P. Matanle (Eds.), Researching Twenty-First Century Japan: New Directions and
Approaches for the Electronic Age. Lexington Books.
Galbraith, P. W. (2013). Maid Cafés: The Affect of Fictional Characters in Akihabara, Japan.
Asian Anthropology, 12(2), 104–125. https://doi.org/10.1080/1683478X.2013.854882
Galbraith, P. W. (2017). RapeLay and the return of the sex wars in Japan. Porn Studies, 4(1),
105–126. https://doi.org/10.1080/23268743.2016.1252159
Galinsky, A. D., & Moskowitz, G. B. (2000). Perspective-taking: Decreasing stereotype
expression, stereotype accessibility, and in-group favoritism. Journal of Personality and
Social Psychology, 78(4), 708–724. https://doi.org/10.1037/0022-3514.78.4.708
Gambino, A., Fox, J., & Ratan, R. (2020). Building a Stronger CASA: Extending the Computers
Are Social Actors Paradigm. Human-Machine Communication, 1(1).
https://doi.org/10.30658/hmc.1.5
Gershon, I., & Bell, J. A. (2013). Introduction: The Newness of New Media. Culture, Theory and
Critique, 54(3), 259–264. https://doi.org/10.1080/14735784.2013.852732
Gilden, A. (2021). Endorsing After Death (SSRN Scholarly Paper ID 3816046). Social Science
Research Network. https://papers.ssrn.com/abstract=3816046
Giles, D. C. (2002). Parasocial Interaction: A Review of the Literature and a Model for Future
Research. Media Psychology, 4(3), 279–305.
https://doi.org/10.1207/S1532785XMEP0403_04
277
Gillespie, T. (2014). The Relevence of Algorithm. In T. Gillespie, P. J. Boczkowski, & K. A.
Foot, Media Technologies: Essays on Communication, Materiality, and Society (pp. 167–
194). MIT Press.
Gillil, N. (2020, October 28). Four trends that show how influencer marketing has evolved in
2020. Econsultancy. https://econsultancy.com/four-trends-that-show-how-influencer-
marketing-has-evolved-in-2020/
Gil-Lopez, T., Shen, C., Benefield, G. A., Palomares, N. A., Kosinski, M., & Stillwell, D.
(2018). One Size Fits All: Context Collapse, Self-Presentation Strategies and Language
Styles on Facebook. Journal of Computer-Mediated Communication, 23(3), 127–145.
https://doi.org/10.1093/jcmc/zmy006
Gitelman, L. (2006). Always Already New: Media, History, and the Data of Culture. MIT Press.
http://ebookcentral.proquest.com/lib/socal/detail.action?docID=3338602
Glaser, B., & Strauss, A. (1967). The Discovery of Grounded Theory: Strategies for Qualitative
Research. Sociology Press.
Goffman, E. (1959). The presentation of self in everyday life. Doubleday.
Goffman, E. (1963). Stigma: Notes on the Management of Spoiled Identity. Touchstone.
Goulding, C. (2005). Grounded theory, ethnography and phenomenology: A comparative
analysis of three qualitative strategies for marketing research. European Journal of
Marketing, 39(3/4), 294–308. https://doi.org/10.1108/03090560510581782
Grady, C. (2020, June 30). Why the term “BIPOC” is so complicated, explained by linguists.
Vox. https://www.vox.com/2020/6/30/21300294/bipoc-what-does-it-mean-critical-race-
linguistics-jonathan-rosa-deandra-miles-hercules
278
Gray, M. L., & Suri, S. (2019). Ghost work: How to stop Silicon Valley from building a new
global underclass. Houghton Mifflin Harcourt.
Griner, D. (2019, April 11). KFC Created a “Virtual Influencer” Colonel Sanders, Mocking
Marketing’s Newest Odd Trend. Adweek. https://www.adweek.com/creativity/kfc-
created-a-virtual-influencer-colonel-sanders-mocking-marketings-newest-odd-trend/
Grizzard, M., & Ahn, C. (2018). Morality & Personality: Perfect and Deviant Selves. In J. Banks
(Ed.), Avatar, Assembled: The Socio and Technical Anatomy of Digital Bodies (pp. 117–
126). Peter Lang.
Grogan, B. (2020, November 16). Virtual Idol Variety Show “Dimension Nova” is a Bizarre
Digital Pop Star Battle. RADII. https://radiichina.com/dimension-nova/
Groom, V., Bailenson, J. N., & Nass, C. (2009). The influence of racial embodiment on racial
bias in immersive virtual environments. Social Influence, 4(3), 231–248.
https://doi.org/10.1080/15534510802643750
Gunkel, D. J. (2018). Robot Rights. MIT Press.
Guzman, A. L., & Lewis, S. C. (2020). Artificial intelligence and communication: A Human–
Machine Communication research agenda. New Media & Society, 22(1), 70–86.
https://doi.org/10.1177/1461444819858691
Haasch, P. (2020, November 11). Meet K/DA, the virtual pop group that teamed up with a fake
influencer who made controversial mental health posts. Insider.
https://www.insider.com/kda-members-league-of-legends-riot-games-what-is-twitter-
seraphine-2020
279
Hackl, C. (2021, January 29). How Brands Can Thrive In The Direct To Avatar Economy.
Forbes. https://www.forbes.com/sites/cathyhackl/2021/01/29/how-brands-can-thrive-in-
the-direct-to-avatar-economy/
Haga Yui. (2022). In Wikipedia.
https://ja.wikipedia.org/w/index.php?title=%E8%8A%B3%E8%B3%80%E3%82%86%E
3%81%84&oldid=87632843
Haileyesus, S. (2020, February 11). 42% of Gen Zers and Millennials Couldn’t Spot a CGI
Influencer. Small Business Trends. https://smallbiztrends.com/2020/02/cgi-
influencers.html
Halfmann, A., & Rieger, D. (2019). Permanently on Call: The Effects of Social Pressure on
Smartphone Users’ Self-Control, Need Satisfaction, and Well-Being. Journal of
Computer-Mediated Communication, 24(4), 165–181.
https://doi.org/10.1093/jcmc/zmz008
Hall, K. A. (2015). The authenticity of social-media performance: Lonelygirl15 and the amateur
brand of Young-Girlhood. Women & Performance: A Journal of Feminist Theory, 25(2),
128–142. https://doi.org/10.1080/0740770X.2015.1057011
Hall, S. (1993). Encoding, decoding. In S. During (Ed.), The Cultural Studies Reader (pp. 90–
103). Routledge.
Hampton, K. N., Lu, W., & Shin, I. (2016). Digital media and stress: The cost of caring 2.0.
Information, Communication & Society, 19(9), 1267–1286.
https://doi.org/10.1080/1369118X.2016.1186714
280
Hampton, R. (2020, November 4). Perhaps You’d Like to Read Some Weird International News
That Has Nothing Do With the U.S. Election. Slate. https://slate.com/human-
interest/2020/11/the-wildest-international-news-stories-no-election-just-whales.html
Han, J. (2021, July 7). Paen 2man5000myeong boyuhan 22se geunyeo, sarami anieotda [22-year
old she who has 25000 fans is not human]. www.donga.com.
https://www.donga.com/news/Society/article/all/20210707/107838477/1
Han, Y. (2021, August 12). “AI chaetbot ingyeokche anida”... Ingwonwi, iruda hyeomo bareon
jinjeong gaka [AI chatbot is not a person"...Human Rights association dismisses Lee
Ruda’s hate speech]. Chosunilbo.
https://www.chosun.com/national/incident/2021/08/12/BQD4CTB7ZRDBHHXHXA4O
UDW4HI/
Hancock, J. T., Naaman, M., & Levy, K. (2020). AI-Mediated Communication: Definition,
Research Agenda, and Ethical Considerations. Journal of Computer-Mediated
Communication. https://doi.org/10.1093/jcmc/zmz022
Hangukgyeongje. (2021, July 18). Ijipteu, tahiti...e siguge ’nomaseukeu’ro segye nubineun
yeoeu jeongche [Egypt, Tahiti...Identity of a woman who is roaming the world with “no
mask” in the current times]. Nate News.
https://news.nate.com/view/print?aid=20210718n12165
Haraway, D. (2016). Staying With the Trouble: Making Kin in the Chthulucene. Duke University
Press.
Haraway, D. (1991). Simians, Cyborgs, and Women: The Reinvention of Nature. Free
Association Books.
281
Hargrove, C. (2018, August 30). See Balmain’s First “Virtual Army” Of Models. Refinery29.
https://www.refinery29.com/en-gb/2018/08/208692/balmain-olivier-rousteing-digital-cgi-
supermodels
Harnes, A. (2019, June 21). Virtual Influencers Are Starting To Take Over—And It’s Really
Creepy. The Inquisitr. https://www.inquisitr.com/5493653/virtual-influencers-are-
starting-to-take-over-and-its-really-creepy/
Hayles, N. K. (1999). How We Became Posthuman: Virtual Bodies in Cybernetics, Literature,
and Informatics. University Of Chicago Press.
Hearn, A. (2010). Structuring feeling: Web 2.0, online ranking and rating, and the digital
‘reputation’ economy. Ephemera: Theory and Politics in Organization, 10(3–4), 421–
438.
Hearn, A., & Schoenhoff, S. (2015). From Celebrity to Influencer: Tracing the Diffusion of
Celebrity Value across the Data Stream. In P. D. Marshall & S. Redmond (Eds.), A
Companion to Celebrity (pp. 194–212). John Wiley & Sons, Inc.
https://doi.org/10.1002/9781118475089.ch11
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world?
Behavioral and Brain Sciences, 33(2–3), 61–83.
https://doi.org/10.1017/S0140525X0999152X
Herndon, H., & Dryhurst, M. (2020, May 14). Episode 3 Trevor McFedries (Brud / Lil Miquela).
https://interdependence.fm/episodes/interdependence-3-trevor-mcfedries-brud-lil-
miquela-nonpatrons
Highmore, B. (Ed.). (2002). The Everyday Life Reader. Routledge.
282
Hill, H. (2019, May 7). Here’s what you need to know about those CGI influencers invading
your feed. Mashable. https://mashable.com/article/cgi-influencers-instagram-what-you-
need-to-know
Hine, C. (2007). Multi-sited Ethnography as a Middle Range Methodology for Contemporary
STS. Science, Technology, & Human Values, 32(6), 652–671.
https://doi.org/10.1177/0162243907303598
H-information. (n.d.). Cyber Singer LUSIA. Retrieved June 20, 2022, from
https://web.archive.org/web/20010331033250/http://www.lusia.com/nonflash/index.html
Hogan, B. (2010). The Presentation of Self in the Age of Social Media: Distinguishing
Performances and Exhibitions Online. Bulletin of Science, Technology & Society, 30(6),
377–386. https://doi.org/10.1177/0270467610385893
Holland & Knight LLP. (2021, February 24). Advertising on Social Media Regulation:
Comments from an FTC Official. JD Supra.
https://www.jdsupra.com/legalnews/advertising-on-social-media-regulation-1638406/
Hollenbaugh, E. E. (2021). Self-Presentation in Social Media: Review and Research
Opportunities. Review of Communication Research, 9, 80–98.
Honea, K. (2020, October 20). LoL marketing has sparked controversy with Seraphine and
mental health. Daily Esports. https://www.dailyesports.gg/lol-marketing-has-sparked-
controversy-with-seraphine-and-mental-health/
Honer, A., & Hitzler, R. (2015). Life-World-Analytical Ethnography: A Phenomenology-Based
Research Approach. Journal of Contemporary Ethnography, 44(5), 544–562.
https://doi.org/10.1177/0891241615588589
283
Hong, S. (1998, January 23). Saibeogasu cheottansaeng [First birth of a cyber singer]. KBS
News. https://news.kbs.co.kr/news/view.do?ncd=3782356
Hong, S., Jahng, M. R., Lee, N., & Wise, K. R. (2020). Do you filter who you are?: Excessive
self-presentation, social cues, and user evaluations of Instagram selfies. Computers in
Human Behavior, 104, 106159. https://doi.org/10.1016/j.chb.2019.106159
HoriPro Inc. (1997, June 7). DK-96. DK-96.
https://web.archive.org/web/19970607011419/http://www.dhw.co.jp/horipro/talent/DK96
/dev_e.html
HoriPro & indecom. (2003, June 23). Digital Kids. Digital Kids.
https://web.archive.org/web/20030623085545/http://www.diki.co.kr/
Hornborg, A. (2017). Artifacts have consequences, not agency: Toward a critical theory of
global environmental history. European Journal of Social Theory, 20(1), 95–110.
https://doi.org/10.1177/1368431016640536
Horton, D., & Wohl, R. R. (1956). Mass Communication and Para-Social Interaction.
Psychiatry, 19(3), 215–229. https://doi.org/10.1080/00332747.1956.11023049
Hsieh, C. (2019, June 21). Meet This Porn Company’s User-Generated “Ideal” Woman
(Warning: This Will Make You Rage-Smash Your Screen). Cosmopolitan.
https://www.cosmopolitan.com/sex-love/a28134490/xhamster-data-ideal-woman-shy-
yume/
Hsu, T. (2019, June 18). These influencers aren’t flesh and blood, yet millions follow them. NZ
Herald. https://www.nzherald.co.nz/lifestyle/these-influencers-arent-flesh-and-blood-yet-
millions-follow-them/75IVYEPX5X7XXJFXOJB2RUP4AU/
284
Huh, S., & Williams, D. (2009). Dude looks like a lady: Online game gender swapping. In W.
Bainbridge (Ed.), Online worlds: Convergence of the real and the virtual. Springer.
Human Zoos: A Shocking History of Shame and Exploitation. (n.d.). CBC. Retrieved June 23,
2022, from https://www.cbc.ca/natureofthings/features/human-zoos-a-shocking-history-
of-shame-and-exploitation
Hussain, Z., & Griffiths, M. D. (2008). Gender Swapping and Socializing in Cyberspace: An
Exploratory Study. CyberPsychology & Behavior, 11(1), 47–53.
https://doi.org/10.1089/cpb.2007.0020
Hwang, H. (2020, June 10). Alleksandeo maekkwinui aembaeseodeoga doen jeonjihyeon [Jeon
Jihyun became Alexander McQueen’s ambassador]. Vogue Korea.
https://www.vogue.co.kr/2020/06/10/%ec%95%8c%eb%a0%89%ec%82%b0%eb%8d%
94-%eb%a7%a5%ed%80%b8%ec%9d%98-%ec%97%a0%eb%b2%84%ec%84%9c%eb
%8d%94%ea%b0%80-%eb%90%9c-%ec%a0%84%ec%a7%80%ed%98%84/
Jackson, L. M. (2018, May 4). Shudu Gram Is a White Man’s Digital Projection of Real-Life
Black Womanhood. The New Yorker. https://www.newyorker.com/culture/culture-
desk/shudu-gram-is-a-white-mans-digital-projection-of-real-life-black-womanhood
Janssen, L., Schouten, A. P., & Croes, E. A. J. (2022). Influencer advertising on Instagram:
Product-influencer fit and number of followers affect advertising outcomes and
influencer evaluations via credibility and identification. International Journal of
Advertising, 41(1), 101–127. https://doi.org/10.1080/02650487.2021.1994205
Jeffries, D., & Jeffries, R. (2017). Marxist Materialism and Critical Race Theory: A Comparative
Analysis of Media and Cultural Influence on the Formation of Stereotypes and
285
Proliferation of Police Brutality against Black Men. Spectrum: A Journal on Black Men,
5(2), 1–22. https://doi.org/10.2979/spectrum.5.2.01
Jenkins, H., Ford, S., & Green, J. (2013). Spreadable Media: Creating Value and Meaning in a
Networked Culture. NYU Press.
Jennings, R. (2021, July 6). TikTok’s catfish problem is worse than you think. Vox.
https://www.vox.com/the-goods/2021/7/6/22561197/coconutkitty-diana-deets-
itsnotdatsrs-asianfishing-deepfake-catfish
Jenson, J., Taylor, N., de Castell, S., & Dilouya, B. (2015). Playing With Our Selves:
Multiplicity and identity in online gaming. Feminist Media Studies, 15(5), 860–879.
https://doi.org/10.1080/14680777.2015.1006652
Jeon, J. (2018, February 25). “syugamaen2” saibeogasu adameun bakseongcheol..."cheoeum nae
eolgullo mudae seotda" [’Sugarman2’ cyber singer Adam is Park Sung Chul..."I
performed in my face for the first time"]. Seoulgyeongje.
https://www.sedaily.com/NewsVIew/1RVUC8W0T3
Jeong, D. (2006, April 7). [Jeongdeokyeonui nemonan sesang] gayogye, saibeo gasuga neomu
manta [[Jeong Deokhyeon’s Square World] Too many cyber singers in the music
industry]. OSEN. https://n.news.naver.com/mnews/article/109/0000030134?sid=001
Jo, J. (2016, March 6). [Beulluseukeurin] segimal saibeogasu ’adam’ui choehu [[The end of
bluescreen] End of the millenium cyber singer “Adam”’s end]. Economic Review.
http://www.econovill.com/news/articleView.html?idxno=283370
Jones, A., Bennett, R. J., & Cross, S. (2015). Keepin’ it real? Life, death, and holograms on the
live music stage. In A. Jones & R. J. Bennett (Eds.), The Digital Evolution of Live Music.
Chandos Publishing.
286
JTBC Entertainment. (2018, February 25). Adam, ssodajineun seoboe yocheongedo hwaldong
motan iyu “gisullyeok bujok” tuyu peurojekteu-syugamaen2 6hoe [Adam, the reason he
was not active despite pouring requests “technological limitation” Two Yoo Project-
Sugarman2 episode 6]. https://www.youtube.com/watch?v=TDW1CLK1rq8
Jung, H. (2021, June 9). Saramboda deo saramgateun “dijiteolhyumeon’i tteunda [More human
than human “digital humans” are getting popular]. Segyebiz.
http://m.segyebiz.com/newsView/20210609514582
Jung, Y. (2019, November 28). [Nnyeonjeonnyuseu] 1998nyeon hyeseonggachi debwihan
“saibeogasu adam” [[News from N years ago] Debuted like a comet in 1998 ‘cyber singer
Adam’]. YTN. https://www.ytn.co.kr/_ln/0134_201911280929127858
Jurgenson, N. (2012, October 29). Strong and mild digital dualism. Cyborgology.
https://thesocietypages.org/cyborgology/2012/10/29/strong-and-mild-digital-dualism/
Kádeková, Z., & Holienčinová, M. (2018). Influencer Marketing as a Modern Phenomenon
Creating a New Frontier of Virtual Opportunities. Communication Today, 9(2), 90–105.
Kang, D. (1999, July 19). [Deo saibeo] netijeun: Saibeogasudeul “ingimori” jaegadong [[The
cyber]: Netizen: Cyber singers “gathering popularity” again. Hangukgyeongjesinmun.
https://www.hankyung.com/news/article/1999071904281
Kang, J., & Wei, L. (2020). Let me be at my funniest: Instagram users’ motivations for using
Finsta (a.k.a., fake Instagram). The Social Science Journal, 57(1), 58–71.
https://doi.org/10.1016/j.soscij.2018.12.005
Kang, S. (2001, May 8). [Eumak]’yaksok’bureun “jero” neun saibeogasu adam [[Music] “Zero”
who sang “Yaksok” is cyber singer Adam. Dongailbo.
https://n.news.naver.com/mnews/article/020/0000062811?sid=103
287
Katz, J., & Csordas, T. J. (2003). Phenomenological Ethnography in Sociology and
Anthropology. Ethnography, 4(3), 275–288. https://doi.org/10.1177/146613810343001
KBS. (2021, July 17). Wontagui gijadeul—Beochueol inpeullueonseo, eodikkaji hwaryongdoel
su isseulkka? [Roundtable of journalists—Virtual influencer, to what extent can they be
utilized?]. In Seonggongyegam Gimbanghuiimnida. KBS.
Khamis, S., Ang, L., & Welling, R. (2017). Self-branding, ‘micro-celebrity’ and the rise of
Social Media Influencers. Celebrity Studies, 8(2), 191–208.
https://doi.org/10.1080/19392397.2016.1218292
Kim, D.-W. (2015). The Evolution of Cyber Singer Viewed from the Coevolution of Man and
Machine. Cartoon and Animation Studies, 39, 261–295.
https://doi.org/10.7230/KOSCAS.2015.39.261
Kim, G. (2021, July 26). [Teukjingju] gyesokdoeneun metabeoseu tuja yeolgi...alchera 13%
geupdeung[[characteristic stock] continuing Metaverse investment interests...Alchera
rises 13%]. Seoulgyeongje. https://www.sedaily.com/NewsView/22P29DVY4F/GA0101
Kim, J. (2022, March 2). Eumwon naego debwihan tgasangingantdeultt gasutpsyeongikkaji
yeongyeok neolpilkka [’virtual humans’ who released music and debuted...singer·acting
widening areas]. Seoulsinmun.
https://www.seoul.co.kr/news/newsView.php?id=20220302024005
Kim, S. (2000, June 8). gasang kaerikteoe saengmyeongnyeogeul... [Life to virtual characters...].
Hangyeorye 21. http://legacy.h21.hani.co.kr/h21/data/L000529/1p3p5t2d.html
Kim, S. (2021, September 21). “chaeunuin jul aratda”..hwajeui gongdaesaeng nugugillae
["Thought he was Cha Eunwoo"..who is the it engineering student]. Hankyung.
https://www.hankyung.com/it/article/2021092169737
288
Kim, Y. (2021, January 13). Chatbot Lee Luda brings ethical concerns around AI. AsiaToday.
http://en.asiatoday.co.kr/view.php?key=20210112002141234
Kinstler, L. (2019). Finding Lena, the Patron Saint of JPEGs. Wired.
https://www.wired.com/story/finding-lena-the-patron-saint-of-jpegs/
Kitchin, R. (1998). Cyberspace: The World in the Wires. Wiley. https://www.wiley.com/en-
us/Cyberspace%3A+The+World+in+the+Wires-p-9780471978626
Kitchin, R., & Dodge, M. (2011). Code/space: Software and everyday life. MIT Press.
Klein, M. (2020, November 17). The Problematic Fakery Of Lil Miquela Explained—An
Exploration Of Virtual Influencers and Realness. Forbes.
https://www.forbes.com/sites/mattklein/2020/11/17/the-problematic-fakery-of-lil-
miquela-explained-an-exploration-of-virtual-influencers-and-realness/
Kracker, D. (2014, June 22). The Face of Her Generation. Ceiling Gallery. http://www.ceiling-
gallery.com/blog/2014/6/14/haga-yui
Kutsugi, K. (n.d.). Kutsugi Kenichi no nijigen kurabu. Retrieved June 20, 2022, from
https://www.kutsugi.com/
Lambert-Beatty, C. (2009). Make-Believe: Parafiction and Plausibility. October, 129, 51–84.
https://doi.org/10.1162/octo.2009.129.1.51
Latour, B. (1994). Where are the Missing Masses? The Sociology of a Few Mundane Artifacts.
In W. Bijker & J. Law (Eds.), Shaping Technology / Building Society: Studies in
Sociotechnical Change. MIT Press.
Latour, B. (1996). On actor-network theory: A few clarifications. Soziale Welt, 47(4), 369–381.
JSTOR.
289
Law, J. (2009). Actor Network Theory and Material Semiotics. In The New Blackwell
Companion to Social Theory (pp. 141–158). John Wiley & Sons, Ltd.
https://doi.org/10.1002/9781444304992.ch7
Lee, B. (2002, October 25). Saibeo gasudeul jongjeok gamchuda [Cyber singers hide traces].
Gyeonghyanggeimseu. https://www.khgames.co.kr/news/articleView.html?idxno=11924
Lee, B. (2022, April 3). “naeeolgul naenwara”...gasangingan, ingan model chosang. Chimhae
nollan [Biz tok] [“give my face back"...virtual influencer, human model rights to likeness
infringement controversy [Biz talk]]. Chosunilbo.
https://www.chosun.com/economy/tech_it/2022/04/04/VINLSHCHYJHWZI5T5MHNY
YKFGA/
Lee, E. (1998, February 3). Noraehaneun “saibeo seuta” itdara tansaeng [Singing “cyber stars”
continue to birth]. Jeonjasinmun. https://www.etnews.com/199802030071?SNS=00002
Lee, J. (2019, July 20). [MHNhwabo] “sinipsagwan guhaeryeong” hwaryeohan bijueorui
dowondaegun, eolgulcheonjae chaeunu [[MHN photoshoot] “Rookie Historian Goo Hae-
ryung” fabulous visual king Dowon, face genius Cha eunwoo]. Moonhwanews.
https://m.news.zum.com/articles/53870541
Lee, J. (2020, November 3). Riot on how League of Legends’ latest character may have gotten a
little too real. Polygon.
https://www.polygon.com/interviews/2020/11/3/21548141/seraphine-controversy-league-
of-legends-twitter
Lee, K. M., & Nass, C. (2003). Designing Social Presence of Social Actors in Human Computer
Interaction. Proceedings of the SIGCHI Conference on Human Factors in Computing
Systems, 289–296. https://doi.org/10.1145/642611.642662
290
Leighton, H. (2019, November 26). What It Means For Virtual Instagram Influencers’ Popularity
Rising. Forbes. https://www.forbes.com/sites/heatherleighton/2019/11/26/what-it-means-
for-virtual-instagram-influencers-popularity-rising/?sh=706d85946e09
Liffreing, I. (2019, December 4). Essence Cosmetics is the latest brand with a virtual Instagram
influencer. Ad Age. https://adage.com/article/digital/essence-cosmetics-latest-brand-
virtual-instagram-influencer/2220351
Ling, R. (2016). Soft coercion: Reciprocal expectations of availability in the use of mobile
communication. First Monday. https://doi.org/10.5210/fm.v21i9.6814
Livingstone, S. (1999). New media, new audiences? New Media and Society, 1(1), 59–66.
Lopez, S., Yang, Y., Beltran, K., Kim, S. J., Cruz Hernandez, J., Simran, C., Yang, B., & Yuksel,
B. F. (2019). Investigating Implicit Gender Bias and Embodiment of White Males in
Virtual Reality with Full Body Visuomotor Synchrony. Proceedings of the 2019 CHI
Conference on Human Factors in Computing Systems, 1–12.
https://doi.org/10.1145/3290605.3300787
Luthera, N. (2020, January 16). Council Post: The Dark Side Of Deepfake Artificial Intelligence
And Virtual Influencers. Forbes.
https://www.forbes.com/sites/forbesbusinesscouncil/2020/01/16/the-dark-side-of-
deepfake-artificial-intelligence-and-virtual-influencers/
macbain. (1999, April 20). Date Kyoko Info. N.d.
https://web.archive.org/web/19990420075559/http://home.inreach.com/macbain/dk96inf
o.htm
291
Maeilgyeongje. (2021, July 20). Ojeonjang★temadonghyang [morning market ★theme
tendencies]. Maeilgyeongje.
https://vip.mk.co.kr/newSt/news/news_view.php?sCode=21&t_uid=23&c_uid=379301
Maeilsinmun. (1998, February 17). Saibeogasu “adam” “ryusia” gwanggotpkaerikteokkaji
ingimori [Cyber singers “Adam” “Lusia” getting advertisements and characters, popular].
Maeilsinmun. https://news.imaeil.com/page/view/1998021714302342360
Malamuth, N. M. (2018). “Adding fuel to the fire”? Does exposure to non-consenting adult or to
child pornography increase risk of sexual aggression? Aggression and Violent Behavior,
41, 74–89. https://doi.org/10.1016/j.avb.2018.02.013
Manovich, L. (2001). The Language of New Media. MIT Press, 28.
Maras, M.-H., & Alexandrou, A. (2019). Determining authenticity of video evidence in the age
of artificial intelligence and in the wake of Deepfake videos. The International Journal of
Evidence & Proof, 23(3), 255–262. https://doi.org/10.1177/1365712718807226
Marcus, G. E. (1995). Ethnography in/of the World System: The Emergence of Multi-Sited
Ethnography. Annual Review of Anthropology, 24, 95–117.
Martey, R. M., & Consalvo, M. (2011). Performing the Looking-Glass Self: Avatar Appearance
and Group Identity in Second Life. Popular Communication, 9(3), 165–180.
https://doi.org/10.1080/15405702.2011.583830
Martey, R. M., Stromer-Galley, J., Banks, J., Wu, J., & Consalvo, M. (2014). The strategic
female: Gender-switching and player behavior in online games. Information,
Communication & Society, 17(3), 286–300.
https://doi.org/10.1080/1369118X.2013.874493
292
Marvin, C. (1988). Chapter 1: Inventing the Expert. In When Old Technologies Were New:
Thinking About Electric Communication in the Late Nineteenth Century. Oxford
University Press.
Marwick, A. E. (2015). You May Know Me from YouTube (Micro-) Celebrity in Social Media.
In P. D. Marshall & S. Redmond (Eds.), A Companion to Celebrity (pp. 333–350). John
Wiley & Sons.
Marwick, A. E., & boyd, danah. (2011). I tweet honestly, I tweet passionately: Twitter users,
context collapse, and the imagined audience. New Media & Society, 13(1), 114–133.
https://doi.org/10.1177/1461444810365313
Marx, L. (2010). Technology: The Emergence of a Hazardous Concept. Technology and Culture,
51(3), 561–577.
Maso, I. (2007). Ch 9 Phenomenology and Ethnography. In P. Atkinson, S. Delamont, A.
Coffey, J. Lofland, & L. Lofland (Eds.), Handbook of Ethnography (pp. 136–144).
SAGE.
Matassi, M., Boczkowski, P. J., & Mitchelstein, E. (2019). Domesticating WhatsApp: Family,
friends, work, and study in everyday communication. New Media & Society, 21(10),
2183–2200. https://doi.org/10.1177/1461444819841890
Matney, L. (2021, October 4). NFT startup Dapper Labs acquires virtual influencer startup Brud |
TechCrunch. TechCrunch. https://techcrunch.com/2021/10/04/nft-startup-dapper-labs-
acquires-virtual-influencer-startup-brud/
McCracken, G. (1989). Who Is the Celebrity Endorser? Cultural Foundations of the
Endorsement Process. Journal of Consumer Research, 16(3), 310–321.
https://doi.org/10.1086/209217
293
McLuhan, M. (1964). Understanding media: The extensions of man. The MIT Press.
McRobbie, A. (2020). Feminism and the politics of resilience: Essays on gender, media and the
end of welfare. Polity Press.
Mears, A., & Finlay, W. (2005). Not Just a Paper Doll: How Models Manage Bodily Capital and
Why They Perform Emotional Labor. Journal of Contemporary Ethnography, 34(3),
317–343. https://doi.org/10.1177/0891241605274559
Melenboim, T. (2019, June 30). Exploring the Blurry Lines Between Humans, VR and AR. The
Times of Israel. https://blogs.timesofisrael.com/exploring-the-blurry-lines-between-
humans-vr-and-ar/
Meta. (2021, October 28). The Facebook Company Is Now Meta. Meta.
https://about.fb.com/news/2021/10/facebook-company-is-now-meta/
Meyrowitz, J. (1985). No sense of place: The impact of electronic media on social behavior.
Oxford University Press.
Mikkola, M. (2018). Pornographic Videogames: A Feminist Examination. In The Aesthetics of
Videogames. Routledge.
Miller-Young, M. (2008). Hip-Hop Honeys and Da Hustlaz. Meridians, 8(1), 261–292.
https://doi.org/10.2979/MER.2008.8.1.261
Milton, J. (2017, March 6). From holograms to headliners – how Gorillaz’s inventive live shows
have transformed. NME. https://www.nme.com/blogs/nme-blogs/holograms-headliners-
how-gorillaz-live-shows-changed-2003855
Miquela. (2019, May 16). Miquela and Bella Hadid Get Surreal | CALVIN KLEIN.
https://www.youtube.com/watch?v=JuTowFf6B9I
294
Mitra, A., & Schwartz, R. L. (2001). From Cyber Space to Cybernetic Space: Rethinking the
Relationship between Real and Virtual Spaces. Journal of Computer-Mediated
Communication, 7(1). https://doi.org/10.1111/j.1083-6101.2001.tb00134.x
Moon, Y. (2000). Intimate Exchanges: Using Computers to Elicit Self-Disclosure from
Consumers. Journal of Consumer Research, 26(4), 323–339.
https://doi.org/10.1086/209566
Moradi, B., & Huang, Y.-P. (2008). Objectification Theory and Psychology of Women: A
Decade of Advances and Future Directions. Psychology of Women Quarterly, 32(4), 377–
398. https://doi.org/10.1111/j.1471-6402.2008.00452.x
Mori, M. (1970). The uncanny valley. Energy, 7(4), 33–35.
Morstatter, F., Kim, D. O. (Donna), Jonckheere, N., Liu, C., Seth, M., & Williams, D. (2021).
I’ll Play on My Other Account: The Network and Behavioral Differences of Sybils.
Proceedings of the ACM on Human-Computer Interaction, 5(CHI PLAY), 277:1-277:18.
https://doi.org/10.1145/3474704
Mosco, V. (2004). The Digital Sublime: Myth, Power, and Cyberspace.
https://doi.org/10.7551/mitpress/2433.001.0001
Mou, Y., & Xu, K. (2017). The media inequality: Comparing the initial human-human and
human-AI social interactions. Computers in Human Behavior, 72, 432–440.
https://doi.org/10.1016/j.chb.2017.02.067
Müller, M., & Schurr, C. (2016). Assemblage thinking and actor-network theory: Conjunctions,
disjunctions, cross-fertilisations. Transactions of the Institute of British Geographers,
41(3), 217–229. https://doi.org/10.1111/tran.12117
295
Mulvey, L. (1989). Visual pleasure and narrative cinema. In Visual and other pleasures (pp. 14–
26). Palgrave Macmillan.
Mulvin, D. (2021). Proxies: The Cultural Work of Standing In. MIT Press.
Mumford, L. (1962). Technics & Civilization. The University of Chicago Press.
Murray, J. H. (2020). Virtual/reality: How to tell the difference. Journal of Visual Culture, 19(1),
11–27. https://doi.org/10.1177/1470412920906253
Myers West, S. (2018). Censored, suspended, shadowbanned: User interpretations of content
moderation on social media platforms. New Media & Society, 20(11), 4366–4383.
https://doi.org/10.1177/1461444818773059
Nagy, J., & Turner, F. (2019). The Selling of Virtual Reality: Novelty and Continuity in the
Cultural Integration of Technology. Communication, Culture and Critique, 12(4), 535–
552. https://doi.org/10.1093/ccc/tcz038
Nakabayashi A. (2019, February 1). Ganso bācharu tarento date kyōko no musume " date ayano
" VTuber debyū [Kyoko Date’s daughter “Ayano Date” debuts on VTuber]. AV Watch.
https://av.watch.impress.co.jp/docs/news/1167731.html
Napoli, P. M., & Caplan, R. (2017). Why media companies insist they’re not media companies,
why they’re wrong, and why it matters. First Monday, 22(5).
Nass, C., Moon, Y., Fogg, B. J., Reeves, B., & Dryer, C. (1995). Can computer personalities be
human personalities? Conference Companion on Human Factors in Computing Systems
- CHI ’95, 228–229. https://doi.org/10.1145/223355.223538
Nass, C., Steuer, J., & Tauber, E. R. (1994). Computers are social actors. Proceedings of the
SIGCHI Conference on Human Factors in Computing Systems, 72–78.
https://doi.org/10.1145/191666.191703
296
Nasubi. (2021). In Wikipedia.
https://en.wikipedia.org/w/index.php?title=Nasubi&oldid=1062944377
Natale, S. (2016a). There Are No Old Media. Journal of Communication, 66(4), 585–603.
https://doi.org/10.1111/jcom.12235
Natale, S. (2016b). Unveiling the Biographies of Media: On the Role of Narratives, Anecdotes,
and Storytelling in the Construction of New Media’s Histories. Communication Theory,
26(4), 431–449. https://doi.org/10.1111/comt.12099
Neff, G., & Nagy, P. (2016). Talking to Bots: Symbiotic Agency and the Case of Tay.
International Journal of Communication, 10(0), 17.
Nelson, L. R. (2016). Choosing illusion: Mediated reality and the spectacle of the Idol in Kōji
Shiraishi’s Shirome. Journal of Japanese and Korean Cinema, 8(2), 140–155.
https://doi.org/10.1080/17564905.2016.1222149
Nikkan Saizō. (2008, July 2). ganso dejitaru aidoru date kyōko wa nan do demo yomigaeru!?
[The original digital idol Kyoko Date revives many times !?]. Nikkan Saizō, July.
https://www.cyzo.com/2008/07/post_679_entry.html
(No. 6). (2018, February 25). In Two Yoo Project Sugarman. JTBC.
Noble, S. U. (2018). Algorithms of Oppression: How Search Engines Reinforce Racism. In
Algorithms of Oppression. New York University Press.
https://doi.org/10.18574/9781479833641
Nowak, K. L., & Fox, J. (2018). Avatars and Computer-Mediated Communication: A Review of
the Definitions, Uses, and Effects of Digital Representations. 25.
297
Obuno, Y. (2022, January 11). What happened to the Japanese man who “married” virtual
character Hatsune Miku? The Mainichi.
https://mainichi.jp/english/articles/20220111/p2a/00m/0li/028000c
Öhman, C., & Floridi, L. (2017). The Political Economy of Death in the Age of Information: A
Critical Approach to the Digital Afterlife Industry. Minds and Machines, 27(4), 639–662.
https://doi.org/10.1007/s11023-017-9445-2
Oliveira, A. B. da S., & Chimenti, P. (2021). “Humanized Robots”: A Proposition of Categories
to Understand Virtual Influencers. Australasian Journal of Information Systems, 25.
https://doi.org/10.3127/ajis.v25i0.3223
O’Sullivan, P. B., & Carr, C. T. (2018). Masspersonal communication: A model bridging the
mass-interpersonal divide. New Media & Society, 20(3), 1161–1180.
https://doi.org/10.1177/1461444816686104
Oware, M. (2011). Brotherly Love: Homosociality and Black Masculinity in Gangsta Rap Music.
Journal of African American Studies, 15(1), 22–39. https://doi.org/10.1007/s12111-010-
9123-4
Paik, P. C., & Shi, C.-K. (2013). Playful gender swapping: User attitudes toward gender in
MMORPG avatar customisation. Digital Creativity, 24(4), 310–326.
https://doi.org/10.1080/14626268.2013.767275
Papacharissi, Z. (2010). A Networked Self: Identity, Community, and Culture on Social Network
Sites. Routledge.
Papacharissi, Z. (2012). Without You, I’m Nothing: Performances of the Self on Twitter.
International Journal of Communication, 6(0), 18.
298
Park, G. (2020, August 27). K/DA, the ‘League of Legends’ K-pop group, is back with a full EP.
Washington Post. https://www.washingtonpost.com/video-games/2020/08/27/kda-league-
legends-k-pop-group-is-back-with-full-ep/
Park, S. (2016, June 27). South Koreans cringe over cyber singer’s resurrection. Aju Business
Daily. http://ajudaily.com/view/20160627104134045
Pat Lok. (2021, February 19). Trevor McFedries (Lil Miquela, Brud) on crypto, NFTs and the
future of music (No. 35) [Livestream]. In Fone a Friend. Twitch.tv/PatLokMusic
Paul, B., & Linz, D. G. (2008). The Effects of Exposure to Virtual Child Pornography on Viewer
Cognitions and Attitudes Toward Deviant Sexual Behavior. Communication Research,
35(1), 3–38. https://doi.org/10.1177/0093650207309359
Paul, K. (2021, October 23). Lyft admits it recorded 4,000 sexual assault claims in long-awaited
report. The Guardian. https://www.theguardian.com/technology/2021/oct/22/lyft-sexual-
assault-reports-uber-ridesharing
Pero, J. (2019, April 15). CGI influencer Lil Miquela makes Coachella debut interviewing J
Balvin. DailyMail. https://www.dailymail.co.uk/sciencetech/article-6924617/CGI-
influencer-Lil-Miquela-makes-Coachella-debut-interviews-artist-J-Balvin.html
Persaud, C. J., & Perks, M. E. (2022). Beauty From the Waist Up: Twitch Drag, Digital Labor,
and Queer Mediated Liveness. Television & New Media, 15274764221080912.
https://doi.org/10.1177/15274764221080912
Peters, B. (2009). And lead us not into thinking the new is new: A bibliographic case for new
media history. New Media & Society, 11(1–2), 13–30.
https://doi.org/10.1177/1461444808099572
299
Petre, C., Duffy, B. E., & Hund, E. (2019). “Gaming the System”: Platform Paternalism and the
Politics of Algorithmic Visibility. Social Media + Society, 5(4), 2056305119879995.
https://doi.org/10.1177/2056305119879995
Pettit, H. (2019, May 2). “Virtual” Instagram star Imma poses in fashion mag with real models –
but she’s completely FAKE. The Sun. https://www.thesun.co.uk/tech/8989729/virtual-
instagram-star-imma-id-fashion-mag/
Pfadenhauer, M., & Grenz, T. (2015). Uncovering the Essence: The Why and How of
Supplementing Observation with Participation in Phenomenology-Based Ethnography.
Journal of Contemporary Ethnography, 44(5), 598–616.
https://doi.org/10.1177/0891241615587382
Pikicast. (2021, February 20). Sarami anirago? Bwadobwado nollaun AI inpeullueonseo [Not
human? AI influencers that are more surprising the more you see]. Facebook.
https://www.facebook.com/allnewpiki/posts/3840305259389890
Pinch, T., & Bijker, T. (2012). The Social Construction of Facts and Artifacts: Or How the
Sociology of Science and the Sociology of Technology Might Benefit Each Other. In W.
E. Bijker Hughes, P. &. Pinch, T. (Ed.), The Social Construction of Technological
Systems (pp. 17–50). The MIT Press.
Pooley, J., & Socolow, M. J. (2013). Checking up on The Invasion from Mars: Hadley Cantril,
Paul Lazarsfeld, and the making of a misremembered classic. International Journal of
Communication, 7, 1920–1948.
Possati, L. M. (2022). Psychoanalyzing artificial intelligence: The case of Replika. AI &
SOCIETY. https://doi.org/10.1007/s00146-021-01379-7
Postman Neil. (1998). Five things we need to know about technological change.
300
PULSE9. (2021a, March 22). ETERN!TY( 이터니티)—I’m Real Official M/V.
https://www.youtube.com/watch?v=XQMg8-Ku3JI
PULSE9. (2021b, August 26). [ETERN!TY] DAIN 이터니티 다인 No Filter( 노필터) MV.
https://www.youtube.com/watch?v=biIVEH7-Rao
PULSE9. (2022, April 7). [ETERNITY] “Paradise ( 파라다이스)” MV Teaser.
https://www.youtube.com/watch?v=N5h_a68DqVs
pureyādo. (2011, September 18). Bācharu aidoru ima mukashi ~ eguchi aimi to haga yui ~
[Virtual idol now past ~Eguchi Aimi and Haga Yui~]. gajetto tsūshin.
https://getnews.jp/archives/141915
Pyman, T. (2020, November 7). CGI models earning developers a fortune by promoting fashion
brands. DailyMail. https://www.dailymail.co.uk/news/article-8923893/Digitally-created-
models-earning-developers-fortune-promoting-fashion-brands.html
Ratan, R. (2019). When Automobiles Are Avacars: A Self-Other-Utility Approach to Cars and
Avatars. International Journal of Communication, 13, 1–19.
Ratan, R. A., Taylor, N., Hogan, J., Kennedy, T., & Williams, D. (2015). Stand by Your Man:
An Examination of Gender Disparity in League of Legends. Games and Culture, 10(5),
438–462. https://doi.org/10.1177/1555412014567228
Ratan, R., & Sah, Y. J. (2015). Leveling up on stereotype threat: The role of avatar
customization and avatar embodiment. Computers in Human Behavior, 50, 367–374.
https://doi.org/10.1016/j.chb.2015.04.010
Reality Check: TINA.org Calls on FTC to Address Virtual Influencers. (2020, June 22). Truth In
Advertising. https://www.truthinadvertising.org/reality-check-tina-callson-ftcto-address-
virtual-influencers/
301
Reeves, B., & Nass, C. (2002). The Media Equation. CSLI Publications.
Reinecke, L., Klimmt, C., Meier, A., Reich, S., Hefner, D., Knop-Huelss, K., Rieger, D., &
Vorderer, P. (2018). Permanently online and permanently connected: Development and
validation of the Online Vigilance Scale. PLOS ONE, 13(10), e0205384.
https://doi.org/10.1371/journal.pone.0205384
RenderHub 3D. (2018, December 14). How To Create Photo-realistic Human Skin Textures.
https://www.youtube.com/watch?v=uGyGH-dPj_8
Richardson, J. (2005). “The Digital Won’t Let Me Go”: Constructions of the Virtual and the Real
in Gorillaz’ “Clint Eastwood”1. Journal of Popular Music Studies, 17, 1–29.
https://doi.org/10.1111/j.1524-2226.2005.00031.x
Richterich, A. (2020). Tracing controversies in hacker communities: Ethical considerations for
internet research. Information, Communication & Society, 23(1), 76–93.
https://doi.org/10.1080/1369118X.2018.1486867
Rivu, R., Zhou, Y., Welsch, R., Mäkelä, V., & Alt, F. (2021). When Friends Become Strangers:
Understanding the Influence of Avatar Gender on Interpersonal Distance in Virtual
Reality. In C. Ardito, R. Lanzilotti, A. Malizia, H. Petrie, A. Piccinno, G. Desolda, & K.
Inkpen (Eds.), Human-Computer Interaction – INTERACT 2021 (Vol. 12936, pp. 234–
250). Springer International Publishing. https://doi.org/10.1007/978-3-030-85607-6_16
Robertson, V. L. D. (2014). Of ponies and men: My Little Pony: Friendship is Magic and the
Brony fandom. International Journal of Cultural Studies, 17(1), 21–37.
https://doi.org/10.1177/1367877912464368
Robinson, B. (2020). Towards an Ontology and Ethics of Virtual Influencers. Australasian
Journal of Information Systems, 24. https://doi.org/10.3127/ajis.v24i0.2807
302
Rogers, E. M. (2003). Diffusion of Innovations (5th edition). Free Press.
Roundtree, C. (2021, April 3). Meet the World’s First AI Robot Rapper Who Sells NFTs. The
Daily Beast. https://www.thedailybeast.com/meet-fn-meka-the-worlds-first-ai-robot-
rapper-who-sells-nfts
“sal~jjak chaeunu? Joshua?” kkonminam aidol yeonseupsaeng 'jeongsejin’ui jeongche" ["li~ttle
bit Eunwoo Cha? Joshua?" Handsome idol trainee ‘Jeong Saejin’s’ identity]. (2021,
August 21). Dispatch. https://www.dispatch.co.kr/2160694
Salisbury, M., & Pooley, J. D. (2017). The #nofilter Self: The Contest for Authenticity among
Social Networking Sites, 2002–2016. Social Sciences, 6(1), 10.
https://doi.org/10.3390/socsci6010010
Sands, S., Campbell, C. L., Plangger, K., & Ferraro, C. (2022). Unreal influence: Leveraging AI
in influencer marketing. European Journal of Marketing, 56(6), 1721–1747.
https://doi.org/10.1108/EJM-12-2019-0949
Savage, S. L. (2015). Just Looking: Tantalization, Lolicon, and Virtual Girls. Visual Culture &
Gender, 10, 37–46.
Sayes, E. (2014). Actor–Network Theory and methodology: Just what does it mean to say that
nonhumans have agency? Social Studies of Science, 44(1), 134–149.
https://doi.org/10.1177/0306312713511867
Schneider, Z. (2021, March 4). DM for Collabs. Medium. https://immerse.news/i-posted-to-
instagram-as-an-aging-robot-and-here-are-some-responses-i-received-9874b6072af6
sdlkfj. (2022, February 10). [Cheopibaeng ttoneun danbal] gwiyeoun imiji 10daehuban~20dae
choban model guin (chobohwanyeong) / gasang inpeullueonseo jejak [[Choppy bangs or
bob] looking for a model in cute image late teens to early twenties (beginners welcome) /
303
virtual influencer creation]. Filmmakers Community.
https://www.filmmakers.co.kr/performerWanted/12264850
Sharkey, A., & Sharkey, N. (2012). Granny and the robots: Ethical issues in robot care for the
elderly. Ethics and Information Technology, 14(1), 27–40.
https://doi.org/10.1007/s10676-010-9234-6
Shaw, A. (2017). Encoding and decoding affordances: Stuart Hall and interactive media
technologies. Media, Culture & Society, 39(4), 592–602.
https://doi.org/10.1177/0163443717692741
Shieber, J. (2019, January 14). More investors are betting on virtual influencers like Lil Miquela.
TechCrunch. https://social.techcrunch.com/2019/01/14/more-investors-are-betting-on-
virtual-influencers-like-lil-miquela/
Shifman, L. (2018). Testimonial rallies and the construction of memetic authenticity. European
Journal of Communication, 33(2), 172–184. https://doi.org/10.1177/0267323118760320
Silverstone, R. (1999). What’s New about New Media?: Introduction. New Media & Society,
1(1), 10–12. https://doi.org/10.1177/1461444899001001002
Simpson, D. (2005, September 8). Gorillaz to tour—In 3D. The Guardian.
https://www.theguardian.com/music/2005/sep/08/gorillaz.popandrock
Sims, C. (2018, May 9). To understand feuds between virtual Instagram models, you need to
understand pro wrestling. The Verge.
https://www.theverge.com/2018/5/9/17332778/instagram-lil-miquela-bermuda-feud-pro-
wrestling
304
Skjuve, M., Følstad, A., Fostervold, K. I., & Brandtzaeg, P. B. (2021). My Chatbot
Companion—A Study of Human-Chatbot Relationships. International Journal of
Human-Computer Studies, 149, 102601. https://doi.org/10.1016/j.ijhcs.2021.102601
Smith, D. (2020). Samsung’s Neon “artificial humans” are confusing everyone. We set the
record straight. CNET. https://www.cnet.com/how-to/samsung-neon-artificial-humans-
are-confusing-everyone-we-set-record-straight/
Smits, T., & Wevers, M. (2021). The agency of computer vision models as optical instruments.
Visual Communication, 1470357221992097. https://doi.org/10.1177/1470357221992097
Sobande, F. (2021). Spectacularized and Branded Digital (Re)presentations of Black People and
Blackness. Television & New Media, 22(2), 131–146.
https://doi.org/10.1177/1527476420983745
Sone, Y. (2017). Hatsune Miku, Virtual Machine-Woman. In Y. Sone (Ed.), Japanese Robot
Culture: Performance, Imagination, and Modernity (pp. 139–166). Palgrave Macmillan
US. https://doi.org/10.1057/978-1-137-52527-7_6
Song, H. (1999, December 28). Saibeoagi, 2000nyeon 1wol 1il tansaeng [Cyberbaby, born on
January 1st, 2000]. KBS News. https://news.kbs.co.kr/news/view.do?ncd=3812439
Song, S. (2019, December 14). CGI Influencer Lil Miquela Criticized For “Sexual Assault”
Vlog. PAPER. https://www.papermag.com/lil-miquela-sexual-assault-vlog-
2641593301.html
Spangler, T. (2020, May 6). Miquela, the Uncanny CGI Virtual Influencer, Signs With CAA
(EXCLUSIVE). Variety. https://variety.com/2020/digital/news/miquela-virtual-
influencer-signs-caa-1234599368/
305
Sparrow, R., & Sparrow, L. (2006). In the hands of machines? The future of aged care. Minds
and Machines, 16(2), 141–161. https://doi.org/10.1007/s11023-006-9030-6
Stahl, M. (2011). The Synthespian’s Animated Prehistory: The Monkees, The Archies, Don
Kirshner, and the Politics of “Virtual Labor.” Television & New Media, 12(1), 3–22.
https://doi.org/10.1177/1527476409357641
Stanley, T. L. (2020, April 10). Why the WHO Turned to a Virtual Influencer to Fight COVID-
19. Adweek. https://www.adweek.com/creativity/why-world-health-organization-virtual-
influencer-knox-frost-covid-19-tips/
Star, S. L. (2010). This is Not a Boundary Object: Reflections on the Origin of a Concept.
Science, Technology, & Human Values, 35(5), 601–617.
https://doi.org/10.1177/0162243910377624
Star, S. L., & Griesemer, J. (1989). Institutional Ecology, ‘Translations’ and Boundary Objects:
Amateurs and Professionals in Berkeley’s Museum of Vertebrate Zoology, 1907-39.
Social Studies of Science, 19, 387–420.
Star, S. L., & Strauss, A. (1999). Layers of Silence, Arenas of Voice: The Ecology of Visible and
Invisible Work. Computer Supported Cooperative Work (CSCW), 8(1), 9–30.
https://doi.org/10.1023/A:1008651105359
Stardust, Z. (2018). Safe for Work: Feminist Porn, Corporate Regulation and Community
Standards. In C. Dale & R. Overell (Eds.), Orienting Feminism: Media, Activism and
Cultural Representation (pp. 155–179). Springer International Publishing.
https://doi.org/10.1007/978-3-319-70660-3_9
Stark, A. (2018). Virtual Pop: Gender, Ethnicity, and Identity in Virtual Bands and Vocaloid.
306
Stassen, M. (2021, April 1). This robot rapper has 9m followers on TikTok. The company that
created him thinks traditional A&R is ‘inefficient and unreliable.’ Music Business
Worldwide. https://www.musicbusinessworldwide.com/this-robot-rapper-has-9-million-
followers-on-tiktok-his-creator-thinks-traditional-ar-is-inefficient-and-unreliable/
Steinhart, E. (2014). Your Digital Afterlives: Computational Theories of Life after Death.
Springer.
Stephenson, N. (1992). Snow Crash. Roc (Penguin Books).
Stewart, R. S. (2019). Is Feminist Porn Possible? Sexuality & Culture, 23(1), 254–270.
https://doi.org/10.1007/s12119-018-9553-z
Suchman, L. (2009, March). Agencies in technology design: Feminist reconfigurations.
Proceedings of 5th European Symposium on Gender & ICT, Digital Cultures:
Participation–Empowerment–Diversity.
Suri, S. (2019, May 16). Bella Hadid kisses CGI “influencer” Lil Miquela in Calvin Klein ad.
DailyMail. https://www.dailymail.co.uk/tvshowbiz/article-7038063/Bella-Hadid-shares-
kiss-fictional-CGI-influencer-Lil-Miquela-Calvin-Klein-ad.html
Tayenaka, T. (2020, July 31). CGI-Created Virtual Influencers Are the New Trend in Social
Media Marketing. Entrepreneur. https://www.entrepreneur.com/article/352937
Taylor, S. H., & Bazarova, N. N. (2021). Always Available, Always Attached: A Relational
Perspective on the Effects of Mobile Phones and Social Media on Subjective Well-Being.
Journal of Computer-Mediated Communication, 26(4), 187–206.
https://doi.org/10.1093/jcmc/zmab004
Teh, C. (2021, August 12). China is tempting customers with its flawless AI idols—Virtual
influencers who don’t gain weight, never age, and keep their computer-generated noses
307
out of controversy. Insider. https://www.insider.com/chinas-flawless-ai-influencers-the-
hot-new-queens-of-advertising-2021-8
Terai Yuki. (2021). In Wikipedia.
https://ja.wikipedia.org/w/index.php?title=%E3%83%86%E3%83%A9%E3%82%A4%E
3%83%A6%E3%82%AD&oldid=82691162
Teurlings, J. (2013). Unblackboxing production. What media studies can learn from actor-
network theory. In J. Teurlings (Ed.), After the Break (pp. 101–116). Amsterdam
University Press. https://doi.org/10.1515/9789048518678-007
The Diigitals. (n.d.). Thediigitals. Retrieved June 21, 2022, from https://www.thediigitals.com
Thomas, V. L., & Fowler, K. (2020). Close Encounters of the AI Kind: Use of AI Influencers As
Brand Endorsers. Journal of Advertising, 0(0), 1–15.
https://doi.org/10.1080/00913367.2020.1810595
Tiffany, K. (2019, June 3). Lil Miquela and the virtual influencer hype, explained. Vox.
https://www.vox.com/the-goods/2019/6/3/18647626/instagram-virtual-influencers-lil-
miquela-ai-startups
Townsend, L., & Wallace, C. (2016). Social media research: A guide to ethics.
https://www.bolton.ac.uk/assets/Uploads/Social-media-ethics-study-Aberdeen-2018.pdf
Travers, C. (2020a, February 23). Who is Virtual Influencer and “Dream Girl” Shy Yume?
VirtualHumans.Org. https://www.virtualhumans.org/article/who-is-virtual-influencer-
and-dream-girl-shy-yume
Travers, C. (2020b, April 22). Who is Virtual Influencer and AR Sneakerhead FNMEKA?
VirtualHumans.Org. https://www.virtualhumans.org/article/who-is-virtual-influencer-
and-ar-sneakerhead-fnmeka
308
Travers, C. (2022, March 24). Instagram Has Verified 35 Virtual Influencers.
VirtualHumans.Org. https://www.virtualhumans.org/article/instagram-has-verified-35-
virtual-influencers
Trepany, C. (2019, October 16). The robot invasion has begun: Meet the computer-generated
influencers taking over Instagram. USA Today.
https://www.usatoday.com/story/life/2019/10/16/cgi-influencers-blur-line-between-
reality-and-fantasy-instagram-advertising/3790471002/
TU Kyoko Date DK96 Archive. (n.d.). Random-Abstract.Com. Retrieved June 21, 2022, from
http://www.random-abstract.com/kyokodate/
Turkle, S. (1995). Life on the screen: Identity in the age of the Internet. Touchstone.
US Department of Justice. (2010). The National Strategy for Child Exploitation Prevention and
Interdiction: A Report to Congress.
http://toolkit.heatwatch.org/policy/files/USDOJ_2010__National_Strategy_to_Prevent_C
SEC.pdf
VirtualHumans.org. (n.d.). VirtualHumans.Org. Retrieved June 21, 2022, from
https://www.virtualhumans.org/
vom Lehn, D., & Hitzler, R. (2015). Phenomenology-Based Ethnography: Introduction to the
Special Issue. Journal of Contemporary Ethnography, 44(5), 539–543.
https://doi.org/10.1177/0891241615595436
Wada, K., & Shibata, T. (2007). Robot Therapy in a Care House—Change of Relationship
among the Residents and Seal Robot during a 2-month Long Study. RO-MAN 2007 - The
16th IEEE International Symposium on Robot and Human Interactive Communication,
107–112. https://doi.org/10.1109/ROMAN.2007.4415062
309
Webster, A. (2020, August 31). Ikea turned a virtual influencer into a physical installation. The
Verge. https://www.theverge.com/2020/8/31/21408626/ikea-tokyo-imma-virtual-
influencer
Whateley, D. (2021, March 17). Influencers are rushing to make money on NFTs using virtual
avatars, digital collectibles, and one-off art. Business Insider.
https://www.businessinsider.com/how-influencers-are-making-money-on-nfts-while-
market-hot-2021-3
Whateley, D., & Bradley, S. (2020, July 20). Virtual influencer startups say they’ve seen a spike
in interest from brands as the pandemic has shut down many human photo shoots.
Business Insider. https://www.businessinsider.com/virtual-influencers-like-lil-miquela-
see-interest-increase-in-pandemic-2020-7
Whitson, J. R. (2018). Voodoo software and boundary objects in game development: How
developers collaborate and conflict with game engines and art tools. New Media &
Society, 20(7), 2315–2332. https://doi.org/10.1177/1461444817715020
Who is Cade Harper? @cadeharper, explained. (n.d.). VirtualHumans.Org. Retrieved June 22,
2022, from https://www.virtualhumans.org/human/cade-harper
will2power. (2018, October). I created Terai Yuki for Genesis 8. Daz 3D Forums.
https://www.daz3d.com/forums/discussion/287171/i-created-terai-yuki-for-genesis-8
Williams, D. (2003). The video game lightning rod. Information, Communication & Society,
6(4), 523–550.
Williams, D., & Kim, D. O. (2019). Third places in the ether around us: Layers on the real world.
In J. Dolley & C. Bosman (Eds.), Rethinking Third Places: Informal Public Spaces and
310
Community Building. Edgar Elgar Publishing. https://www.e-elgar.com/shop/rethinking-
third-places
Williams, R. (2014). The Technology and the Society (1974). In J. McGuigan (Ed.), Raymond
Williams on Culture & Society: Essential Writings (pp. 139–160). SAGE Publications
Ltd. https://doi.org/10.4135/9781473914766
Wills, K. (2019, March 7). Meet Lil Miquela: The AI influencer taking over Instagram. Evening
Standard. https://www.standard.co.uk/lifestyle/esmagazine/lil-miquela-ai-influencer-
instagram-a4084566.html
Winner, L. (1988). Do Artifacts Have Politics? In The Whale and the Reactor. University of
Chicago Press.
Winner, L. (1993). Upon Opening the Black Box and Finding It Empty: Social Constructivism
and the Philosophy of Technology. Science, Technology, & Human Values, 18(3), 362–
378.
WIRED Staff. (2005, July 1). Keeping It (Un)real. Wired.
https://www.wired.com/2005/07/gorillaz-2/
Wise, D. (2020, November 28). The Life and Death of the Instagram Influencer Who Never
Was. Variety. https://variety.com/2020/digital/news/ziv-schneider-idfa-doclab-sylvia-
1234841619/
Wissinger, E. (2009). Modeling Consumption: Fashion modeling work in contemporary society.
Journal of Consumer Culture, 9(2), 273–296. https://doi.org/10.1177/1469540509104377
WKtokyo. (2020, September 12). IKEA Harajuku with imma: The Recap EN.
https://www.youtube.com/watch?v=s586r8PMlSk
Wolfe, C. (2010). What is Posthumanism? U of Minnesota Press.
311
World Entertainment News Network. (2019, May 20). Calvin Klein apologizes for “queer-
baiting” ad with Bella Hadid. Canoe. https://canoe.com/life/fashion-beauty/calvin-klein-
apologizes-for-queer-baiting-ad-with-bella-hadid
xHamster. (2019, June 20). Meet Shy, the woman of your dreams! XHamster.
https://xhamster.com/blog/posts/990663
Xie, T., & Pentina, I. (2022). Attachment Theory as a Framework to Understand Relationships
with Social Chatbots: A Case Study of Replika. Hawaii International Conference on
System Sciences. https://doi.org/10.24251/HICSS.2022.258
Yee, N., & Bailenson, J. (2007). The Proteus Effect: The Effect of Transformed Self-
Representation on Behavior. Human Communication Research, 33(3), 271–290.
https://doi.org/10.1111/j.1468-2958.2007.00299.x
Yee, N., Bailenson, J. N., & Ducheneaut, N. (2009). The Proteus Effect: Implications of
Transformed Digital Self-Representation on Online and Offline Behavior.
Communication Research, 36(2), 285–312. https://doi.org/10.1177/0093650208330254
Yee, N., Ducheneaut, N., Yao, M., & Nelson, L. (2011). Do men heal more when in drag?:
Conflicting identity cues between user and avatar. Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems, 773–776.
https://doi.org/10.1145/1978942.1979054
Yesnet Co.,LTD. (2002, June 1). Saida kaerikteo meil bonaegi [CYDA send character email].
https://web.archive.org/web/20020601075114/http://www.cyda.co.kr/
Yoo, H. (2021, July 20). [Teukjingju]jaieonteuseutep, beochyueol hyumeoneuro metabeoseu
seonjeom “gangse” [[characteristic stocks] Giant Step takes charge of Metaverse with
312
virtual humans ’strong’]. Asiagyeongje.
https://www.asiae.co.kr/article/2021072009400308434
Yoo, J. (2021, July 13). [Wikeulliripoteu 38ho] “neibeoweptun woryoil sunwi” 6wi girokan
weptunui jeongche, almyeon alsurok nollamneyo [[Weekly report issue 38] “Naver
Webtoon monday ranking” 6th webtoon’s identity, the more surprising the more you
know]. Wikitree. https://www.wikitree.co.kr/articles/668430
Ytre-Arne, B., & Moe, H. (2021). Folk theories of algorithms: Understanding digital irritation.
Media, Culture & Society, 43(5), 807–824. https://doi.org/10.1177/0163443720972314
Yurieff, K. (2018, June 25). Instagram star isn’t what she seems. But brands are buying in. CNN
Business. https://money.cnn.com/2018/06/25/technology/lil-miquela-social-media-
influencer-cgi/index.html
Zaborowski, R. (2016). Hatsune Miku and Japanese Virtual Idols. The Oxford Handbook of
Music and Virtuality, 111–128. https://doi.org/10.1093/oxfordhb/9780199321285.013.7
Zhang, Y. (Gavin), Dang, Y. (Mandy), Brown, S. A., & Chen, H. (2017). Investigating the
impacts of avatar gender, avatar age, and region theme on avatar physical activity in the
virtual world. Computers in Human Behavior, 68, 378–387.
https://doi.org/10.1016/j.chb.2016.11.052
313
Appendix A: List of Virtual Influencers
Table 5
List of Virtual Influencers (accessed on March 31, 2022)
# Name @handle Last active* Verified** followers following posts
1 Lu Do Magalu @magazineluiza
1 5,800,000 229 2,307
2 Miquela Sousa @Lilmiquela
1 3,000,000 1,912 1,182
3 Knox Frost @knoxfrost 20210627 1 673,000 516 148
4 Thalasya Pov @thalasya_ 20211001 478,000 1 225
5 Imma @Imma.gram
356,000 334 630
6 Leya Love @leyalovenature
336,000 639 602
7 Bermuda @Bermudaisbae 20200921 1 277,000 534 268
8 Shudu @Shudu.gram
1 225,000 5,030 102
9 Aliza Rex @Alizarexx 20211111 198,000 316 160
10 Koffi @koffi.gram 20210611 177,000 81 21
11 Blawko @Blawko22 20200917 1 145,000 1,025 162
12 Liam Nikuro @Liam_nikuro
131,000 28 109
13 Rozy Oh @rozy.gram
1 123,000 5,824 245
14 Maria @soymar.ia
1 101,000 34 89
15 Amara @amara_gram 83,800 19 82
16 Ruby9100m @ruby9100m 81,300 1,899 994
17 Binxie @Itsbinxie 20210610 65,900 99 223
18 Zinn @plusticboy
53,400 83 242
19 Ion Göttlich @iongottlich
1 53,400 473 304
20 Emry Paige @_emry.paige_ 51,500 606 330
21 Serah @serahreikka 51,000 140 217
22 Lila Ziyagil @lilaziyagil
46,400 119 147
23 Esther Olofosson @esther.olofsson
44,700 1,041 415
24 Jedy Vales @jedyvales
39,900 34 212
25 Ria @ria_ria_tokyo
38,400 54 55
26 Pol @pol.songs 2021077 32,100 863 88
27 te'resa @virtual_teresa 30,700 325 358
28 Zoe Dvir @zoedvir 20210726 27,200 3,197 172
29 Ninocence @ninocence
27,100 2,850 408
30 AI Angelica @ai_angelica 20210430 26,600 - 120
31 Essential @essentialxo 20210401 25,700 3 33
32 Aya Stellar @aya.stellar
25,600 382 413
33 Liv Os @soylivos 20200202 19,300 62 27
34 Kenna @thisis.kenna 20200327 17,200 70 18
35 Maya @maya_cgi 20191222 17,100 334 47
36 Rae @here.is.rae 16,300 141 272
37 Aliona Pole @aliona_pole
16,000 2 115
38 Reah Keem @reahkeem 20200920 15,300 - 52
39 Ailynn @ai_ailynn
14,000 10 77
40 Lil Wavi @lil_wavi
14,000 528 18
41 Ezmusgita @ezmusgita 20211211 13,500 4,529 99
42 Sasha SHE @sasha_____she 20210907 13,400 95 103
314
43 J-Nice
Prempelli 🇬🇬🇬🇬
@opalslutuniverse 20211112 12,200 232 51
44 Aoi Prism @aoiprism 20201022 12,000 142 140
45 Ruby @ruby.economics 20200707 11,700 85 174
46 Ivaany @ivaany.h
11,700 1,004 227
47 Rui @ruuui_li
11,500 10 78
48 Ayayi @ayayi.iiii
11,100 11 91
49 Lisa Gen @lisa_gen_lg
11,000 38 336
50 Phoenix
McEwan
@phoenixmcewan 20210528 10,500 260 118
51 SUA @sua_to_z
10,400 952 83
52 Yua Han @_hanyua
10,300 27 36
53 Ella Stoller @itsellastoller 20210531 10,000 318 93
54 Dagny @dagny.gram 20210618 9,251 49 19
55 Pippa Pei @pippapei 20210412 8,575 55 95
56 Lewis Newman @lewis_hiro_newm
an
20201018 8,558 2 90
57 Bejby Blue @bejby.blue
8,529 86 182
58 Alice Mikoni @alicemikoni 20200318 8,409 88 18
59 Kemi @kemiunruly 20210801 7,668 604 22
60 Ella @ella.imagination
7,859 28 184
61 Maya @mayaaa.gram 20200824 7,836 17 47
62 Uca @xx_uca_xx 20200806 6,036 14 56
63 Sina @sensualsina
4,885 392 59
64 Sui-D @sui.daily
4,676 77 556
65 Candice Klubb @candiceklubb 20210215 4,515 27 8
66 Ana Tobor @ana.tobor 20211231 4,477 381 99
67 Hannah Stein @hey_mrs.stone
4,438 355 130
68 Amanda @amanda_bims 20220124 4,411 444 189
69 Ava Azura @avaazura 20211120 4,367 1,671 8
70 Meme @meme.konichiwa
4,101 398 118
71 Yona @y0000000000na 20210903 4,052 146 38
72 Lena @iam.lena.official
3,953 90 488
73 Brittany Rose @brittany_rose_offi
cial
3,865 2,079 59
74 polishboy08 @polishboy08 20210525 3,863 1 151
75 Yasmine Varma @yasminevarma 20210528 3,800 190 32
76 Aya @cahaya.gram 20200114 3,604 132 48
77 Kim Zulu @kimzulu_
3,489 695 147
78 Brenn @brenn.gram 20200910 3,345 49 15
79 Mikirah Muse @mikirahmuse
3,250 1,635 56
80 Ellie @ell_ii_e
3,084 184 73
81 Asu @asuu.gram 20220113 2,650 9 37
82 Aba Wils @abawils
2,539 3,422 198
83 Ono @ono.gram 20201213 2,146 312 234
84 Kira @kira.faux 20200402 2,090 - 39
85 Viena Numera @viena_numera
1,853 400 52
86 Milla @milla_sakurai 20200406 1,707 26 41
87 Sara Kosmos @sara.kosmos 20200807 1,643 149 59
88 Geneva Numera @geneva_numera
1,595 779 48
89 Veronica @thereal.veronica
1,505 22 89
315
90 Dana Beauvoir @danabeauvior 20210121 1,291 538 188
91 Aba Wils @abawils
1,263 2,833 89
92 Zeevaah @zeevaah 20210613 1,262 168 106
93 KAROLIN XS @karolinxs 20200515 1,196 52 39
94 Robin Abree @robinabree 20211014 1,178 253 110
95 Asya Strike @asyastrike 20190918 1,077 5 12
96 Nikki Aura @nikki.a.i
1,034 596 22
97 Hayley @hayleysoblessed 20210810 994 1,846 97
98 Faith Fawkes @faithfawkes 20210406 927 137 36
99 Ary @aryuforreal 20210404 559 256 65
100 Lérraje @lerraje_
507 182 46
101 Naïry @naairy.h 20201017 456 115 36
102 Wooju @woo.ju.like
323 3 7
103 Akuchi @digi.akuchi 20210127 297 318 32
104 Jaide @jaide.dao
212 373 38
105 Mia Locklace @mialocklace 95 56 325
Notes. The list has been created using this dissertation’s definition of virtual influencers,
however with some flexibility (e.g., some digital twins are included). The list is not
comprehensive.
*The order is year, month, date. For example, 20210627 means June 27, 2021. Date is only
indicated for virtual influencers who have been dormant for more than two months.
**Virtual influencers who have earned the verified badge from Instagram, awarded to public
figures. 1 = verified.
Abstract (if available)
Abstract
This dissertation investigated how humanness and agency are being negotiated by various human and nonhuman actors in the sociotechnical phenomenon of virtual influencers, defined as photorealistic CGI “humans” who are, or aspire to be, social media influencers. “Virtually human” technological nonhumans and technology-augmented humans are increasingly co-inhabiting social environments. This undermines existing research, industry, and regulatory practices that stem from anthropocentric assumptions, calling for renewed attention on the supposed ontological divide between humans and nonhumans. Interlaced with multiple contextual and performative ambiguities, virtual influencers are compelling cases of this contextual shift. The dissertation drew on findings from phenomenological ethnographic research that incorporated observant participation of various digital sites, semi-structured interviews with global human participants, and archived digital data, such as social media posts, news, podcasts, and derivative creative contents. Despite commonly being narrated as novel, analysis of virtual influencers’ (pre)history and their variegated definitions pointed at many continuing patterns of power, intermingled with discursive practices around the future. Multi-directional approach to networks of virtual influencers demonstrated their socially meaningful, “real” relational effects. It also showed how their virtual bodies can strategically connect and disconnect from directly or discursively connected physical bodies to escape accountability. Inclusion of nonhuman actors in networks suggested that despite the phenomenon’s supposed liberating immateriality, virtual influencers come into being through networks of bodies, things, and places, some of which were made hidden through historical structures of oppressive exploitation. I propose meta-authenticity, defined as the desire or achievement of authenticity in practices of extreme inauthenticity, as an accountability-motivated concept for critically assessing boundary-crossing media’s “real” consequences without losing experiential nuances. I argue for a shared sense of responsibility that fully recognizes each actor’s capacity to make difference, including our own. Guidelines for ethical engagement and implications on the future of human communication culture are also discussed.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Self-presence: body, emotion, and identity extension into the virtual self
PDF
Living with the most humanlike nonhuman: understanding human-AI interactions in different social contexts
PDF
Gathering understanding: players, systems and developers in Magic: the gathering
PDF
"Wonders and wishes": contexts and influences of Black millennials' childhood television viewership
PDF
Upvoting the news: breaking news aggregation, crowd collaboration, and algorithm-driven attention on reddit.com
PDF
The search for authenticity in influencer marketing
PDF
Sunsetting: platform closure and the construction of digital cultural loss
PDF
Gamification + HCI + CMC: effects of persuasive video games on consumers’ mental and physical health
PDF
Towards social virtual listeners: computational models of human nonverbal behaviors
PDF
Socially-informed content analysis of online human behavior
PDF
Artificial intelligence for low resource communities: Influence maximization in an uncertain world
PDF
Home screen home: how parents of children with disabilities navigate family media use
PDF
Political coming-of-age in the era of WeChat: understanding the ethnic media ecosystem and group politics of first-generation Chinese Americans
PDF
A framework for research in human-agent negotiation
PDF
Social value orientation, social influence and creativity in crowdsourced idea generation
PDF
Benefits of social networking in online social support groups
PDF
These are their stories: two decades of Showrunner production, content, and context in Law & Order: SVU
PDF
Communicating organizational knowledge in a sociomaterial network: the influences of communication load, legitimacy, and credibility on health care best-practice communication
PDF
Motivation for participating in virtual religious communities: developing social capital
PDF
The power of social media narratives in raising mental health awareness for anti-stigma campaigns
Asset Metadata
Creator
Kim, Do Own (Donna)
(author)
Core Title
Virtually human? Negotiation of (non)humanness and agency in the sociotechnical assemblage of virtual influencers
School
Annenberg School for Communication
Degree
Doctor of Philosophy
Degree Program
Communication
Degree Conferral Date
2022-08
Publication Date
07/24/2024
Defense Date
04/28/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
artificial intelligence,authenticity,CGI influencer,computer-mediated communication,human-machine communication,nonhuman,OAI-PMH Harvest,parasocial,representation,self-presentation,social media,virtual being,virtual human,virtual influencer
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Jenkins, Henry (
committee chair
), Williams, Dmitri (
committee chair
), Ananny, Mike (
committee member
)
Creator Email
doownkim@usc.edu,kimdoown@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC111375336
Unique identifier
UC111375336
Legacy Identifier
etd-KimDoOwnDo-10958
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Kim, Do Own (Donna)
Type
texts
Source
20220728-usctheses-batch-962
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
artificial intelligence
authenticity
CGI influencer
computer-mediated communication
human-machine communication
nonhuman
parasocial
representation
self-presentation
social media
virtual being
virtual human
virtual influencer