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
/
Preparing for the future of work: exploring worker perceptions of the impact of automation
(USC Thesis Other)
Preparing for the future of work: exploring worker perceptions of the impact of automation
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Preparing for the Future of Work:
Exploring Worker Perceptions of the Impact of Automation
by
Daniel L. Ahlgren
Rossier School of Education
University of Southern California
A dissertation submitted to the faculty
in partial fulfillment of the requirements for the degree of
Doctor of Education
August 2021
© Copyright by Daniel L. Ahlgren 2021
All Rights Reserved
The Committee for Daniel L. Ahlgren certifies the approval of this Dissertation
Scott Lever
Jennifer Phillips
Helena Seli, Committee Chair
Rossier School of Education
University of Southern California
2021
iv
Abstract
Automation of human tasks has begun to fuel a massive workforce transition labeled the “Fourth
Industrial Revolution”, because it will fundamentally change the dynamic of how people work,
and the types of jobs they will do (Vermeulen et al., 2018). The purpose of the study was to
explore individual worker demographics and perspectives about the potential impact of
automation and how those perspectives relate to the individual’s willingness to accept change,
and accept and use automation in order to assess potential opportunities for better planned
organizational change. The study employed a survey designed to produce results of workers’
level of understanding of the impact of automation, their willingness to accept change, and
accept and use automation. The study resulted in multiple correlational relationships indicating
the importance of workers having an understanding of automation, how it will impact them in the
future, and how organizations can better prepare workers for changes that come with automation.
Based on this study, the following is recommended, to (a) use educational interventions to
support knowledge and skill development to improve future acceptance and use of automation;
(b) support acceptance and use automation through enhancing perceived value of automation and
worker efficacy; and (c) develop and enhance organizational systems and processes to support
the acceptance and use of automation.
Keywords: the future of work, automation, intelligent automation, human augmentation,
technology, up-skilling, re-skilling, digital dexterity
v
Dedication
To my parents who instilled in me the importance of faith and education from a young age. More
specifically, this achievement is dedicated to my mother, who is a master educator, and has
dedicated her life to educating others. She is the embodiment of a lifelong learner.
vi
Acknowledgements
Thank you to my wife Scherrie, who has supported me through many pursuits, over many years,
and has been a source of constant encouragement through all of them. To my sisters and
brothers, who have provided just the right amount of humor and encouragement at just the right
times. To Dr. Seli, my Dissertation Chair, thank you for your openness to new concepts,
believing in me, and helping me get to the finish line. To my other family members, for believing
in my success and supporting my continuous improvement. And to my dear friends, colleagues,
mentors, teachers, and professors, without you I would not have had the strength to come this far.
As one much wiser than I once said:
Two are better than one, because they have a good return for their labor: If either of them
falls down, one can help the other up. But pity anyone who falls and has no one to help
them up. Also, if two lie down together, they will keep warm. But how can one keep
warm alone? Though one may be overpowered, two can defend themselves. A cord of
three strands is not quickly broken. (Holy Bible, New International Version, 1978/2011,
Ecclesiastes 4:9–12)
vii
Table of Contents
Abstract .......................................................................................................................................... iv
Dedication ....................................................................................................................................... v
Acknowledgements ........................................................................................................................ vi
List of Tables .................................................................................................................................. x
List of Figures ............................................................................................................................... xii
Chapter One: Overview of the Study .............................................................................................. 1
Related Literature................................................................................................................ 2
Study Context...................................................................................................................... 4
Purpose of the Study and Research Questions .................................................................... 4
Importance of the Study ...................................................................................................... 5
Overview of Theoretical Framework and Methodology .................................................... 5
Definitions of Terms ........................................................................................................... 7
Organization of the Dissertation ......................................................................................... 8
Chapter Two: Literature Review .................................................................................................... 9
Historical Progression of Automation ................................................................................ 9
Workforce Predictions for Automation ............................................................................ 15
Benefits of Adopting Automation ..................................................................................... 28
Pressures and Challenges for Adopting Automation ........................................................ 32
Theoretical and Conceptual Framework ........................................................................... 44
Chapter Two Summary ..................................................................................................... 51
Chapter Three: Methodology ........................................................................................................ 52
Research Questions ........................................................................................................... 52
Overview of Methodology ................................................................................................ 53
The Researcher.................................................................................................................. 53
viii
Data Source: Survey ......................................................................................................... 54
Data Collection Procedures............................................................................................... 57
Ethics................................................................................................................................. 61
Chapter Four: Results ................................................................................................................... 63
Quantitative Survey Overview .......................................................................................... 63
Study Participants ............................................................................................................. 65
Results Research Question One: Are There Differences in an Individual Worker’s
Perception and Understanding of Automation Based on Their Demographic
Background? ............................................................................................................... 76
Results Research Question Two: Are There Differences in an Individual Worker’s Level
of Acceptance of Change and Acceptance and Use of Automation Based on Their
Demographic Background? ........................................................................................ 80
Results Research Question Three: What Are the Perceptions of Individual Workers
Related to Automation? .............................................................................................. 82
Results Research Question Four: Which Tasks and What Percentage of One’s Job Does
an Individual Worker Perceive can be Automated? ................................................... 95
Results Research Question Five: Is There a Relationship Between an Individual Worker’s
Perception and Understanding of the Impact of Automation, Level of Acceptance of
Change and Level of Acceptance and Use of Automation? ..................................... 101
Chapter Four Summary ................................................................................................... 106
Chapter Five: Recommendations and Discussion....................................................................... 109
Recommendations for Practice ....................................................................................... 109
Limitations and Delimitations......................................................................................... 130
Recommendations for Future Research .......................................................................... 131
Implications for Equity: Focus on Ethics and Balancing Profit, Efficiency, and Social
Good to Ensure the Most Vulnerable Are Considered as a Part of the Automation
Strategy ..................................................................................................................... 134
Conclusion ...................................................................................................................... 136
References ....................................................................................................................... 138
Appendix A: Survey Protocol and Items .................................................................................... 151
ix
Appendix B: Survey Protocol Crosswalk ................................................................................... 166
Appendix C: 2017 U.S. North American Industrial Classification System Descriptions .......... 167
x
List of Tables
Table 1: Theoretical Framework Survey Item Categories and Subcategories……………….... 56
Table 2: Data Analysis Approach by Research Question……………………………………... 59
Table 3: General Demographic Data and Frequencies………………………………………… 68
Table 4: Demographic Employment Data and Frequencies…………………………………… 71
Table 5: Self-Described Occupations…………………………………………………………. 73
Table 6: General Education Data and Frequencies……………………………………………. 74
Table 7: Self-Described Degree Data………………………………………………………….. 75
Table 8: Impact of Automation, Career Changes, and Skills Changes ANOVA Results…..…. 79
Table 9: Acceptance of Change (AC) and Acceptance and Use of Automation (AUA)
ANOVA Results……………………………………………………………………….. 81
Table 10: Personal Exposure to Prior Automation…………………………………………….. 83
Table 11: Behavioral Intent to Improve the Workplace Through Automation………………... 84
Table 12: Understanding of Impact of Automation (UIA), Jobs and Tasks Descriptive Data… 86
Table 13: Perception of Career Changes (PCC), Role, Job, and Career Descriptive Data……. 87
Table 14: Perception of Skills Changes (PSC), Skills, Education, and Training Descriptive
Data…………………………………………………………………………………….. 89
Table 15: Items From Acceptance and Use of Automation (AUA) Descriptive Data………… 91
Table 16: Perception of Responsibility for Providing Training on New Automation………… 93
Table 17: Percentage Expecting to Have Automation Implemented and to Receive Training... 94
Table 18: Percent of Tasks or Duties Deemed Replaceable or Already Replaced……………. 97
Table 19: Self-Described Results of Daily Tasks or Duties Deemed Replaceable……………. 99
Table 20: Self-Described Responses of Types of Tasks or Duties Deemed Replaceable……... 100
Table 21: Table of Correlations and Significant Relationships………………………………... 105
Appendix B: Survey Protocol Crosswalk……………………………………………………… 166
xi
Appendix C: 2017 U.S. North American Industrial Classification System Descriptions……... 167
xii
List of Figures
Figure 1: Conceptual Framework………………………………………………………………..50
1
Chapter One: Overview of the Study
Automation of human tasks has begun to fuel a massive workforce transition labeled the
“Fourth Industrial Revolution,” because it will fundamentally change how humans and
technology interact (Vermeulen et al., 2018). Research has indicated that the massive shift will
encompass a broad range of technologies such as artificial intelligence, data science, quantum
computing, and robotics and will change the dynamic of how people work, and the types of jobs
they will do (Vermeulen et al., 2018). As a part of the coming changes, the global management
consulting firm, McKinsey, predicts that as many as 400 to 800 million people (worldwide)
could be displaced as a result of automation, and 75 to 375 million people (worldwide) will have
to learn new skills and change occupations (Manyika et al., 2017a). A 2017 study on the
susceptibility of jobs to computerization found that over the next 10 to 20 years, 47% of total
U.S. employment could be automated (Frey & Osborne). Brookings, another leading research
firm, estimates 25%, 36%, and 39% of U.S. job types over the next few decades will be exposed
to high, medium, and low automation potential, respectively, with nearly 70% of tasks facing
substitution (Muro et al., 2019). Another McKinsey report predicts that advances in technology
supporting automation could affect most occupations, and that 30% of activities performed by
60% of all occupations could be automated (Manyika et al., 2017b).
As automation ushers in this “Fourth Industrial Revolution,” organizations will be faced
with complex challenges in identifying critical gaps in worker competencies, how to influence
the adoption of automation, and understanding better ways for evaluating how people and
technology will work together (Hawksworth et al., 2019; Poitevin, 2018; Vermeulen et al.,
2018). The concept of automation now extends beyond traditional thoughts about robots
replacing humans on factory assembly lines to places such as doctor’s and law offices.
2
Automation will affect both blue and white-collar jobs, although not necessarily in a
proportionate way for men and women, or for different races (Borry & Getha-Taylor, 2018;
Lent, 2018).
Related Literature
The study seeks to analyze is how individual worker backgrounds as well as their
perceptions about the potential impact of automation relate to their acceptance of change and
acceptance and use of automation. Assessing the potential impact of automation will require
thinking about internal and external pressures, various theories on acceptance of change and
technology acceptance, and understanding how individuals perceive automation. Predicting and
planning for the full implications of adopting automation will be difficult and will require leaders
to think holistically about the required change within their organizations (Makridakis, 2017).
In 1965, Gordon Moore predicted computing power would grow exponentially as
integrated circuits become smaller and less expensive, doubling in capacity and becoming twice
as cheap, every two years (Moore, 1998). As computing power, and thereby automation has
improved, organizations have sought adoption because of the swath of potential benefits such as
operational efficiency, standardization, or financial savings (Borry & Getha-Taylor, 2018). By
some estimates, 91% of organizations are already using some form of automation (Chao et al.,
2018) and by 2022, 20% of workers may rely on technology to perform most of their job (Mok,
2018). While this may not seem initially concerning, automation and artificial intelligence (AI)
(i.e., intelligent automation) technologies have been rapidly replacing many of tasks that were
traditionally handled by people, including both routine and non-routine functions (Kim et al.,
2017; Lindborg, 2017).
3
Rather than having clarity and understanding around how automation will impact jobs in
the future, there is still considerable disagreement and speculation among researchers,
technologists, and economists over how secure the current job market is, and what the future will
hold (Lent, 2018). According to Lent (2018), one area of disagreement is about the strength of
the job market. This disagreement is primarily due to the lack of data for the number of jobs that
have already been replaced by automation, as baby boomers retire, and businesses continue their
recovery from the last U.S. recession (Lent, 2018). Critics have also argued that large, sweeping
predictions of job losses, lack consideration for the varying levels of risk associated with specific
tasks and the likelihood of their replacement by machines (Arntz et al., 2017).
While there seems to be a general lack of consensus about the full effect automation will
have on individual workers, researchers have agreed that predicting the broad implications of the
workforce transformation will be challenging (Makridakis, 2017). Gartner, a leading IT research
and advisory company, predicts that there will be three areas of traditional work that will thrive
through the transition including (a) automated intelligence in routine work; (b) worker digital
dexterity; and (c) the internal and external “gig” economy (Tay & Aggarwal, 2018) in which
workers accept alternative work arrangements such as temporary or single tasks jobs (Lent,
2018, p. 207). These changes will likely benefit the average person, but they will also impact
displaced workers who do not have the resources to learn new skills or pursue another career
path (Brynjolfsson & McAfee, 2014; Lent, 2018).
While organizations may be tempted to defer implementing automation, it is much more
likely that mass unemployment may come as a result of a lack of automation, rather than from its
implementation (Manyika et al., 2017a). Organizations must begin to see things more
holistically, balancing their desire for profit, greater potential productivity, and concern for the
4
overall well-being of workers (Makridakis, 2017). More broadly, the inability of organizations to
find a balance between automation and the welfare of human workers will likely lead to a
reduction in the purchasing power of workers, who also serve as consumers in the market at
large (Makridakis, 2017). Organizations will have to begin to balance automation decisions with
consideration for social equity along with economic, efficient, and effective processes (Borry &
Getha-Taylor, 2018). Organizations can no longer view investment in technology and in people
as mutually exclusive, but rather, as one activity (Poitevin, 2018). To begin solving the problem,
leaders will be required to assess social and cultural factors of how people and technology will
work together as a part of their technology implementation planning (Poitevin, 2018).
Study Context
The study focused on a specific population of the workforce. This included professional
workers working at least 40 hours per week or more, below the level of Vice President, who
worked within an industry defined by the North American Industry Classification System
(NAICS) as defined by the U.S Census Bureau (2017) (Appendix C). This study only included
individuals over the age of 18, of any gender and race, who lived and resided within the United
States.
Purpose of the Study and Research Questions
The purpose of the study was to explore individual worker demographics and
perspectives about the potential impact of automation and how those perspectives relate to the
individual’s willingness to accept change, and accept and use automation in order to assess
potential opportunities for better planned organizational change. The questions that guided the
study were the following:
5
1. Are there differences in an individual worker’s perception and understanding of
automation based on their demographic background?
2. Are there differences in an individual worker’s level of acceptance of change and
acceptance and use of automation based on their demographic background?
3. What are the perceptions of individual workers related to automation?
4. Which tasks and what percentage of one’s job does an individual worker perceive can be
automated?
5. Is there a relationship between an individual worker’s perception and understanding of
automation, acceptance of change, and level of acceptance and use of automation?
Importance of the Study
Although there has been considerable focus on the potential impact of automation on the
workforce, much of the research has focused on macro-level concepts such as the number of
people, types of jobs, industries, or tasks that could be replaced by automation, rather than on the
individual workers themselves, and their perception and needs in preparing for pending change.
This study will focus on understanding the underlying perceptions of workers, and provide
potential areas to prioritize resources, training, and investments in order to support organizational
change management. A failure to explore individual perspectives will perpetuate the gap in the
existing research and leave organizations unprepared and lacking an understanding of the needs
of workers.
Overview of Theoretical Framework and Methodology
There are two primary theories that will serve as the basis for this study’s theoretical
framework: the Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh et
al., 2003), and Acceptance of Change (AC) (Di Fabio & Gori, 2016). UTAUT has been used to
6
test potential acceptance of new technology for users by measuring various elements of
Performance Expectancy (PE), Social Influence (SI), and Effort Expectancy (EE), as well as
attitudes and behavior toward using technology while accounting for Facilitating Conditions
(FC) and Behavioral Intention (BI) (Venkatesh et al., 2003). The Acceptance of Change (AC)
construct is heavily grounded in psychology and the phenomenon of change and focuses on
embracing change from a positive standpoint (Di Fabio & Gori, 2016). AC includes dimensions
of change behaviors including Cognitive Flexibility (CF), Change Seeking (CS), and
Predisposition to Change (PC), Support for Change (SC), and Positive Reaction to Change (RC)
(Di Fabio & Gori, 2016). Both the UTAUT and AC focus on the individual, and their propensity
to adopt and accept change, albeit, automation or change itself. Each of these theories provided
part of the survey items for answering the study’s research questions and the guiding principles,
concepts, and served as valid and reliable scales to measure variables of interest.
This study was a quantitative research study conducted using the survey design method
which tests for associations among variables and typically produces descriptions of attitudes,
opinions, and trends of a population (Creswell & Creswell, 2018). Because of the exploratory
nature of the study, no hypotheses were advanced. The survey was constructed using existing
scales from UTAUT and AC to minimize validity issues and increase the reliability of the
results. The survey also introduced new items that were designed to probe for general
perceptions of automation using Likert-type and open-ended self-responses. Following this
research design allowed the analysis to focus on potential relationships, significant correlations,
and testing of the conceptual framework. Although the term “automation” was used specifically
in the research questions, in terms of this research and the methodology employed, the terms
“technology” and “automation” were considered interchangeable, as the understanding of
7
automation in this context is enabled by supporting technology that is taking the place of human
labor.
Definitions of Terms
This section contains standard industry definitions for the purpose of providing
clarification for the use of technological terms throughout this document. The terms listed are not
intended to be an exhaustive list, but rather provide the most frequently used terms in this study.
The following are generally accepted definitions of the following terms:
• Automation refers to, “automatically controlled operation of an apparatus, process, or
system by mechanical or electronic devices that take the place of human labor”
(Merriam-Webster, n.d.).
• Artificial intelligence (AI) refers to, “the capability of a machine to imitate intelligent
human behavior” (Merriam-Webster, n.d.).
• Intelligent automation refers to artificial intelligence and automation enabled systems
with the ability to synthesize large quantities of information, automate workflows and
processes, and adapt and learn, including decision making capabilities (Schatsky &
Mahidhar, 2014).
• Technology refers to, “methods, systems, and devices which are the result of scientific
knowledge being used for practical purposes” (Collins, n.d.).
• Reskilling refers to, “the process of learning new skills so you can do a different job, or of
training people to do a different job” (Cambridge University Press, n.d.).
• Up-skill is, “to provide (someone, such as an employee) with more advanced skills
through additional education and training” (Merriam-Webster, n.d.).
8
Organization of the Dissertation
This dissertation will follow a standard five-chapter approach including this chapter, the
introduction; and subsequent chapters including a literature review, methodology, results, and
recommendations and discussion. Chapter two, literature review, focuses on current research
from empirical sources, relevant books, and white papers (e.g., industry research reports), and
other reports to provide a background on existing research efforts and key concepts. Chapter
three, methodology, provides an overview of the methodology, discussion of the researcher, a
description of the data source, participants, instrumentation, data collection and analysis
procedures, validity and reliability considerations, and ethics. Chapter four, results, includes an
overview of the results, provides answers for individual research questions, and a summary of all
results. Chapter five, recommendations and discussion, expands on the results and provides
specific recommendations for the field, and includes a review limitations and delimitations of the
study, and recommendations for future research.
9
Chapter Two: Literature Review
This literature review is designed to provide an overview of multiple dimensions of
automation. The first part of the review contains the historical context of automation leading up
to today. The review also covers workforce predictions for coming automation in the future
including macro (large-scale), tasks, jobs, and occupational based skill, wage, and demographic
predictions; as well as current reactions to automation. Potential benefits of adopting automation
are reviewed based on recent research as well as current pressures and challenges facing
organizations in adopting automation. Workforce considerations for leaders regarding ethics,
social good, and potential societal impact and current strategies for preparing for automation are
also highlighted. The last section of this literature review contains an overview of the two
foundational theories, Acceptance of Change (AC), and the Unified Theory of Acceptance and
Use of Technology (UTAUT), and which served as the basis for the theoretical and conceptual
frameworks.
Historical Progression of Automation
The context and history of automation is a primary facet in understanding the
underpinnings of automation of today. This section highlights the history of automation over the
past few centuries and provides an overview of some of the critical historical turning points in
the last 50 years. This section also includes descriptions of technological advancements over the
past 25 years that have enabled advanced forms of automation, which are predicted to rapidly
transform the nature of work for humans over the next five to 10 years.
The First Machine Age
The first half of the 19
th
century in the United States was dominated by increased
automation replacing labor-intensive processes, known as the “first machine age” (Acemoglu &
10
Restrepo, 2018; Brynjolfsson & McAfee, 2014). In the first machine age, mechanical power
dominated technological innovations such as the steam engine, combustion engines,
automobiles, continuous production lines and widespread electricity powering homes and
appliances (Brynjolfsson & McAfee, 2014; Makridakis, 2017). Continuing through World War
II, agricultural inventions such as the cotton gin, horse-powered harvesters, reapers, and plows,
further replaced or reduced monotonous, physically demanding manual labor, and eventually led
to the invention of the tractor (Acemoglu & Restrepo, 2018; Autor, 2015). These seemingly
rudimentary inventions allowed farmers and factories to become more productive and
considerable cost savings that eventually enabling the invention of the computer (Acemoglu &
Restrepo, 2018).
The first general purpose programmable digital computer was invented in 1946, called
the Electronic Numerical Integrator and Computer (ENIAC), which could calculate a broad
range of complicated problems (Makridakis, 2017; Swaine & Freiberger, 2020). Between the
1950s and 1970s, computers permeated the business landscape starting with the International
Business Machines (IBM) Corporation’s business computers, followed by electronic data
processing, time-sharing computers, and the invention of the microprocessor (Makridakis, 2017).
As workers began to have more time away from work, this generated new demands for leisure
and manufactured goods, which in turn raised the threshold for higher education and technical
training (Autor, 2015).
The Second Machine Age
As the United States entered the 1980s, computers were introduced into the mainstream
automation mix, allowing for interconnectivity through modems, serving as the catalyst for the
“second machine age” (Brynjolfsson & McAfee, 2014, p. 7) and the basis for Intelligent
11
Automation (IA) (Autor, 2015; Makridakis, 2017). What made computers so successful was their
ability to replicate and perform simple tasks by following strict protocols and programming
procedures (Autor, 2015). As computing power grew, so did their capacity to more accurately
and quickly perform routine work, continuously replacing routine human tasks (Autor, 2015).
The second machine age provided the right conditions for development of early forms of
IA through traditional information exchange systems, called Structured Data Interactions (SDIs)
(Brynjolfsson & McAfee, 2014; EY, 2017). SDIs follow well-structured integration protocols,
forming the base of data transformation tools, relational database management systems
(RDBMS), web services, and application programming interfaces (APIs) (EY, 2017). As
computer processing power and software capacity increased, SDI was then able to incorporate
more complex rules driven scripting processes which led to Robotics Process Automation (RPA)
(EY, 2017; Patel, 2018). In its earliest form, RPA used software-based robotics through physical
equipment to perform human functions (EY, 2017). RPA has since progressed from performing
individual tasks through isolated robotic equipment to multifaceted and interconnected enterprise
level software-based solutions (Eggers et al., 2017; EY, 2017; Patel, 2018).
By 1993, technological innovation would once again reach new heights, with the
invention of the Intel Pentium microprocessor (Makridakis, 2017). Credited as a primary driver
early forms of AI, the microprocessor provided the computing power needed to automate tasks
beyond standard RPA, further replacing redundant tasks in far more industries than just
manufacturing (Brynjolfsson & McAfee, 2011; Makridakis, 2017; Patel, 2018). Prior to this,
computers had primarily served as data processors with few exceptions, such as the first neural
net device that read handwritten digits in 1990 (Makridakis, 2017). The microprocessor allowed
for advanced computer applications to extend beyond the constraints of prior technology into
12
advancements in AI and IA such as robotic vision, chess, and speech capabilities (Makridakis,
2017). By the late 2000s computers could make complicated decisions and rapidly analyze large
data with pinnacle inventions such as Google’s first self-driving car and the IBM Watson
computer winning at Jeopardy in the early 2010s (Makridakis, 2017). Technology progressed
from merely being able to investigate cognitive processes and simulate intelligence to a continual
increase in capacity to learn and grow (Makridakis, 2017; Wisskirchen et al., 2017).
Automation in the 2020s
The speed of automation has continued to increase along with the exponential reduction
of time between technological inventions and their general widespread use (Makridakis, 2017).
For additional perspective, there was over 200 years between the invention of the steam engine
in 1707 and the Ford automobile in 1908 whereas, there was over 90 years between the
introduction of electricity and the resulting improvements on productivity in factories
(Makridakis, 2017). In the last 25 years, computers have progressed from processing simple
zeros and ones to enabling sophisticated calculations and algorithms used today, such as
employing neural network interfaces enabling computers to see, understand, and engage in
conversation (Makridakis, 2017).
In today’s technological landscape, Intelligent Automation (IA) is further supported by
the advanced technologies of Machine Learning (ML) and advanced Artificial Intelligence (AI)
(Autor, 2015; Brynjolfsson & McAfee, 2014). Using AI and cognitive computing capabilities, IA
assists with decisions and actually making changes in the process by learning from experience
and making adjustments through cognitive-like processes (Patel, 2018). Unlike RPA, IA is not a
standard (desktop) computer solution and requires greater integration with other systems and
additional resources to build and maintain (Patel, 2018). IA can be applied in the context of
13
assisting humans (through scripting activities), replacing activities or tasks (through executing
processes), and deciding (through assisting with decisions) through what is called back-end and
front-end automation (Patel, 2018).
Back-end automation follows strict rules-based controls, much like SDI, to assist or
replace human tasks and typically involves RPA engineering of desktop (computer) office
activities such as consolidating data from multiple sources and automating sequences of manual
steps in a process (Patel, 2018). Back-end automation can include assisting with or replacing
some human tasks through the use of virtual workers, replacing manual repetitive human
activities, or manipulating existing software to further automate and complete existing processes
(Patel, 2018). Front-end automation is a more advanced form of IA and often leverages digital or
virtual assistants to replace human tasks with computer-generated characters that simulate human
interaction such as answering questions or providing guidance (Patel, 2018). Unlike back-end
automation, front-end automation replaces or helps with heavily judgement-based decision-
making tasks (Patel, 2018).
Intelligent Automation (IA) is now capable of mimicking human actions as seen through
the emergence of Machine Learning (ML). With ML, the machine is programmed and trained by
providing large amounts of curated examples to establish a baseline or “grounded truth” (Autor,
2015, p. 25). The computer system processes information and learns by trial and error, taking in
vast amounts of data and analyzing them against the baseline to determine normal or abnormal
data markers (Eggers et al., 2017). The machine then applies statistical and algorithmic based
analysis to identify information deemed necessary by the operator and learns how to handle
originally unanticipated variations to make more advanced predictions (EY, 2017; Varian, 2014).
ML systems like the IBM Watson computer have enabled semi-automated treatment and
14
diagnosis within the healthcare field through their ability to analyze millions of medical reports
and records to search for patterns, develop benchmarks, and provide a rapid comparison and
greater variety of treatment options (Cohn, 2013).
Today, ML is no longer confined to routine tasks and is capable of uncovering
commonalties that exist in larger data sets (sometimes referred to as “big data”) (Frey & Osborn,
2017, p. 259), and processing non routine, non-rule-based queries through even more advanced
technologies (Brynjolfsson & McAfee, 2011). Advancements in computer vision, which is the
ability for computers to identify scenes, objects, human faces, or scanned images, has also aided
with bulk data intake in ML (Eggers et al., 2017). Natural Language Processing (NLP) leverages
learning algorithms and statistical methods to scan and analyze unstructured information such as
text, terms, people, or places to construct meaning, intent, or sentiment (Eggers et al., 2017; EY,
2017). Natural Language Generation (NLG) is able to generate text as people speak and fill in
fields into structured forms and is aiding with things such as a company’s financial analysis (EY,
2017).
Software and AI-powered technologies are now able to handle complex tasks such as
conducting inventories, coordinating logistics, preparing taxes, retrieving information, translating
documents, writing reports, and even diagnosing diseases (Acemoglu & Restrepo, 2018). Rules
based systems have now progressed and are capable of solving routine functions as well as more
challenging problems such as recognizing human speech and providing automatic transcription
(Eggers et al., 2017). RPA is now capable of mirroring human intelligence and undertaking
increasingly complex tasks such as data collection, verification, and data entry (EY, 2017: Patel,
2018). Advanced forms of speech recognition have also enabled machine translation of speech or
text into different languages (Eggers et al., 2017).
15
The most advanced AI today are decision systems that employ a wide range of
technology and algorithms to solve complex, multidimensional, inter-related problems to
formulate decisions (EY, 2017). These systems are predicted to continue to employ deep
learning and cognitive capabilities, apply statistical models, and recognize patterns, while
integrating complex algorithms to make choices (EY, 2017). As a branch of ML, deep learning
mimics the human brain through artificial neural networks (Nitin et al., 2017). These AI-
Decision systems are now being used to help people decide on the best course of action for
something like shipping a package to a customer, while factoring in constraints such as weather,
distance, and real-time availability (EY, 2017).
Workforce Predictions for Automation
Although Gordon Moore (1998) helped to predict the rate of technological change,
predicting the full implications of automation will be challenging (Makridakis, 2017). The
largest portion of recent research around automation deals with the scale of its impact, potential
scenarios of employment growth, availability of future work, impact on skills and wages, and
how to manage the workforce transition (Manyika et al., 2017a). This section provides an
overview of current predictions around automation. The section breaks down large scale (macro
level) predictions, task, job, and occupational (micro level) predictions, as well as skill, wage,
and demographic predictions and current worker reactions to automation.
Need for More Jobs After Initial Automation-Based Displacement
Although the topic of automation has seen an increase in attention in recent years,
concerns are far from novel. The economist John Maynard Keynes (1930, 1978) wrote about his
concern for automation replacing human labor in what he called “technological unemployment”
in his 1930 article about economic predictions for the future (p. 325). Keynes (1930, 1978)
16
believed rapidly growing technological efficiencies 100 years beyond 1930 would reduce the
need for human labor by 75% and significantly raise the standard of living. Keynes predicted that
the reduced need for human labor would result in temporary unemployment and a period of
adjustment, as efficiencies outpaced the ability for business to find new types of labor for
humans to perform (Keynes, 1930, 1978). As it stands today, in general, automation has
distinctly impacted economies in three primary ways including (a) displacing labor; (b) raising
the productivity rate of labor, by requiring fewer inputs (e.g., materials) and increasing outputs
and quality; and (c) lifting the short-term Gross Domestic Product (GDP) and overall economic
investment as adoption of automation takes place (Manyika et al., 2017a).
History has shown that massive job losses from implementing automation are atypical,
and that technological innovation typically leads to greater demand for employment after the
period of initial displacement (Brynjolfsson & McAfee, 2014; Manyika et al., 2017a). This
displacement is predicted to cause workers to transition and learn new skills, with the potential
for upwards of 375 million changing occupations, and another 400 to 800 million workers
having to find new jobs by 2030 (Manyika et al., 2017a). Advanced economies such as the US
could expect anywhere from 16 to 54 million workers changing occupations, and up to one third
of workers learning new skills in the same timeframe (Manyika et al., 2017a).
During the period of transition, it is predicted that the demand for labor and wages will be
initially depressed, but will eventually be counteracted by a productivity effect (Acemoglu &
Restrepo 2018), or “scale effect” (Acemoglu & Restrepo, 2017, p. 9). Rather than massive job
losses from specific technology, jobs lost is predicted to be offset by a greater number of jobs
created (Vermeulen et al., 2018). Automation is predicted to have a much larger effect on low-
wage and low-skilled workers, although those same workers will likely have to transition to jobs
17
that that require more creativity, intellectual capacity, or greater social intelligence, which are not
as susceptible to replacement (Frey & Osborn, 2017). Workers who were initially displaced, will
later enjoy greater employment opportunities and income due to the cost savings generated by
automation and an increased demand for the tasks that cannot be automated (Acemoglu &
Restrepo, 2018; Vermeulen et al., 2018). As productivity increases, organizational capital and
improvements from automation accumulates, which further increases the demand for different
types of labor (Acemoglu & Restrepo, 2018).
Although predictions indicate that up to half of workforce activities could be automated,
less than 5% of occupations have activities that can be completely automated (Manyika et al.,
2017a). As the technology advances, humans are more likely find themselves working alongside
robots, rather than experiencing complete replacement (Eggers et al., 2017). By 2022, one in five
workers could rely on automation to perform most of their job (Mok, 2018) because, while
automation can provide a good substitute for human labor, it does a better job complimenting
and enhancing human capacity to perform (Autor, 2015; Eggers et al., 2017). Rather than
complete replacement, humans are likely to find themselves transitioning to new types of jobs,
occupations, and that will require them to learn new skills for different tasks (Autor, 2015;
Manyika et al., 2017a). New labor types, tasks, and activities will surface, changing the dynamic
of the human labor share contribution and creating unique challenges for leaders to match the
skills required for using new technologies (Acemoglu & Restrepo, 2018).
Through 2030, McKinsey predicts an 8 to 9% increase in demand for jobs that were
previously non-existent, offsetting any decline in other sectors caused by automation (Manyika
et al., 2017a). The largest net share of job growth is predicted to include jobs across healthcare,
IT, professionals (e.g., analysts, scientists, engineers), executives and managers, educators in
18
economies that are emerging, artists who meet the growing demand of those with higher
incomes, construction professionals and tradespeople, and manual laborers in unpredictable
environments (Manyika et al., 2017a). Middle-skill jobs that require a mixture of physical
activity and vocationally foundational skills like abstract problem solving, adaptability, literacy,
empathy, and cognitive flexibility, are predicted to persist (Acemoglu & Restrepo, 2018; Autor,
2015). Architects, electricians, engineers, carpenters, and those in skilled trades and construction
could account for up to 200 million new jobs as a demand for physical buildings to support
critical infrastructure and gaps in housing increase (Manyika et al., 2017a). As energy efficiency
becomes more necessary, investments in wind, solar, or other renewable energy technologies
could require an additional 10 million construction workers, installers, and jobs in manufacturing
(Manyika et al., 2017a).
Overall, the predicted worldwide workforce transition could generate $23 trillion in
consumer spending in local and export economies, creating up to 280 million new jobs in
consumer goods, and up to 85 million jobs in education and health through 2030 (Manyika et al.,
2017a). As products become less expensive to make, and services are cheaper to acquire,
spending will increase, thereby offsetting reductions seen in sectors that previously made goods
or provided services (Vermeulen et al., 2018). As workers upskill and automation replaces
routine tasks, personal disposable income is predicted to increase by over 51% through 2026,
increasing demand for labor in other categories that are making products and applying new
technology (Vermeulen et al., 2018). Technology spending could increase by over 50%, with
information technology (IT) spending accounting for at least half of the overall growth and
creation of up to 50 million high-wage technology sector jobs (Manyika et al., 2017a).
19
Different Task, Job, and Occupational Susceptibility to Automation
Much of the research in the past 10 years has been focused on the types of jobs that will
be affected by technological automation. Research has focused on tasks, jobs, and occupations
that could potentially be affected, using various models to help with predictions. Rather than
looking to one job category, researchers are looking across multiple categories in hopes of
accurately predicting people and jobs that will be affected.
In order to help standardize research, categories from data sets such as the Bureau of
Labor Statistics (BLS) North American Industry Classification (NAICS) occupational sectors
within Standard Occupational Classifications (SOCs) occupational groups have been used as a
baseline (Vermeulen et al., 2018). Within occupational groups in research, jobs are typically
classified as routine, semi-routine, or non-routine (Manyika et al., 2017b). Occupations have also
been classified based on (a) workers directly or indirectly susceptible to changes in labor demand
or income; (b) those who will develop, enhance, and maintain new technology (such as robots);
and (c) occupations that will emerge and provide innovative services or products that use,
extend, or recombine technology (Lordan & Neumark, 2018; Vermeulen et al., 2018).
Although all occupational groups are expected to see some level of full or partial
replacement of tasks, the greatest impact from automation is predicted to be for workers who
perform the most predictable, physical, and routine tasks (Kim et al., 2017; Manyika et al.,
2017b, Muro, et al., 2019). The greatest number (over 10 million) of jobs by occupation that are
predicted to be affected by automation include office and administrative support, sales, food
preparation and serving, and transportation and material movers (Vermeulen et al., 2018). The
greatest susceptibility to replacement by occupation include farming, fishing, and forestry,
installation, maintenance, and repair, construction and extraction, building and grounds cleaning
20
and maintenance, transportation and material moving, and production (Vermeulen et al., 2018).
Other potential occupations for replacement include data collectors and processors in sectors
such as paralegal work, mortgage origination, back-office processing, and accounting (Manyika
et al., 2017a).
Low-skilled occupations such as factory equipment operators and warehouse and factory
material movers are far more susceptible to robotization, whereas middle skilled occupations
such as plant operators, broadcast equipment operators, and parking attendants are likely to be
most susceptible to replacement by software (Webb, 2020). Mobile Robots (MRs) and Machine
Learning (ML) technology could potentially affect 702 occupations, with the greatest risk falling
on those in routine and labor-intensive occupations such as cashiers, cooks, technicians, and
assemblers (Frey & Osborne, 2017). Over the next 20 years, AI may be able to more closely
match human intelligence, performing more and more tasks that were traditionally handled by
people, which would challenge human jobs and become a serious competitor (Kim et al., 2017;
Makridakis, 2017). As AI increases in its ability to perform, it will continue to disrupt a much
wider range of occupations and high-skilled workers than technology such as robots or software
(Webb, 2020). Unlike robots or software, AI is predicted to have a greater impact on higher-
skilled older workers who are more highly educated, but will be dependent on the labor market,
education supply channels, occupational mobility, and human capital investment (Webb, 2020).
While the aforementioned research highlights the most likely to be affected, there are still
many tasks that will continue to be challenging for machines to replace (Fleming, 2019; Frey &
Osborn, 2017). At the baseline, jobs that require more dexterity, technical skill, or creativity are
not as susceptible to automation (Vermeulen et al., 2018). Tasks that are challenging for
machines are sometimes referred to as “computerization bottlenecks” and include tasks across
21
creative intelligence (arts, originality), perception and manipulation (hand or finger dexterity
functions, confined or awkward workspaces), and social intelligence (negotiation, social
perception, persuasion, caring or assisting others) (Frey & Osborn, 2017). Bessen (2016)
provides the example of the introduction of automated teller machines (ATMs), which was
coupled with an increase in bank branches and new bank tellers, who specialized in tasks that
ATMs were incapable of performing.
Jobs that require developing or managing people potentially represent a 9% risk of
replacement, decision-making and problem solving 18%, working with diverse stakeholder
groups 20%, and operating machines or conducting physical tasks in unpredictable environments
26% (Manyika et al., 2017a). The lowest risk includes occupations require thought processes that
are harder for machines to replicate such as therapists, healthcare workers, artists, advisors,
management types, and those who perform physical labor in unpredictable environments (Frey &
Osborn, 2017; Manyika et al., 2017b). The list of least susceptible includes workers in heavily
manual occupations such as postal carriers, podiatrists, and barbers in occupations that require
dexterity that is challenging for machines (Brynjolfsson & McAfee, 2014; Webb, 2020).
Workers in occupations such as plumbing, gardening, and child and elder care workers may also
be safe, at least through 2030, due to lower wages, difficulty to automate, and their generally
unpredictable operational environments (Manyika et al., 2017a).
Other Automation Susceptibility Prediction Considerations
While many researchers attempt to make proper estimates, there are still some residual
risks in estimation approaches because of the variation of skill required across different
workplaces within one particular occupation (Arntz et al, 2017). To reduce the potential risk of
overestimating replacement and susceptibility to automation, some researchers are now
22
incorporating measures for variation based on occupation-level verse job-level (Arntz et al.,
2017). Using the job-level approach has been shown to help balance predictions for the relative
size of job-creation and job-destruction, reducing the risk of replacement down to 9% (Arntz et
al., 2017).
Socio-economic influences, through “bounded automation,” could also account for a
reduction of adoption and shape the diffusion of technology across occupations (Fleming, 2019,
p. 24). Under bounded automation, robotization, is dependent on the price of labor, where
workers in one location may be willing to work for a lower wage, thereby increasing the value of
human labor over machines and reducing the value of investing in a robot to complete the job
(Fleming, 2019). In this scenario, the cheaper substitute of human labor will always drive the
level of investment in technology (Fleming, 2019). In another scenario, workers who demand
higher wages and benefits, and push back against automation, are more likely to encourage
organizations to automate, because organizations will be incentivized to remove the human
component altogether (Fleming, 2019). Some recent examples of this have shown labor disputes
and strikes resulting in automating the jobs of entire workforces, to neutralize threats to the
company (Fleming, 2019).
Jobs unworthy of automating will potentially continue, either due to considerably low
prices or low skill demanded, or because unions or other protective systems make it inherently
difficult to challenge the current arrangements in place (Fleming, 2019). Middle skill workers
who perform tacit skills, requiring judgement, flexibility, or common sense could remain viable
due to the challenges in machines face in performing them (Autor, 2015; Hemous & Olsen,
2014). In a more practical sense, some people may still prefer to speak with a live person,
especially if their issue deals with something substantial like a compromised bank account
23
(Fleming, 2019). While commercial airlines may technically be ready to replace commercial
airline pilots, humans may not be ready for the reality that a machine is flying without any
human intervention (Fleming, 2019).
Workers in advanced societies have traditionally protected themselves by becoming more
elite, and forging alliances with those in power to make it more challenging to completely
replace them through automation (Fleming, 2019). For those workers facing partial or semi-
automatable occupations, in which jobs are altered or restructured, workers can leverage their
unique expertise (Fleming, 2019). Leveraging unique expertise can also be combined with a
strategy of gaining additional credentials which may help maintain a niche contribution for
human labor component (Fleming, 2019).
Uneven Distribution of Those Affected by Automation
Even though researchers vary in their estimations of tasks, jobs, or occupations that will
be impacted by automation, studies have indicated that some workers are more vulnerable than
others. Automation is predicted to affect both blue and white-collar jobs, although not
necessarily in a proportionate way for men and women, or different races (Borry & Getha-
Taylor, 2018; Lent, 2018; Muro, et al., 2019). Evidence shows that globalization and automation
has caused income inequality from inefficient redistribution channels, rising job insecurity, and
higher unemployment in advanced economics (like the United States) since the Great Recession
of the late 2000s (Colombino, 2019). This has led to an uneven distribution of rewards for
workers with higher demand skills (Brynjolfsson, & McAfee, 2014) and in emerging economies,
educated workers and those with advanced degrees have already seen an increase in demand
across all levels of employment (Manyika et al., 2017a).
24
Workers at Greatest Risk for Impact and Potential Replacement
Research suggests the workers who face the greatest risk for replacement from
automation are low-skilled, low-wage workers (Vermeulen et al., 2018). These workers typically
occupy lower paying roles in industries such as service, transportation, production, or office
administration (Fray & Osborne, 2017; Muro et al., 2019). Young workers (between 16 and 24),
men, underrepresented communities and those in lower population rural and metropolitan areas
face increased risk because these groups occupy more automatable occupations (Muro et al.,
2019). More specifically, ethnic and racial groups such as Hispanic, Black, and Native American
workers have a higher risk because they occupy a greater number of jobs at risk (Borry & Getha-
Taylor, 2018; Muro et al., 2019).
As of 2019, Brookings reported that 44% of American workers (53 million) between the
ages of 18 and 64 were low-wage earners, which means that nearly half of the working class
were making annual income of around $18,000 (Ross & Bateman, 2019). Nearly 20 million low-
wage workers had a high school diploma or less and seven million between 18 and 24 did not
have a college degree and were not in school (Ross & Bateman, 2019). Based on almost 400
metropolitan areas in the United States, low-wage workers made up larger shares of the
workforce in southwestern parts of the United States, though there were still hundreds of
thousands existing in large metropolitan areas such as Seattle, Washington D.C., and San
Francisco (Ross & Bateman, 2019).
While some workers in the Brookings report were only temporarily categorized as low-
wage earners, many of those workers statistically do not move on to higher paying jobs (Ross &
Bateman, 2019). The most likely to remain in the low-wage category continues to include people
of color, women, and those with lower levels of education (Ross & Bateman, 2019). Existing
25
conditions for disadvantaged workers in inner cities and poverty prone areas will only exacerbate
issues for workers who are displaced by automation (Lent, 2018). More concerning is that 27
million in the category of low-wage are a substantial or primary contributors for the total income
in their family (Ross & Bateman, 2019). To further complicate the problem, there has been an
uneven representation from traditionally underrepresented groups in organizational leadership
and positions of power (Borry & Getha-Taylor, 2018). This creates challenges for workers who
are already lacking the resources to protect themselves, because those in a position of power are
also in a position responsible for making the decisions that will affect underrepresented and
vulnerable groups (Borry & Getha-Taylor, 2018).
Although automation generally raises the standard of living for society overall, models
for perpetual growth predict that automation will lead to business owners gaining while affected
workers struggle (Prettner, 2017). If wages for low-wage and low-skilled workers begin to rise,
susceptibility also increases, because organizations are further incentivized to reduce their labor
costs and turn to automation (Lordan & Neumark, 2018). As positions are eliminated, there are
fewer low-skilled positions available with a greater number of job seekers for non-existent
positions (Cords & Prettner, 2018). Moreover, displaced workers who are not able to increase
their skill level remain vulnerable both to subsequent replacement and underemployment
(Lordan & Neumark, 2018). What further complicates the situation is that workers facing
replacement are already facing underrepresentation and have fewer resources to prepare
themselves for transition or finding employment (Lent, 2018).
Workers at Reduced Risk for Impact and Potential Replacement
Even though low-skilled, low-wage workers may be in the most vulnerable group, it does
not mean that all models predict a dismal future. Much of the impact for disadvantaged workers
26
will be primarily based on differing levels of adoption across occupations based on the timing of
technological advancements and implementation (Lent, 2018). In a scenario where low-skilled,
low-wage workers are also faced with little automation, they will maintain the status quo
(Hemous & Olsen, 2014). If wages begin to rise, the labor category will likely see depressed
growth and reduced share of the labor market, but as automation levels off, wages may sill prove
to be a motivating attractor of employers for some types of low-skilled labor (Hemous & Olsen,
2014). In the long-term, automation may initially displace low-skilled labor, but only for workers
who already receive high wages in comparison to the relative cost of automation (Hemous &
Olsen, 2014).
Workers already employed in high-skilled positions are predicted to generally benefit
from lower unemployment rates and increased wages (Cords & Prettner, 2018). Tasks that
require more cognition and are non-routine are typically conducted by higher-skilled workers
who are paid more (Vermeulen et al., 2018). Part of the protection afforded high-skilled workers
is because automation is a great substitute for low-skilled workers and a poor substitute for high-
skilled workers (Cords & Prettner, 2018; Lankisch et al., 2017). As labor is more plentiful in the
low-skilled market, unemployment of low-skilled workers increases, and wages are driven lower
coupled with rising wages for high-skilled workers (Cords & Prettner, 2018). While high-skilled
and educated workers may initially enjoy a period of reduced risk, those same workers will need
to continue to add value in their organizations to keep up with continued technological
innovation to maintain relevance (Lordan & Neumark, 2018; Muro et al., 2019).
Current Workforce Reactions to Automation
With all of the interest around the potential effect of automation on jobs, skills, and
workers, researchers are also looking to gauge workers perceptions of coming automation.
27
Evidence from a 2019 study suggests that workers have a generally positive outlook on
automation, and almost half of those surveyed (18,813) believed automation has made their job
easier, reduced their risk of injury, improved their quality of work, and made work more
interesting (Ipsos). Another study of 6,477 full-time employees found mostly positive
perceptions of better work outcomes, simplified and enhanced processes, improved job
satisfaction, and reports of more time to focus on tertiary job functions, which contributed to
greater job creation, rather than elimination (Davenport, 2019). The Davenport (2019) study also
noted that almost 70% of 6,477 full-time employees were interested in learning new skills, but
many of those same workers were concerned with job changes, and the unknown level of effort
for learning new skills. Almost one third of workers from the Ipsos study reported that they were
worried that automation may put their employment at risk and that they felt unevenly prepared or
trained for new skills needed for new automation (Ipsos, 2019).
Another way that workers are reacting to the introduction of automation is through
finding other types of work that are more temporary in nature. Workers are now accepting work
in the “gig” economy (Tay & Aggarwal, 2018). With the gig economy, workers accept
alternative work arrangements such as temporary, or single task jobs, such as TaskRabit or Uber
(Lent, 2018, p. 207). The gig economy allows for much more flexibility, and some buffer to the
impacts of automation, but does not traditionally offer benefits such as health care for temporary
workers (Lent, 2018). With gig work, workers have more freedom to pick what they do, when
they will do it, and for how much pay they will take for the work (Brown et al., 2018). Gig
workers are also seen as freelancers or consultants who typically do not want to subscribe to
historically based 9:00 a.m. to 5:00 p.m. work hours and constrained work weeks (Brown et al.,
2018).
28
Benefits of Adopting Automation
While some research has been focused on the potential elimination of jobs through
automation, Autor (2015) highlights that the goal of automation has always been to reduce
human labor, but not necessarily to eliminate employment. Until recently, it was not possible to
reduce costs and improve speed and quality at the same time through traditional automation, but
developments in cognitive technologies and IA are predicted to enable organizations to start to
realize all three (Eggers et al., 2017). As technology increases so does its value; providing faster,
more reliable work processes and decreased costs which incentivizes organizations to adopt
advanced automation (Autor, 2015). Organizational priorities are now focusing on the swath of
benefits that IA can provide, such as improvements in accuracy, speed, service continuity,
processing efficiency, ease of use, workforce agility, scalable infrastructure, and strategic focus
(Patel, 2018).
Expected Increases in Organizational Performance
A 2018 study from Accenture of 200 U.S. Federal Government executives indicated that
47% of respondents noted wanting to use automation to reduce manual process steps as a critical
priority, 37% accelerating end-to-end processing, 36% improving consistency and reducing
errors, 36% simplifying data entry and system interaction, and 30% improving reporting,
correspondence, and output generation (Patel). The reason for this focus on performance may be
that while automation can be a good substitute for labor, it almost always increases performance,
throughput, productivity, and safety (Chui et al., 2017). These increases are partially generated
by the reduction of repetitive work, which enables organizations to use people for higher value
work, reduce costs, increase speed, and enhance organizational impact (Eggers et al., 2017).
29
Machine based augmentation is predicted to provide relief from the parts of the jobs that
were least satisfying which generally leads to more meaningful and enjoyable work (Mok, 2018).
Through augmentation, organizations will be able to maintain higher quality strategic focus and
devote more time to higher-level decision-making responsibilities, reducing the need for human
oversight of repetitive yet operationally necessary tasks (Patel, 2018). Technology is now able
perform tasks more rapidly and accurately than people, leaving people to additional time to
perform higher level functions for the work that remains (Eggers et al., 2017).
In some cases, machines have reduced process cycle-time by up to 46% for a task such as
reviewing and processing applications, and increased workflows and processing efficiency
through analytics with potential cost reductions of up to 80% (Patel, 2018). Software robots are
now helping with calculations, email, and creating reports and documents. ML is being used in
manufacturing for predicting maintenance and repairs, thereby extending equipment life and
reducing expenditures by up to 10 to 20% (Chui et al., 2017). Farmers are using advanced
algorithms to better predict the types of seeds, nutrients, soil, and yield potential for crops, and
account for adjustments in supply channels (Chui et al., 2017). Increases in cognitive automation
is reducing risks of humans missing key patterns by providing greater data accuracy for
predicting issues, helping to improve allocation of resources, enabling real-time monitoring and
tracking, data-driven decision making, and increased organizational effectiveness through
manual pattern recognition (Eggers et al., 2017). Advanced neural networks are also assisting in
pattern and anomaly detection and are starting to be capable of understanding situational context
and contributing to the decision-making process without human intervention (Eggers et al.,
2017).
30
Increased Customer Engagement and Satisfaction
Better cognitive engagement is predicted to increase relationships with customers through
use of virtual or digital assistants who helping with activities such as customer service or
marketing campaigns and are now handling human interaction, reducing human workloads and
enabling people to focus on other higher-level tasks (Chui et al., 2017; Eggers et al., 2017, EY,
2017). As people have their manual tasks reduced, they are able to focus on more valuable work,
which is further enhanced through additional cognitive insights from machines and virtual
assistants (Eggers et al., 2017). Organizations are using these opportunities to capitalize on
agility and scalable infrastructure capacity that IA provides allowing for automation of processes
with minimal disruption to ongoing operations (Patel, 2018).
Using automation has also been shown to reduce customer service wait times, increase
customer engagement, free up people to work in other roles within the organization, and allow
for reallocation of budgets from reduced costs (Eggers et al., 2017). Because automated robots
can work 24 hours a day, seven days a week, and are dependable and predictable, they offer
continuity of service with some evidence pointing to up to a 60% reduction in the need for
human interaction and a 30% improvement in speed of service (Patel, 2018). AI and virtual
agents are predicted to significantly improve in interpreting text, voice, and understanding
languages while being able to respond intelligently without human aid (Chui et al., 2017; EY,
2017). RPA and IA are becoming easier to use and maintain leading to the increased use of
digital assistants to help workers keep up with the demands of tracking multiple sources of data
(Patel, 2018).
31
Economic and Societal Benefits
One of the primary economic benefits of automation is the potential jobs that may be
created in quaternary economic sectors to support increased demand from those who were
previously spending time making the goods or providing services (Makridakis, 2017; Vermeulen
et al., 2018). As discretionary income and free time rises, people would potentially spend it on
adjacent sectors such as travel and leisure activities, theater, arts, cultural events, sports, or
entertainment (Vermeulen et al., 2018). Much like Keynes predicted in the 1930s, others predict
that workers will likely see benefits as with the last industrial revolution, where workweeks were
reduced from seven to five days and required fewer hours (Makridakis, 2017).
Makridakis (2017) compares a potential scenario to that of the aristocrats of ancient
Greece who’s use of slaves allowed them to spend all of their time philosophizing in discussions
of democracy and exercising while slaves bore the brunt of the work. In today’s context,
machines would serve as the labor, allowing more than just the societal elite to have additional
disposable income and time for leisure activities (Makridakis, 2017). Additionally, those who
perform dangerous tasks may benefit from a safer and more enjoyable working environment
through the automation of risky parts of their work (Lent, 2018).
By one estimation, implementing task-based automation, an organization such as the U.S.
Government could see an annual savings (out of 4.3 billion work hours) somewhere between
96.7 million federal hours and $3.3 billion on the low end, and 1.2 billion hours and $41.1 billion
on the high end (Eggers et al., 2017). IA has been shown to aid in up to a 28% reduction in
processing time and a 40% increase in productivity, while reducing errors by 50% and
empowering workers to focus on value added tasks and improve data-driven decision-making
capabilities (Patel, 2018). RPA, through automated data entry, scheduling, speech recognition,
32
NLP, and automated customer service reduces the need for people to spend time on manual
tasks, paperwork, and existing backlogs (Eggers et al., 2017). Technologies like self-driving AI
vehicles that can be programmed to obey speed limits, do not get tired, do not drink alcohol or
use drugs, and are not easily distracted, will help reduce human error and improve safety
(Makridakis, 2017). These factors all contribute to the primary value proposition of automation,
which is to amplify, substitute, or supplement human tasks and improve productivity
(Makridakis, 2017).
Pressures and Challenges for Adopting Automation
Over the last two centuries, automation and technology has not made human labor
obsolete, but rather, has reflected a steady increase of the employment-to-population ratio
(Autor, 2015). Researchers are predicting that the failure to automate will more likely cause mass
unemployment than be the underlying the reason for it (Makridakis, 2017; Manyika et al.,
2017a). Automation may even lead to job shortages if organizations fail to plan for the
implications of their decisions (Lent, 2018). The true test of managing automation
implementation, innovation, and technological advancement will be in the ability of leaders to
ensure workers have the right skills required of them in their new jobs (Manyika et al., 2017a).
Automation has now become a necessity to maintain a competitive advantage in the
marketplace, requiring organizations to place greater emphasis on investing and preparing
workers for automation (Makridakis, 2017; Manyika et al., 2017a; Poitevin, 2018). As new
forms of automation are created, there will be a greater need for additional skill and human
capital knowledge (Acemoglu & Restrepo, 2018). Wheelan (2019) describes human capital as
the net total of skill of an individual, what is left when stripping away their job, assets,
possessions, which may include creativity, education, experience, qualifications, or intelligence.
33
Organizations who commit to even incremental investments in human capital can create large
competitive advantages against their peers since human capital is a scarce resource that enables
organizational productivity on a large scale (Wheelan, 2019). As organizations seek to make the
right investments in their workers, they will be challenged with how to equip people with the
right skills and train them for new roles (Lund et al., 2019) and will require building an
organizational knowledge base of worker competencies through continuous evaluation of
research and models across multiple disciplines (Cosman, 2016).
Educational systems are likely to experience increase pressure for creating pathways to
properly prepare workers for new job and skill requirements (Acemoglu & Restrepo, 2018).
Governments, institutions, and businesses will be forced to rethink business and educational
models that have not changed for many years (Manyika et al., 2017a). In general, occupations
and jobs at a lower risk for automation will require greater increases in education and skills than
those at greater risk (Manyika et al., 2017a). To make things more complicated, educational
mismatches will create an imbalance in the labor market, making it difficult for organizations to
adjust and realize the full benefits of automation (Acemoglu & Restrepo, 2018).
Organizational leaders will also be challenged with how to identify technology that will
provide the greatest anticipated benefit, while balancing the need for additional skills
(Makridakis, 2017). Research has shown that significant investments in technology does not
reduce the need for proper planning and most implementations fail because people do not end up
accepting or using new technologies (Mitchell et al., 2012). In most cases, leaders will have to
spend more energy to uncover skill gaps and determine the most effective means of motivating
their workforce to want to learn and use new technology (Lund et al., 2019).
34
As automation continues along the inevitable path of adoption, leaders will have to
reconsider business processes, talent management strategies, the true needs of their workers as
individuals, and the types of talent and skills that will be needed in conjunction with the jobs lost
and those to be created (Manyika et al., 2017a). Organizational success in balancing automation
and human skill needs may eventually come down to the ability to create a well-rounded strategy
to hire talented and motivated employees who are willing to maintain high levels of performance
and continue to innovate (Makridakis, 2017). Unfortunately, organizations who succeed in one
aspect of implementation may find that they will continue struggle to find adequate workers to
perform other remaining tasks (Makridakis, 2017).
Strategies and Considerations for Workforce Planning
Although humans have historically looked for value in the work they do, Keynes (1930,
1978) believed the greater challenge would be in helping people find meaning once their
workload has been reduced as a result of automation. Keynes (1930, 1978) went as far as
predicting a future that included 3-hour work shifts and 15-hour work weeks, yet also
highlighted the struggle humans would endure as they adjusted to their newfound time freedoms
and economic prosperity. Over the last 90 years, technological advancements in manufacturing,
transportation, mining, and agriculture have greatly reduced the need for humans as Keynes has
predicted, while also requiring greater emphasis on understanding how to increase worker
satisfaction to accompany the reduction in human labor requirements (Details provided in
Historical Progression of Automation section). This section does not provide specific
recommendations associated with this study, but rather, emphasizes effective strategic and
tactical methods for leaders to consider in conjunction with a comprehensive planning for
35
automation. Recommendations for this study are covered in chapter five, recommendations and
discussion.
Planning Considerations and Strategies
Although there may be a propensity for organizations to want to slow the pace of
automation to avoid job losses, proper planning can aid in realizing automation’s potential
benefits (Manyika et al., 2017a). Authors from McKinsey recommend planning considerations
include how to sustain investment, develop training, and how to provide workers with the means
to ease the transition, support income disparity, and collaborate across private and public sectors
around mutual solutions (Manyika et al., 2017a). Organizations should consider developing an
automation center of excellence (COE) that will be responsibility for governance, skill
development, idea generation, organizational support, and process assessment around all
automation (Patel, 2018). This would include a comprehensive governance structure to manage
organizational change and ensure best practices are used, automation is implemented properly,
processes are updated, fluctuations in demand for services are managed, proper communication
with stakeholders is maintained (Patel, 2018).
In order to understand where to start with automation, organizations should conduct a
current state assessment, creating a heat map (i.e., a visual representing varying levels of
intensity through colors) of the areas of greatest potential benefits from automation as well as
designating jobs or tasks as partially or fully-replicable, or augmentable (Chui et al., 2017).
Some initial tasks for partial replacement will likely include highly manual, routine, and
predictable tasks, such as data collection and processing, whereas jobs denoted as replaceable
should only include jobs that can be completely automated from end-to-end (Chui et al., 2017;
36
Eggers et al., 2017). Other tasks should be split up into supervisorial and non-routine, to separate
tasks that only humans can perform (Eggers et al., 2017).
As leaders survey their options, they should consider the types of technology and tasks
that would provide workers with relief from mundane tasks and allow them to concentrate on
more valuable work (Eggers et al., 2017). Some of this relief can come through partial
automation of tasks, as well as through augmentation, which is a means of extending or
complimenting human actions (Eggers et al., 2017). With augmentation, humans and computers
achieve a mutual benefit because machines are more capable of performing actions such as
rapidly analysis, scanning, or helping to predict anomalies and patterns, allowing more rapid
human intervention and decision making (Eggers et al., 2017).
Categorizing jobs as partially replaceable, fully replaceable, or augmentable, will allow
organizations to decide on a cost based (labor reducing) strategy, a value (human augmentation
and complementing) strategy, or combination of both (Eggers et al., 2017). Organizations may
also benefit from conducting a review of external organizations in similar industries to uncover
areas of improvement and to aid in avoiding the potential loss of competitive advantage (Chui et
al., 2017). According to Patel (2018), external analysis should include conducting a market scan
and a capacity study to enable selection of proper automated tools for the organization.
Organizations will then need to build a suitable infrastructure to support the rollout for virtual
environments, computer servers or hosting environments, and consider installation, service, and
future maintenance requirements (Patel, 2018). Part of the plan should factor in security,
monitoring, and risk mitigation to ensure compliance and risk reduction for robot and human
interaction (Patel, 2018). As organizations finalize their plans, Patel (2018) recommends
37
ensuring to review the original rationale for implementing automation, which may have been a
desire to improve productivity, operational efficiency, quality, or the like.
Strategies for Easing Worker Transition
Because many occupations have the potential for at least partial automation over the next
30 years, leaders must come up with strategic talent management and recruiting plans to aid in
maintaining competitiveness while systematically deploying automation (Chui et al., 2017). As
adoption takes place, it will become more critical for leaders to not only focus on the benefits
that can be achieved, but also to address some of the other larger human capital related issues
that arise (Chui et al., 2017). This may include organizations working with policy makers and
educators to help assess critical skill gaps and develop programs for investing and funding
human capital and lifelong learning programs to smooth out worker transitions (Chui et al.,
2017). This will require new polices to support demand, investment and innovation, while
focusing on occupational sectors that are predicted to see the most disruption (Manyika et al.,
2017a).
Rethinking Education and Training
One of the key strategies for preparing workers for the transition will be in rethinking the
way organizations view and invest in education, job training, retraining, and understand role
fluidity for workers within organizations, and those between jobs (Mok, 2018; Wright & Schultz,
2018). Research has demonstrated that leaders who can successfully reemploy workers within a
year, can offset initial employment losses, reducing further risks of unemployment, lower wages,
and decreased labor demand through generating increased wage, productivity, and overall
economic growth (Manyika et al., 2017a). Companies and governmental and societal entities will
have to have a common goal of helping to reform education and training around new technology
38
and incorporate career intervention and planning for those most vulnerable (Borry & Getha-
Taylor, 2018; Lent, 2018).
Brookings, a leading public policy research organization, recommends promoting a
constant learning mindset, embracing new growth and technology, helping communicate and
facilitate the adjustment, creating a mitigation plan, and looking for ways to reduce hardship for
those affected (Muro et al., 2019). Smoothing out technological unemployment will require a
focus on initiatives such as stimulation for new ventures, collaboration across multiple
stakeholders, and investment in training for current workers to learn new skills or gain
certifications (Muro et al., 2019; Poitevin, 2018; Vermeulen et al., 2018). Part of this will require
modeling and developing future transition plans that focus on complimentary type automation
(i.e., human augmentation) versus substitutionary (e.g., job replacement) (Vermeulen et al.,
2018). In order to achieve this, organizations will need to help employees more easily transition
to new roles and opportunities and to provide skill and competency gap assessments along with
learning tools, resources, and time to improve (Mok, 2018). For workers who cannot be retained,
organizations and governments may consider creating programs to subsidize employment,
sponsor transition support for displaced workers, or by providing pathways for workers to find
jobs in emerging sectors (Muro et al., 2019; Vermeulen et al., 2018).
As a part of a comprehensive approach, there will be a need to evaluate investing in on-
the-job training and upskilling (Manyika et al., 2017a). Training and upskilling opportunities will
need to be seen as part of a comprehensive approach stretching across the career spectrum (i.e.,
early, mid, late career), rather than only at the onset of hiring (Manyika et al., 2017a). This will
also require workers to continue to acquire new skills and shift their expectations away from
traditional norms, ways of working, and to begin focusing on what will be needed of them in the
39
future (Manyika et al., 2017a; Mok, 2018). In all of this, well planned training should be seen as
the pinnacle for ensuring organizations remain differentiated, at least on the basis of having a
knowledgeable and skilled workforce (Aguinis & Kraiger, 2009).
As a part of their overall societal responsibility, companies are now finding that it is in
their best interest to train and prepare their workers for the new world of work (Manyika et al.,
2017a). Because of the benefits and cost savings of automation, organizations who plan and
support workers well, will have an opportunity to reinvest capital into more automation, and
growing their organization through additional jobs at differing skill levels (Vermeulen et al.,
2018). Advanced economies such as the Unites States will need to embrace a more fluid view of
the labor market by enhancing job matching capacity to pair workers who have specific skills to
companies with associated positions (Manyika et al., 2017a). This will be a substantial move
away from the “one size fits all” approach taken by many organizations today and require new
mechanisms for designing talent and reward systems around the unique needs of individuals in
order to retain and attract the right mix of workers (Mok, 2018).
Economic Solutions to Ease Transition
Another potential solution that could help buffer against the inherent volatility from
automating would be to provide something like universal basic income, work finding assistance,
or portable benefits, as a means of transition assistance between jobs (Colombino, 2019;
Manyika et al., 2017a; Vermeulen et al., 2018). Within the scope of an income support policy,
there are essentially two types that could be considered, conditional (or categorical), which
depends on the recipient’s level of income and is therefore subject to conditions, and
unconditional basic income (UBI), which is void of any requirements of the recipient
(Colombino, 2019). For conditional basic and partial basic income guarantees to work, there
40
would need to be major changes to current polices such as having greater selectivity, less
protection, reduction of guarantees, designing greater incentives to work, and employing better
testing for eligibility (Colombino, 2019).
Critics of UBI argue that people would be motivated to work less, also known as the
welfare or poverty trap, however, this could be accounted for and factored into UBI models and
policies (Colombino, 2019). Proponents of UBI do not deny that there would have to be changes
to current systems (Colombino, 2019). Taxation, for example, would have to become progressive
and rise along with the gross income and current social assistance programs and expenditures
would need to be evaluated to see which might be replaced to help offset the massive costs
(Colombino, 2019). Welfare systems would have to be updated and new polices and standards
created to match the reallocation, retraining, and cost of new skills necessary for the global
economy (Colombino, 2019). Combining multiple welfare systems in to one basic income
program may also help to reduce political manipulation, fraud, and administrative costs
associated with current systems, because it would allow for better financial tracking and controls
around an individual program (Colombino, 2019). Research also indicates that the supply of
labor could increase because there would be a reduction or elimination of the welfare-trap and
additional investments in occupation, health, and education (Colombino, 2019).
In order to pay for basic income support, something like a “robot tax” (Arntz et al., 2017,
p. 157) has been proposed to help fund partial protection and to help buffer against mass
unemployment, wage disparity, and to help insulate those at greatest risk (Wright & Schultz,
2018; Vermeulen et al., 2018). A robot tax would essentially be applied to companies who
replace workers with automation, and the monies generated would be used to help regulate the
introduction of AI and provide for those who became unemployed due to robotization,
41
computerization, or automation (Vermeulen et al., 2018). Although on the surface, a robot tax
may seem like a valid solution, additional research around this should be conducted as this type
of policy may incentivize organizations to send their capital to countries without such a tax,
resulting in a counterproductive outcome (Vermeulen et al., 2018).
Motivating Workers Despite Constant Change
As organizations adopt AI and machines to perform work previously performed by
humans, they will need to consider how to motivate workers to adopt new behaviors and learn
new skills to increase comfort and reliance on new technology (Mok, 2018). While
understanding the types of skills that will be required for the workforce of the future is
important, strategies for learning and motivation are different. Learning is a multidimensional
process that produces long-term change for a person (Alexander et al., 2009), whereas
motivation is about the learner taking action toward achieving goals (Pintrich, 2003). In
evaluating motivation, research suggest that successful leaders evaluate motivational constructs
through the lens of multiple theories such as self-determination theory (SDT), the Unified
Theory of Acceptance and Use of Technology (UTAUT), and the technology acceptance model
to name a few (Lee et al., 2015; Mitchell et al., 2012; Rezvani et al., 2017).
Self-determination theory (SDT) was developed by Deci and Ryan in the 1980s, which
posits a link between competence and motivation (1985). With SDT, individual motivation is
based on intrinsic and extrinsic rewards, which are in essence the internal or external factors that
motivate individuals to act (Ryan & Moller, 2017). These concepts also involve achievement
motivation (McClelland, 1985; McClelland et al., 1976) which can be thought in terms of a
person’s focus on continual improvement and an attraction to accomplishing things that require
some degree of moderate risk or challenge (Sternberg, 2017).
42
Many of these theories have underlying foundations in areas of psychology, such as
Pintrich’s (2013) model that helps leaders evaluate concepts such as motivation based on self-
determination (i.e., a sense of autonomy, competence, and control over their learning), social-
cognitive constructs of self-efficacy and higher levels of interest and intrinsic motivation (e.g.,
value, directional goals), self-regulation by enhancing metacognition (i.e., self-awareness and
regulation), helping people understand what motivates them (i.e., unconscious needs and implicit
motives), and thinking though organizational context and culture. Albert Bandura’s (1986) work
in the context of Social Cognitive Theory (SCT) introduced the concept of human agency and
self-efficacy and is useful in understanding the underlying individual’s belief in achieving
personal goals. Bandura’s (2000) more recent work may be more useful leaders to understand the
idea of collective efficacy, which is the shared belief of a group that can impact an individual’s
belief in accomplishing their own goals.
As a part of evaluating automation plans and whether they will be successful or not,
researchers have already begun to study the underlying potential across many of the theories.
Performance Expectancy, Perceived Enjoyment and Effort Expectancy have shown some
indications of motivation through Perceived Usefulness of technology and intrinsic motivation
(from easier to use technology), but very little value from using extrinsic motivation (i.e.,
rewards and punishment) (Lee et al., 2015). Leaders who provide direct intervention during
automation implementation (transitional and transformational) has been shown to improve
perceived competence, usefulness of the technology, satisfaction, and more importantly,
motivation to continue use of the new system beyond training and implementation (Rezvani et
al., 2017). Research also suggests a corollary link between improving the skills of low-skilled
workers through intentional organizational designed learning the use of informal employee
43
collaboration (Kim et al., 2015). Organizations who can bolster workers positive perception of
organizational support may enjoy increased motivation, attitudes, and behaviors for accepting
and using new technology (Mitchell et al., 2012).
Regardless of the theory, leaders will need to reiterate and emphasize the continued need
for workers themselves which can help build emotional commitment and confidence,
communicate organizational support, and increase the likelihood for adopting new work
processes and technology (Davenport, 2019). This will require a focused organizational effort on
protecting workers, investing in cultivating increased individual skill levels, and seeing workers
as individuals (Fleming, 2019). Organizations can reiterate the value that individual workers
bring by reframing the meaningfulness of work, allowing workers to contribute in the decision
making and learning design process, sharing as much relevant information about the coming
changes as possible, and providing continuous performance feedback (Spreitzer et al., 2012).
Allowing workers to be a part of the decision-making process has been shown to increase
intrinsic motivation, thereby increasing skill utilization and job satisfaction (Boxall et al., 2015).
Organizations that have been able to effectively communicate the importance of individual
workers has been shown to have a direct impact on employee health, engagement, reduction of
burnout, and in motivating them to embrace new roles (Spreitzer et al., 2012).
To help increase motivation and attract top talent, organizations can consider changing
the way workers are paid, from paying the position or role, to paying the person based on their
skill level (Mok, 2018). This will require rethinking and adjusting compensation models to
account for the flexibility demanded in digital businesses because of frequent changes in needed
roles, team composition, and varying levels of skills and competencies (Mok, 2018). To help
with this effort, organizations will soon be able to use Machine Learning (e.g., predictive
44
analytics, big data, advanced algorithm modeling capabilities) to effectively evaluate and
measure skill and competency-pay (Mok, 2018). As organizations move toward viewing workers
as individuals who are paid based on their commensurate skill level, they should also enjoy the
benefits of a more engaged and motivated workforce, which will further aid in worker retention
and engagement (Mok, 2018).
Theoretical and Conceptual Framework
This section provides and overview of the theoretical basis and conceptual framework of
the study. The theoretical framework represents a selection of an existing theory, or theories that
serve as the foundation or blueprint for the construction of a research design, study, and basis of
conceptual framework (Grant & Osanloo, 2015). For this research, the theoretical framework
consisted of the Acceptance of Change (AC) model (De Fabio & Gori, 2016) and the Unified
Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh et al., 2003). The
conceptual framework typically presents a new approach to solving a given problem and
provides a logical structure for understanding how different theoretical framework concepts and
independent variables may be related to one another (Grant & Osanloo, 2015). The conceptual
framework includes elements of UTAUT and AC, as well as new concepts and potential
independent variables that were used as the lens for analyzing data collected, and understanding
potential relationships. This section describes the underlying constructs that served as the
conceptual framework for this study.
Theoretical Framework 1: Acceptance of Change
Di Fabio and Gori’s (2016) Acceptance of Change (AC) model was a theoretical model
that was used for this study. Acceptance of Change (AC) is heavily grounded in psychology and
the phenomenon of change. Prior to Di Fabio and Gori’s (2016) work on Acceptance of Change
45
(AC), much of the traditional research centered on the resistance to change. Rather than shy
away from change, AC points to embracing changes based on the theory that accepting change
has a positive impact a person’s well-being (De Fabio & Gori, 2016). The AC construct focuses
on embracing change from a positive standpoint and includes dimensions of change behaviors
(Di Fabio & Gori, 2016). The hypothesis of the AC study was based on the notion that traditional
approaches were focused on resistance to change and therefore failed to provide a means for
evaluating acceptance of change using positive measures such as openness to change (Di Fabio
& Gori, 2016).
The original AC study included five scales that were designed to evaluate Predisposition
for Change (PC), Support for Change (SC), Change Seeking (CS), Positive Reaction to Change
(RC), and Cognitive Flexibility (CF) (Di Fabio & Gori, 2016). Using the five dimensions, the
authors developed and tested an Acceptance of Change Scale (ACS) to evaluate an individual’s
overall willingness to accept change. These items asked participants to rate their ability to see
things from different points of view, make the most of situations, use coping mechanisms, how
likely they were to look for opportunities to change their situation or routine, how they tolerated
negative aspects of change, and how flexible they were to changing their mind.
Di Fabio and Gori’s (2016) study included 261 participants who completed a five-point
Likert-type scale across 20 items (four questions per dimension). Across the five dimensions, the
total variance was accounted for by Predisposition to Change at 32.58%, Support for Change at
10.46%, Change Seeking at 7.36%, Positive Reaction to Change at 6.65%, and Cognitive
Flexibility at 6.15% (Di Fabio & Gori, 2016). Results indicated AC may be specifically
important for individuals needing to construct their own resources to deal with change.
46
Theoretical Framework 2: Unified Theory of Acceptance and Use of Technology (UTAUT)
The Unified Theory of Acceptance and Use of Technology (UTAUT) effectively
evaluates the potential acceptance of new technology for users and forms one unified model from
multiple theories and concepts from social psychology, motivation theory, Perceived Behavioral
Control, human behavior, innovation, and Social Cognitive Theory (SCT) (Venkatesh et al.,
2003). UTAUT integrates and incorporates eight models including the following: technology
acceptance model, theory on reasoned action, theory of planned behavior, motivational model, a
model that combines the theory of planned behavior and the theory on technology acceptance
model, the innovation diffusion theory, the model of personal computer utilization, and Social
Cognitive Theory (SCT) (Venkatesh et al., 2003). Employing Likert-type items and scoring,
UTAUT allows for the evaluation of Performance Expectancy (PE), Social Influence (SI), Effort
Expectancy (EE), and potential attitudes toward using new technology (Venkatesh et al., 2003).
The hypothesis for UTAUT was that combining the multiple theoretical constructs would
provide a model that would serve as an effective means for managers to assess the likelihood of
successful implementation of new technology. The original UTAUT study included 215
participants from four organizations. Two of the organizations (entertainment and telecom
services) implemented non-mandatory to use technology, and the other two (banking, and public
administration) had a mandatory use policy for new technology (Venkatesh et al., 2003). Data
was initially collected and cross referenced, with further testing of an additional 133 voluntary
and mandatory users in financial services and retail electronics (Venkatesh et al., 2003). The
research included voluntariness, experience, age, and gender as moderating influences.
Dependent variables included behavioral intention, and intended use behavior (Venkatesh et al.,
2003). Independent variables were included as scales to test for Performance Expectancy (PE),
47
Effort Expectancy (EE), Attitude Toward Using Technology (ATUT), Social Influence (SI),
Facilitating Conditions (FC), Self-Efficacy (SE), Anxiety (ANX), and Behavioral Intention (BI)
(Venkatesh et al., 2003).
The UTAUT study was conducted and resulted in an empirical model that supported the
use of the authors designed measures. There were two primary measures which were (a)
determining intended use; which included Performance Expectancy (PE), Effort Expectancy
(EE), and Social Influence (SI); and (b) determining behavioral intent and conditions that
facilitated potential use (Venkatesh et al., 2003). Performance Expectancy (PE) appeared to be
the primary determinant for those who most intended to use technology, with varying intensity
based on age and gender, with more significance for younger male workers (Venkatesh et al.,
2003). Effort Expectancy (EE) was moderated by age and gender, with increased significance for
older workers and women, decreasing commensurate with their level of experience (Venkatesh
et al., 2003). Social Influence (SI) and Behavioral Intention (BI) were shown to have
contingency upon all four moderators and therefore, noted as nonsignificant without the
inclusion of moderators (Venkatesh et al., 2003). Facilitating Conditions (FC) only showed
significance when paired with experience and age, such as older workers with more experience
(Venkatesh et al., 2003).
Conceptual Framework
The literature reviewed in this study indicates that there will be a significant amount of
personal change required for people accompanying the implementation of new automation in the
workplace. Adoption of automation will require new approaches for anticipating and planning
for how to provide support for the transition of humans and machines working together. This will
48
require a dramatically different level of understanding of what will be needed for individual
workers across education, training, and organizational constructs and models.
In order to establish an understanding of the impact of automation, three survey scales
were developed. These scales included items created to evaluate an individual worker’s
Understanding of Impact of Automation (UIA) (Jobs and Tasks), Perception of Career Changes
(PCC) (Role, Job, Career), and Perception of Skills Changes (PSC) (Skills, Education, Training).
As these three scales were newly developed, there were no specific hypothesis associated,
although it could be inferred that there was an expectation of varying levels of associated results.
Given that this study was concerned with understanding how well workers would accept and use
automation, each of these scales were critical to answering the research questions in conjunction
with use of the trusted and reliable scales from Di Fabio and Gori’s (2016) Acceptance of
Change (AC) model and Venkatesh et al.’s (2003) Unified Theory of Acceptance and Use of
Technology (UTAUT).
While this study employed the original Acceptance of Change Scale (ACS) from Di
Fabio and Gori’s (2016) study, in order to better understand how workers perceived accepting
and using automation, this study employed an adapted version of the Unified Theory of
Acceptance and Use of Technology (UTAUT). Based on the results of the original UTAUT
study, the authors recommended five sub-scales for future use. For this study, the trusted and
reliable Likert-type scale items recommended by the authors were adapted to test for (a)
Perceived Usefulness (U); (b) Perceived Ease of Use (PEU); (c) Perceived Behavioral Control
(PBC); (d) Compatibility (C); and (e) Anxiety (ANX) (Figure 1). Rather than evaluating the
potential acceptance of future technology, this study adopted the term “automation” in its place,
given that this study asked participants specifically about automation. This resulted in a new
49
adapted scale designed to test for Acceptance and Use of Automation (AUA) as depicted in
Figure 1.
While the underlying theories of UTAUT have been used across a variety of research,
and UTAUT itself has been broadly adopted for assessing the likelihood of use of new
technology across various industries, there has not been a study that incorporates the AC model
for automation. Additionally, a study has not been conducted which incorporates demographic
variables and how those may impact the individual worker’s overall perception of the impact of
automation, perception of potential career changes, or perception of needed skills changes. This
study’s conceptual framework introduces the concept that multiple variables influence and
contribute to an individual’s willingness to accept the change that is required as a part adopting
and using new automation. This conceptual framework as shown in Figure 1 effectively
incorporates evaluation across several concepts and posits a new way of assessing organizational
and individual readiness for implementing automation.
50
Figure 1
Conceptual Framework
Impacts
Impacts
Impacts
Impacts
• Understanding of Impact of Automation
(Jobs and Job Tasks) (UIA)
• Perception of Career Changes
(Role, Job, Career) (PCC)
• Perception of Skills Changes
(Skills, Education, Training) (PSC)
Perception & Understanding
of Impact of Automation
• Perceived Usefulness (U)
• Perceived Ease of Use (EOU)
• Perceived Behavioral Control (PBC)
• Compatibility (C)
• Anxiety (ANX)
Acceptance and Use of Automation
(AUA) (adapted from UTAUT)
Acceptance of Change (AC)
• Predisposition for Change (PC)
• Support for Change (SC)
• Change Seeking (CS)
• Positive Reaction to Change (RC)
• Cognitive Flexibility (CF)
Demographic Variables
• Gender
• Age
• Race
• Education
• Technical Education
• Length of Time in Current
Occupation
• Total Length of Work Experience
• Number of Current Jobs
• Level in Organization
• Exposure to Automation
• Organizational Size
• Income
51
Chapter Two Summary
This literature review has provided a comprehensive overview of the history of
automation through the first and second machine age into the modern era of advanced Intelligent
Automation and AI-Decision systems of the future. The review covered multiple recent
foundational studies on current macro (large scale) workforce automation predictions, tasks,
jobs, and occupational predictions; skill, wage, and demographic level predictions; as well as
some current reactions in the workforce. Benefits of adopting automation, and current pressures
and challenges facing organizations was discussed based on recent research and literature. The
review also highlighted strategies and considerations for workforce and transition planning.
Lastly, Acceptance of Change (AC) and the Unified Theory of Acceptance and Use of
Technology (UTAUT) theoretical frameworks were described along with the details of the
conceptual framework that served as a guide for the development of this study.
52
Chapter Three: Methodology
Effective evidence-based research allows for proving or disproving of assumptions and
hypothesis, and implores the researcher to abandon unsubstantiated claims, test theories and
concepts, using valid and reliable measures to uncover potential relationships among variables
while seeking to minimalize and account for bias (Creswell & Creswell, 2018). The purpose of
the study was to explore worker perspectives on the potential impact of automation and how
those perspectives impact the worker’s willingness to accept change and accept and use
automation. This chapter discusses the study’s research questions, methodology, and provides
methods used to reduce potential bias of the research. The chapter also provides details on data
sources including participants, instrumentation, collection procedures, and analysis methods.
Measures used to increase validity and reliability as well as ethical considerations and safeguards
are also presented and described.
Research Questions
The questions that guided the study were the following:
1. Are there differences in an individual worker’s perception and understanding of
automation based on their demographic background?
2. Are there differences in an individual worker’s level of acceptance of change and
acceptance and use of automation based on their demographic background?
3. What are the perceptions of individual workers related to automation?
4. Which tasks and what percentage of one’s job does an individual worker perceive can be
automated?
5. Is there a relationship between an individual worker’s perception and understanding of
automation, acceptance of change, and level of acceptance and use of automation?
53
Overview of Methodology
This study was a quantitative research study with some open-ended items. Quantitative
studies aim to produce generalizable knowledge, test theories about processes and human
behavior, patterns, and employ a deductive approach (Creswell & Creswell, 2018). Conducting a
quantitative survey study helped the current study reach its conceptual goals by making it
possible to sample a greater number of participants, to test variables of interest, and collect an
adequate data pool for use in data analysis and economy of design and rapid turnaround
(Creswell & Creswell, 2018). Following this research design allowed the focus of the analysis to
be on potential relationships and significant correlations that emerged from the conceptual
framework’s hypothesized relationships. The conceptual framework posited that there were
relationships between demographic variables and perception and understanding of automation
related constructs, the Acceptance and Use of Automation (AUA), and Acceptance of Change
(AC). The conceptual framework also posited that there was a relationship between AC and
AUA, and between perception and understanding of automation related constructs.
The Researcher
In order to ensure the minimization of personal bias, a reflexive evolution was conducted,
which Creswell and Creswell (2018) define as a reflection of how a researcher’s personal
background, values, and potential bias may shape interpretations over the course of a study. One
of the primary ways to minimize bias is by conducting a comprehensive literature review and
subsequently, developing the theoretical and conceptual framework to serve as a baseline
foundation and conceptualization for a study, and provide opportunity to review relevant existing
instrumentation, research designs, and references for interpreting data (Merriam & Simpson,
2000; Rocco & Plakhotnik, 2009). A broad literature review of empirical and professional
54
sources was conducted in conceptualizing this study in order to minimize the personal bias of the
researcher in the research development process and methodological approach.
The entire study was peer reviewed, by my dissertation committee and by an official
Institutional Review Board (IRB) process. Potential survey bias in the design, data collection,
and analysis was minimized through the use of existing valid and reliable survey items from
tested theories. Survey bias was also minimized through pilot testing, conducting peer reviews,
and receiving dissertation committee approval of protocols before beginning survey
administration. I was distanced from participants who are providing the primary quantitative
survey data to be analyzed. Individual participant names and identifying information were not
collected. I collected enough survey data to provide the ability to engage in limited inferential
statistics.
Data Source: Survey
This study was conducted using the survey design method. The survey design method
was appropriate for this study because it tests for associations among variables and typically
produces descriptions of attitudes, opinions, and trends of a population (Creswell & Creswell,
2018). The use of surveys reduced feasibility constraints and allowed for maximum participation
while minimizing potential bias and increasing validity and reliability for providing rich data for
analysis and answering the research questions.
Participants
The study sought 200 to 300 participants in order to engage in inferential analysis and
was able to collect a total of 64 completed and viable surveys out of 291 partially completed
surveys. The survey was open to any adult of 18 years old or older, working within the United
States, and allowed for maximum participation while the survey was open. The study used crowd
55
sourced convenience sampling in order to maximize the data collected and targeted potential
participants through snowball type recruitment via emails and survey links shared on LinkedIn,
Facebook, and through word-of-mouth referrals.
Instrumentation
The data collection method included a survey with items that focused on potential
perception of coming automation and attitudes toward adoption including open-ended items. The
open-ended items were specifically related to gathering data on enumeration of and types and
percentages of tasks deemed replaceable by automation. The survey also employed Likert-type
selection choices for participants designed to enable a simplified approach to analysis and
correlation. Surveys included existing adapted items from Venkatesh et al.’s (2003) Unified
Theory of Acceptance and Use of Technology (UTAUT) and De Fabio and Gori’s (2016)
Acceptance of Change (AC) model as well as items developed by the researcher. UTAUT and
AC theories served as the basis of the theoretical and conceptual framework to provide the
guiding principles and concepts as well as valid and reliable scales to measure the variables of
interest for use in identifying key themes for answering the research questions.
A new scale entitled Acceptance and Use of Automation (AUA) was designed based on
adapted items recommended by the authors of the original UTAUT study. Selected AUA items
were adapted to replace the term “technology” with “automation.” Three new survey constructs
were also introduced including (a) Understanding of Impact of Automation (UIA) (Jobs and
Tasks); (b) Perception of Career Changes (PCC) (Role, Job, Career); and (c) Perception of Skills
Changes (PSC) (Skills, Education, Training). Survey items are available in Appendix A: Survey
Protocol and Items and Appendix B: Survey Protocol Crosswalk. Table 1 displays the chosen
categories for which survey items were adapted.
56
Table 1
Theoretical Framework Survey Item Categories and Subcategories
Theory Primary theoretical category Theoretical subcategory
Acceptance of change (AC) Predisposition for change (PC)
Support for change (SC)
Change seeking (CS)
Positive reaction to change (RC)
Cognitive flexibility (CF)
Acceptance and use of
automation (AUA)—adapted
from the unified theory of
acceptance and use of
technology (UTAUT)
Performance expectancy (PE) Perceived usefulness (U)
Effort expectancy (EE) Perceived ease of use
(EOU)
Facilitating conditions (FC) Perceived behavioral
control (PBC)
Compatibility (C)
Attitude toward using
automation (ATUA)
Anxiety (ANX)
57
Data Collection Procedures
The survey was developed following suggested survey design methods discussed in
Designing Quality Survey Questions (Robinson & Leonard, 2019), which served as the baseline
for the creation of high-quality, intentionally worded items for eliciting useful data to answer this
study’s research questions. Surveys were created in the University of Southern California (USC)
Qualtrics software platform and key concepts were divided into sections, including an
introduction; a description of key terms and concepts; and a description of the section’s purpose.
The survey was designed to take no longer than 10 minutes and was administered virtually to
enable participants to complete the survey at their convenience.
In order to increase the response rate, I posted multiple times requesting for participation
on LinkedIn and Facebook groups during the data collection window. I sent out notices regular
intervals on social media platforms to increase the potential reach and maximize the number of
potential participants during the survey response period. I did not provide any incentives to
participate in the survey. Survey results were not formally offered to participants as individual
participant names, emails, or contact methods were not collected.
Data Analysis
The survey data collection period was open from December 2020 through February 2021.
Once the data collection period was concluded, the data was downloaded and analyzed. The data
was cleaned, and any incomplete surveys were removed from the data set. Items from the open-
ended items were analyzed for greater themes and categories, based on enumeration of item
submissions. The remaining data was run through IBM SPSS Statistics analysis software for
descriptive and inferential statistical analysis. SPSS was employed to calculate descriptive
statistics such as a mean, standard deviation median, mode, as well as inferential statistics such
58
as Pearson’s r, Cronbach’s alpha statistic, and ANOVA. Data was processed using a best fit
approach, employing analytical testing through the software to determine statistical significance,
confidence intervals and effect size. Statistical significance is the reported likelihood that the
observed scores are not due to random chance, but rather, reflect a pattern (Creswell & Creswell,
2018). A confidence interval is the range of potential values denoting the potential level of
uncertainty of a particular item’s score, whereas effect size highlights the strength of the
conclusion about associations of variables or groups that is independent and describes whether
data is representative a population (Creswell & Creswell, 2018). Table 2 provides the intended
analysis by research question.
59
Table 2
Data Analysis Approach by Research Question
Research question Method of analysis
RQ1: Are there differences in an individual
worker’s perception and understanding of
automation based on their demographic
background?
Analyses of variance (ANOVAs)
RQ2: Are there differences in an individual
worker’s level of acceptance of change and
acceptance and use of automation based on
their demographic background?
Analysis of variance (ANOVAs)
RQ3: What are the perceptions of individual
workers related to automation?
Descriptive statistics, means and
standard deviations for all constructs
RQ4: Which tasks and what percentage of
one’s job does an individual worker perceive
can be automated?
Enumeration
RQ5: Is there a relationship between an
individual worker’s perception and
understanding of automation, level of
acceptance of change, and level of acceptance
and use of automation?
Simple regression
60
Validity and Reliability
Validity in quantitative research refers to the ability to draw useful and meaningful
inferences from the collected scoring of specific data instruments (Creswell & Creswell, 2018).
Reliability refers to the confidence one can have internal consistency of scored survey items on a
survey instrument in a research study typically measured by consistency of response constructs
across time, administration, and scoring (Creswell & Creswell, 2018). In order to increase
validity and reliability of the data collected in this study, survey items were developed using a
combination of adapted existing valid and reliable instruments from UTAUT and AC items.
Items created by the researcher had at least two items per category and were also tested for
internal consistency.
In the original UTAUT framework, the study used several elements to increase reliability
and validity including: pre-testing, cross reference testing, testing across multiple points in time,
Partial Least Squares (PLS), and 48 separate validity tests that yielded an acceptable rating for an
internal consistency reliability through Cronbach’s alpha coefficient greater than .70 (Venkatesh
et al., 2003). The researchers administered pre-tested questionnaires with nine to 15 scale items
from the eight models at three different points of time (Venkatesh et al., 2003). The points in
time were measured from: post-training, one-month post-implementation, and three months after
implementation; with actual use measured within the six-months post-training (Venkatesh et al.,
2003). The unified theory was applied, tested, and was noted as useful in helping leaders to
understand potential technology acceptance and adoption rates, accounting for 70% of variation
of intended use of new technology (Venkatesh et al., 2003).
The AC study employed multiple measures and tests for factor analysis, sampling
adequacy, factor loading, the Pearson correlation coefficient (for intercorrelation), scale
61
reliability and internal consistency through the Cronbach’s alpha coefficient, goodness-of-fit
indexes, and verification of concurrent validity using Pearson’s r coefficient (Di Fabio & Gori,
2016). Cronbach’s alpha coefficient provided good internal consistency values above .72 (for the
lowest dimension) with an overall value across the five dimensions of .88 (Di Fabio & Gori,
2016).
Along with the valid, and reliable survey items from existing scales, I designed survey
items and open-ended items that were introduced and used in this study. The survey items were
intended to probe for general perceptions of the impact of automation on one’s job, job tasks,
changes to career, needed education and training, and perception of who is responsible for
education and training. There were also items probing for behavioral intent. The open-ended
items were intended to probe for the percentage of job tasks deemed replicable, or to have
already been replaced by automation. These items were used to inform correlations between the
variables, rather than underlying measures of a given theory. All survey items were reviewed by
the dissertation committee and the entire survey construct went through multiple iterations of
reviews (i.e., question construction, language and syntax, order of questions, clarity). Prior to
beginning data collection, the survey was pilot tested by my dissertation candidate peers who
contributed to quality assurance, and made suggestions for changes of survey structure, flow, or
language (verbiage). In order to avoid ethical issues related to data collection, I asked those who
participated in the peer review pilot testing not to take the final survey for data collection.
Ethics
Leaders will need to prepare for the workforce transition as described by Vermeulen et al.
(2018) as the “Fourth Industrial Revolution,” which is a massive shift in the dynamics of how
people will work in the future, the types of jobs they will do, and how humans and technology
62
will work together. Based on this study’s design, the results will increase the knowledge base
around worker perceptions of the impact of automation to help inform planning considerations
for implementing automation. Both organizations and workers should benefit from the study
results as it provides current perceptions and trends in thinking from a broad range of workers.
However, there could be harm caused to workers, such as those in lower socio-economic groups,
if results are used by organizations to target investment in automation that will only affect
specific types of workers. With these considerations in mind, the scope of the research study and
the specific survey questions were purposefully designed to ensure the study was not swayed or
influenced by any outside interests.
The study only enrolled adult participants age 18 and older on a strictly voluntary basis,
and no data was gathered prior to the approval of the University of Southern California (USC)
Institutional Review Board (IRB). Based on USC IRB standards this study was an exempt study,
meaning that recorded information did not allow for identification of subjects (directly or
indirectly linking), and participant responses were not to be disclosed or place any participants at
risk based on the information they provided. While surveys asked for some demographic
information, participant names were not gathered, to ensure anonymity. Participants were
provided upfront information on the intended use of all data collected. Informed consent was
obtained through participants agreeing to continue at the start of the survey and participants were
permitted to exit the survey at any time. Participants received the name and information of the
principal researcher as well as the purpose and intent as a part of the social media postings.
63
Chapter Four: Results
The purpose of the study was to explore worker perspectives of the impact of automation
and how those perspectives impact the worker’s willingness to change and adopt and use
automation. In order to achieve the study’s primary purpose, research was conducted according
to the methodology discussed in chapter three to produce results to answer the study’s five
research questions. Chapter four is divided into eight sections. The first section discusses the
results of the quantitative survey development, content, deployment, and preparation process for
data analysis. The second section provides an overview of study participants which includes the
descriptive results of participant demographic, employment, occupation, and education data.
Sections three through seven describe the results in answering the study’s five research
questions. The final section provides a summary of the chapter.
Quantitative Survey Overview
To answer the study’s primary purpose, research questions were used to frame and
develop a quantitative survey. The survey relied on the theoretical and conceptual framework
based on the Acceptance of Change (AC) (Di Fabio & Gori, 2016), and the Unified Theory of
Acceptance and Use of Technology (UTAUT) (Venkatesh et al., 2003). The survey included the
use of existing valid and reliable survey items and scales from AC and UTAUT, and introduced
new constructs developed for this study, designed to aid in gathering additional information from
participants.
Survey Development and Content
The survey was administered through the University of Southern California’s (USC)
Qualtrics platform. The survey protocol included an introduction (section I) that qualified
candidates based on age, location (within United States), level in organization, and level of
64
education (presented in Appendix A: Survey Protocol and Items). Section II included items
designed for this study that probed for general understanding of automation and perception of
potential impact, behavioral intent, and perceived responsibility for acquiring new skills,
education, or training. Section II also included three newly created scales including
Understanding of Impact of Automation (UIA) (Jobs and Tasks) (four items), Perception of
Career Changes (PCC) (Role, Job, Career) (three items), and Perception of Skills Changes (PSC)
(Skills, Education, Training) (two items).
Section III used the Acceptance of Change (AC) scale which included ten items in their
original form. AC included pairs of items (two each) that probed for the following:
Predisposition for Change (PC), Support for Change (SC), Change Seeking (CS), Positive
Reaction to Change (RC), and Cognitive Flexibility (CF). Section IV included a selection of
items from Unified Theory of Acceptance and Use of Technology (UTAUT) that were modified
to replace the term “technology” with “automation.” The selection of items from UTAUT led to
the development of a new scale designed to probe for Acceptance and Use of Automation
(AUA). The AUA twenty item scale was made up of the following UTAUT scales and items that
probed for the following: Performance Expectancy (PE); Perceived Usefulness (U) (five items),
Effort Expectancy (EE); Perceived Ease of Use (EOU) (five items), Attitude Toward Using
Automation (ATUA); Anxiety (ANX) (four items), and Facilitating Conditions (FC) which
included Compatibility (C) (three items), and Perceived Behavioral Control (PBC) (four items).
The final part of the survey, section V, asked participants secondary demographic
questions. Questions related to race and ethnicity, educational history and prior technical
training, country of origin and English as a primary language of fluency, employment status
(full, part-time, or unemployed) and number of jobs, time spent in current and previous
65
occupation or industry, and income level. At the conclusion, participants were presented with a
message thanking them for their participation.
Survey Deployment and Preparation for Analysis
The survey was open for participation between December 2020 and February 2021 and
was advertised through LinkedIn, Facebook, and by using snowball type messaging to potential
participants. Upon closure of the data collection period, the data were uploaded into SPSS
software to begin cleaning and analysis. The data were then cleaned to remove incomplete
surveys and those outside of the target participant group.
Survey items were grouped into associated sub-scales of two to five items each. The
Anxiety (ANX) sub-scale was reverse coded to enable accurate comparison analysis. Cronbach’s
alpha was run for all scales to measure internal reliability. The lowest observed Cronbach’s alpha
score was .691 with all other scale means above .70 and several scales above .80, indicating
strong internal consistency across all scales. The resulting internal reliability therefore provided
validity for the use of five primary scales for analysis consisting of three original scales, and the
AC and AUA scales, which were based on the theoretical and conceptual framework and
research methodology (Table 1). Remaining questions and items were run through the SPSS
software to gather an initial understanding of descriptive statistical data (i.e., frequencies, means
and standard deviations). The data were documented and in some cases categories were
collapsed and recoded into logical groups as described in the context of related research
questions.
Study Participants
There was a total of 283 survey participants within the initial survey data pool before
reduction based on target characteristics and removal of incompletions. The survey targeted
66
workers at the “Director” level and below based on information from the literature review that
highlighted types of workers more likely to be impacted by automation. Participants who
selected “Vice President” or above; including Senior Vice President or Partner, C level executive
(CIO, CTO, COO, CMO, Etc.), President or CEO, or owner; were removed from the sample.
Thirty-one participants were removed as they self-selected their level in their organization as
above the “Director” level. Out of the remaining responses, 163 participants were observed to
have completed the survey through at least section IV, which meant they completed the primary
demographic items and the primary scales.
To increase internal reliability, the data was further paired down to only include
participants who selected that they were currently working at least 40 hours per week, and whose
selections for the survey items with Likert type scoring, selected strongly agree, agree, disagree,
or strongly disagree. The result of the effort was a final sample of 64 total viable participant
surveys for data analysis. While the final sample was much smaller than the initial survey
participant total, removing participants who did not know or preferred not to answer on the
critical items and scales removed the possibility of low internal consistency for conducting
ANOVA or regression analyses.
General Demographics of Participants
Overall, there was a generally even split of participants who identified as female (n = 31)
or male (n = 32) with one participant identifying as non-binary. Participants ranged in age from
18 to above 65 with a somewhat normal distribution of participants between 24 and 44 years of
age with roughly 44% (n = 28) of participants below 35 and 56% (n = 36) above. From an ethnic
and racial diversity perspective, there was an underrepresentation of those who identified as
67
other than white with only 4.7% (n = 3) identifying as from Hispanic origin, and only 20.3%
(n = 13) identifying as Asian, Black or African American, or Biracial or Multiracial.
Fourteen-point one percent of participants identified as military veterans. As for income,
the results were generally stratified across brackets with 57.1% (n = 36) earning below $100,000,
and 42.9% (n = 27) earning above $100,000, with one participant abstaining from answering.
Table 3 displays the descriptive statistical results of the general demographic data of participants.
68
Table 3
General Demographic Data and Frequencies
Variable Characteristic n % N
Gender Female 31 48.4
Male 32 50.0
Non-binary 1 1.6
Age 18–23 1 1.6
24–29 15 23.4
30–34 12 18.8
35–39 11 17.2
40–44 13 20.3
45–49 3 4.7
50–55 7 10.9
56–64 1 1.6
65+ 1 1.6
Hispanic origin Yes 3 4.7
No 61 95.3
Race Asian 7 10.9
Black or African American 3 4.7
Biracial or multiracial 3 4.7
White 51 79.7
Veteran status Yes 9 14.1
No 55 85.9
Personal annual income before taxes $30,000 to $39,999 6 9.4
$40,000 to $49,999 3 4.7
$50,000 to $59,999 6 9.4
$60,000 to $69,999 4 6.3
$70,000 to $79,999 4 6.3
$80,000 to $89,999 7 10.9
$90,000 to $99,999 6 9.4
$100,000 to $149,999 19 29.7
$150,000 or more 8 12.5
I prefer not to answer 1 1.6
69
Employment and Occupational Data of Participants
Given the constraints of the final participant sample all participants reported working at
least 40 hours a week or more. For work experience, there was strong representation from
participants with six to 10 years of experience (18.8%; n = 12), 11 to 15 years (21.9%; n = 14),
and 20 plus years of experience (35.9%; n = 23). Most workers (40.6%; n = 26) were mid-level
employees (non-managers), with managers representing the second largest group (32.8%;
n = 21). Forty-five-point three percent of workers (n = 29) were employed with a private
company or business and 31.3% (n = 20) worked for publicly traded companies. From an
industry perspective, there was a generally robust representation across 10 categories with
professional, scientific, and technical services representing the largest group (23.4%; n = 15),
health care and social assistance as the second largest (18.8%; n = 12), and all other categories
having between 1.6% (n = 1) to 7.8% (n = 5) representation.
Most workers (53.1%; n = 34) came from companies with less than 1000 employees with
42.4% (n = 25) coming from companies with over 1000, and 7.8% (n = 5) abstaining from
answering. Ten workers (15.6%) indicated that they worked more than one job, with the
remaining workers from the sample reporting as working only one job. There were three almost
evenly split groups among workers indicating less than two years (23.4%; n = 15), between two
to five years (23.4%; n = 15), and six to ten years (25%; n = 16) in their current occupation.
As far as occupational representation, there were 62 out of 64 participants who self-
described their occupation (Table 5). Prior to analysis, the self-described text responses were
evaluated and grouped into like types where possible (e.g., combining similar responses such as
“business consultant” and “consultant”). Responses then were counted for enumeration and
recorded. The exercise resulted in a broad representation of 44 occupations as shown in Table 4,
70
with the largest groupings falling under consulting or finance which had six workers from each
occupation respectively. Tables 4 and 5 display the descriptive statistical results of employment
and self-described occupational data of participants, respectively.
71
Table 4
Demographic Employment Data and Frequencies
Variable Characteristic n % N
Length of work experience More than 1 year, but less than 3 years 4 6.3
3–5 years 7 10.9
6–10 years 12 18.8
11–15 years 14 21.9
16–20 years 4 6.3
20+ years 23 35.9
Level in organization Entry level (non-manager) 7 10.9
Mid-level (non-manager) 26 40.6
Manager 21 32.8
Senior manager 4 6.3
Director 6 9.4
Place of employment Publicly traded company 20 31.3
Private not-for-profit, tax-exempt, or
charitable organization
5 7.8
Private company or business 29 45.3
State or local government employee 7 10.9
Federal government employee (Non-Military) 1 1.6
Active-duty U.S. Military 2 3.1
Industry Arts, entertainment, and recreation 3 4.7
Education services 5 7.8
Finance and insurance 4 6.3
Health care and social assistance 12 18.8
Manufacturing 3 4.6
Other public services (except public
administration)
4 6.3
Public administration 5 7.8
Professional, scientific, and technical services 15 23.4
Transportation and warehousing 3 4.7
All other industries with n = 1 (1.6% each) 10 15.6
Size of organization Up to 1,000 employees 34 53.1
Between 1,000 & 10,000 11 17.2
Between 10,000 & 25,000 6 9.4
Above 25,000 8 12.5
No answer 5 7.8
Number of jobs I work only one job 54 84.4
I work two jobs 10 15.6
72
Variable Characteristic n % N
Length of time in current
occupation
Less than 2 years in same occupation 15 23.4
2–5 years 15 23.4
6–10 years 16 25
11–15 years 5 7.8
16–20 years 5 7.8
20+ years 8 12.5
73
Table 5
Self-Described Occupations
Occupation n Occupation n Occupation n
Administrative assistant 1 Exercise physiologist 1 Program management 1
Analyst 1 Financial planning;
advisory
1 Public information
manager
1
Animator 1 Finance (Analysis,
management,
compliance)
6 Public safety 1
Bank manager 1 Higher education
administrator
1 Quality assurance
analyst
1
Case management 2 Lobbying 1 Real estate accountant 1
Civil engineering project
manager
1 Marketing 2 Recruiter 1
Clinical laboratory scientist 1 Medical assistant 1 Registered nurse 1
Communications director 2 Medical biller 1 Sales 1
consultant 6 Military officer 1 Software developer 1
Controller 1 Operations manager 1 Sonographer 1
Credit products officer 1 Pharmacist 1 Supervisor 1
Data coordinator, data
analyst
1 Physician assistant 1 Systems engineer 1
Database administrator 1 Political advertiser 1 Training 2
Education coordinator 1 Product manager 3 University instructor 1
Educator 1 Production or warehouse 3
Note. There was a total of 62 participants who described their occupation N = 62.
74
Educational Data of Participants
As far as educational representation, a majority (50%; n = 32) of participants had a
master’s degree or above, which included three who had a doctoral degree (Table 6). Degree
types ranged across 45 different types with strong representation across Science, Technology,
Engineering, and Math (STEM) (46.2%; n = 12) and Liberal Arts (42.3%; n = 11) for bachelor’s
degrees (n = 26), and Business (53.6%; n = 15) and Liberal Arts (35.7%; n = 10) for master’s
degrees (n = 28) (Table 7). Twenty-three-point four percent (n = 15) of workers had a technical
background which ranged from Computer Science or IT related degrees or certifications (n = 9),
and other engineering or technical certification or courses outside of computer science (n = 6)
(Table 6). From a diversity standpoint, only three participants (4.7%) indicated they were the
first in their family to attend college, and seven (10.9%) and 15 (23.4%) participants noted they
were the first in the family to graduate with a bachelors or master’s degree respectively (Table
6). Table 6 and 7 display the descriptive statistical results of general education and self-described
degree data of participants, respectively.
Table 6
General Education Data and Frequencies
Variable Characteristic n % N
Highest level of education
completed
Less than a bachelor’s degree 6 9.4
Bachelor’s degree 26 40.6
Master’s and above 32 50.0
Order in family to attend college First in my family to attend college 3 4.7
First in my family to graduate college 7 10.9
First in my family to obtain a graduate
degree
15 23.4
Not the first in my family for any of the
reasons listed
39 60.9
Indicated having technical
background or certification
Yes 15 23.4
No 49 76.6
75
Table 7
Self-Described Degree Data
Category Description n % N
Associate’s degree type
(n = 1 self-described)
Visual communication 1 100
Bachelor’s degree type (n
= 26 self-described)
STEM (computer science, math, engineering,
biology, medical sonography)
12 46.2
Liberal arts (theology, education, art, history,
English, psychology, political science)
11 42.3
Business 3 11.5
Master’s degree type
(n = 28 self-describe)
Business (MBA, tax, finance, management, or
other business type)
15 53.6
Liberal arts (national security, criminal justice,
journalism, literary studies, media studies,
geopolitics, theology, divinity, education)
10 35.7
Sciences (physiology and kinesiology, public
health, general science)
3 10.7
Doctoral degree type
(n = 3 self-described)
Education 1 33.3
Leadership 1 33.3
Pharmacy 1 33.3
76
Results Research Question One: Are There Differences in an Individual Worker’s
Perception and Understanding of Automation Based on Their Demographic Background?
Research question one was developed to explore if there were differences between an
individual worker’s perception and understanding of automation based on their demographic
background. In order to answer the research question, 12 independent demographic variables and
three dependent variables were selected to be compared. The 12 independent variables that were
used to answer the research question included gender, age, race, education, amount of technical
education, length of time in current occupation, length of work experience, number of current
jobs, level in organization, exposure to prior automation, organizational size, and income. The
dependent variables were represented by scales including (a) Understanding of Impact of
Automation (UIA) (Jobs and Tasks); (b) Perception of Career Changes (PCC) (Role, Job,
Career); and (c) Perception of Skills Changes (PSC) (Skills, Education, Training).
Based on the sample size of participants (n = 64), the independent demographic variable
results were recoded to enable the use of one-way bivariate ANOVAs as multivariate analysis
was not possible. In preparing the data for analysis, the 12 independent demographic variables
were evaluated for a logical cut off point to conduct the ANOVA. For nominal items, other than
gender, that had more than two original categories, the median was used to bifurcate the
categories. For items such as organizational size, or income, the results were bifurcated using the
most logical numerical split based on the number of participant responses.
The three scales were compared against each of the independent variables (factors) in
SPSS using the one-way ANOVA to determine a relational significance above .05 and Tukey’s
post-hoc tests to examine the specific areas where statistical differences existed. A one-way
ANOVA was conducted, using each of the three scales to separately compare whether an
77
individual’s demographic background had an effect on their level of understanding of the impact
of automation, their perception of potential career changes, and perception of potential skills,
education or training. The result of the ANOVA included two groups that had a significant effect
at the p < .05 level. These groups were the number of current jobs and Understanding of Impact
of Automation (p = 0.011), and race and Perception of Skills Changes (p = 0.002) (Table 8). The
post-hoc comparisons using the Tukey test for mean score were not presented in this analysis due
to the small sample size.
While the comparison of race and perception of needed skills’ changes resulted in the
most significant p value (p = 0.002), the difference could be attributed to the limitation of the
small sample size, and breakdown of the bivariate groups. For race, the participant sample
included 79.7% (n = 51) White participants, and 20.3% (n = 13) Asian, Black or African
American, Biracial or Multiracial (Table 3). Number of jobs and understanding the impact of
automation had a similar limitation with 84.4% of participants (n = 54) having only one job,
whereas 15.6% (n = 10) reported having two jobs (Table 4). Therefore, the results are
inconclusive as to whether race has an effect on the Perception of Skills Changes as a result of
automation, or if working one or two jobs has an effect on understanding the impact of
automation.
The remainder of the ANOVA results, based on the survey sample, indicate that
demographics had no specific effect on Understanding of Impact of Automation, Perception of
Potential Career Changes, or Perception of Potential Skills, education or training. Based on the
literature review and cut-off numbers, the categories that could have resulted in potential
differences would have been with gender, level in organization, and education. For gender, there
was almost an even split with reported female participants at 48.4% (n = 31), male with 50%
78
(n = 32) and one non-binary (Table 3). For level in organization, 51.6% participants (n = 33)
reported being a mid-level (non-manager) worker or below, and 48.4% (n = 31) reported they
were managers or above (Table 4). As for education, almost the entire sample (91.6%; n = 58)
had at least a bachelor’s degree, with 50% (n = 32) associated with a bachelor’s degree or below,
and 50% (n = 32) with a masters or above (Table 4). Table 8 displays the results of the analysis
of variance (ANOVA) for demographics, Understanding of Impact of Automation (UIA),
Perception of Career Changes (PCC), and Perception of Skills Changes (PCC).
79
Table 8
Impact of Automation, Career Changes, and Skills Changes ANOVA Results
Independent variable
Understanding of
impact of automation
(p-value)
Perception of career
changes (p-value)
Perception of skills
changes (p-value)
Gender .237 .611 .163
Age .117 .231 .468
Race .639 .574 .002*
Education .454 .947 .880
Technical education .669 .885 .128
Length of time in
current occupation
.773 .246 747
Total length of work
experience
.978 .794 .977
Number of current
jobs
.011* .703 .360
Level in organization .788 .253 .589
Exposure to
automation
.102 .842 .164
Organization size .500 .932 .458
Income .643 .365 .800
* p value significant at p < .05 level.
80
Results Research Question Two: Are There Differences in an Individual Worker’s Level of
Acceptance of Change and Acceptance and Use of Automation Based on Their
Demographic Background?
Research question two was designed to evaluate if any differences existed between an
individual worker’s level of accepting change and level of acceptance and use of technology
based on their demographic background. The 12 independent demographic variables that were
used for research question one were used in addition to two different dependent variables for the
comparisons. The dependent variables were represented by two complete scales including
Acceptance of Change (AC), and Acceptance and Use of Automation (AUA).
The process for preparing the independent demographic variables for one-way bivariate
ANOVAs also remained the same as research question one and the post-hoc comparisons using
the Tukey test for mean score were not presented due to small sample sizes in specific
categories. The sub-scales from AC and AUA were combined to produce a scale to enable the
analysis. The two scales were then run through SPSS against the independent demographic
variables using the one-way ANOVAs to determine if an individual’s demographic background
had an effect on their acceptance of change or acceptance and use of automation.
Based on the ANOVA test results, no groups reflected statistical significance above the
p < .05 level (Table 9). The two groups that had results close to the 95% confidence interval
included level in organization and AC (p = .078), and number of current jobs and AUA
(p = .064). The results indicate that there was no observable difference in the effect the
independent demographic variables had on acceptance of change or the acceptance and use of
automation. As with the results in research question one, the lack of significant differences could
be a result of the sample, which was comprised of workers who roughly shared very significant
81
personas across education and experience levels. Table 9 displays the results of the analysis of
variance (ANOVA) for demographics, Acceptance of Change (AC), and Acceptance and Use of
Automation (AUA).
Table 9
Acceptance of Change (AC) and Acceptance and Use of Automation (AUA) ANOVA Results
Independent variable
Acceptance of change
(AC) (p-value)
Acceptance and use of
automation (AUA)
(p-value)
Gender .233 .531
Age .725 .626
Race .185 .601
Education .187 .565
Technical education .960 .992
Length of time in current occupation .537 .369
Length of work experience .501 .750
Number of current jobs .367 .064
Level in organization .078 .439
Exposure to automation .294 .325
Organization size .624 .458
Income .949 .431
82
Results Research Question Three: What Are the Perceptions of Individual
Workers Related to Automation?
The third research question was designed to determine descriptive data for individual
perceptions and general understanding of automation. Survey questions and items sought data on
a worker’s prior exposure to automation, level of Understanding of the Impact of Automation
(jobs and tasks), Perceived Career Changes (role, job, career), Perceived of needed Skills
Changes (skills, education, training), perception of whether workers or employers were
responsible for providing training on new automation, expectation to receive training for new
automation, and behavioral intent of workers to improve the workplace through automation.
Analysis was conducted on the survey items from the Acceptance and Use of Automation
(AUA) scale that probed for Performance Expectancy (PE); Perceived Usefulness (U), Effort
Expectancy (EE); Perceived Ease of Use (EOU), Attitude Toward Using Automation (ATUA);
Anxiety (ANX), and Facilitating Conditions (FC); Compatibility (C) and Perceived Behavioral
Control (PBC). The tables in each respective section provide descriptive data associated with
each survey item or question and associated written analyses.
83
Personal Exposure to Prior Automation
In seeking to measure general understanding of automation, the survey asked participants
to indicate their prior personal exposure to automation. Eighty-four-point-four percent (84.4%)
of participants reported they could think of a time when automation was implemented in their
workplace (Table 10). Given the sample size, this adds credibility for confidence in participants
responding to subsequent survey items and questions specifically related to automation.
Table 10
Personal Exposure to Prior Automation
Response n %
I can think of a time when new automation was implemented in my
workplace
54 84.4
I am not aware of any new automation being put in place at any of
my workplaces for which I have worked
10 15.6
84
Behavioral Intent to Improve the Workplace Through Automation
In addition to asking participants about prior exposure to automation, the survey sought
data to gauge the potential behavioral intent of participants as to whether they would seek to
improve their workplace through automation, given the opportunity. Almost the entire sample
(93.8%, n = 60) of participants indicated that they would always, almost always, or at least
sometimes make a suggestion to improve things in their workplace with new automation, with
only 6.3% (n = 4) participants indicated that they almost never make a suggestion (Table 11).
This provides additional validity that most participants had an understanding of automation,
regardless of whether or not they had experience automation in their workplace (as discussed
above).
Table 11
Behavioral Intent to Improve the Workplace Through Automation
Variable Response n %
If I identified an opportunity to
improve things in my
workplace through new
automation, I would:
Always make a suggestion 20 31.3%
Almost always make a suggestion 21 32.8
Sometimes make a suggestion 19 29.7
Almost never make a suggestion 4 6.3
Note. Items in italics represent survey anchors.
85
Understanding of Impact of Automation, Jobs and Job Tasks
Results across Understanding of Impact of Automation (UIA) for jobs and tasks were
generally positively skewed toward participants indicating an understanding of both the types of
jobs and tasks that may be affected in general as well as how their own job or tasks could be
affected. The data reflect that 89% of participants or greater indicated they had an understanding
of both the general types of jobs and tasks that could be affected or replaced as well as how their
job could be affected (Table 12). Means and standard deviations across all survey items reflected
higher Liker-type selections across strongly agree and agree, with several items reflecting
identical percentages. There were no responses reflecting strong disagreement with any item
statement, and there was only one response of do not know for understanding types of jobs that
might be affected. Only 9.4% (n =6) and 10.9% (n =7) of participants disagreed to having an
understanding of how their job may be affected and the tasks or duties with a job that could be
affected or replaced by automation. Table 12 displays the descriptive statistical results of
Understanding of Impact of Automation (UIA) for jobs and job tasks.
86
Table 12
Understanding of Impact of Automation (UIA), Jobs and Tasks Descriptive Data
Variable
In general, I
understand the
types of jobs that
may be affected
by automation.
I understand how
my job may be
affected by
automation.
In general, I
understand the
duties or tasks
within a job that
may be replaced
by automation.
I understand the
duties or tasks
within my job
that that may be
replaced by
automation.
M 1.80 1.73 1.86 1.77
SD .671 .623 .587 .556
n % n % n % n %
Strongly agree 19 29.7 23 35.9 16 25.0 19 29.7
Agree 41 64.1 35 54.7 41 64.1 41 64.1
Disagree 3 4.7 6 9.4 7 10.9 4 6.3
Strongly disagree 0 0 0 0 0 0 0 0
Do not know 1 1.6 0 0 0 0 0 0
Note. Underlined and italicized words reflect the emphasis used in the survey to help participants
distinguish the difference between the intended purpose of survey items.
87
Perception of Career Changes (PCC), Role, Job, and Career
As far as Perception of Career Changes, results indicate that most participants disagreed
that automation would cause them to change roles, jobs, or their career. The results were mixed
across survey items for the strongly agree choice, with 12.5% (n = 8) indicating automation
would cause them to change roles, but only one response (1.6%) for changing jobs, and no
responses (0%) for needing to make a career change (Table 13). The results indicate that there
were far more participants who indicated that automation would cause a change in their job role,
but not necessarily a change in their job or career. Table 13 displays the descriptive statistical
results of Perception of Career Changes (PCC) for role, job, and career changes.
Table 13
Perception of Career Changes (PCC), Role, Job, and Career Descriptive Data
Variable
Automation will cause
me to have to change
roles within my
workplace.
Automation will cause
me to have to change
jobs.
I expect to make
career changes due to
automation.
M 2.66 3.25 3.20
SD .859 .735 .780
n % n % n %
Strongly agree 8 12.5 1 1.6 0 0
Agree 13 20.3 5 7.8 12 18.8
Disagree 37 57.8 38 59.4 29 45.3
Strongly disagree 5 7.8 17 26.6 21 32.8
Do not know 1 1.6 3 4.7 2 3.1
Note. Underlined and italicized words reflect the emphasis used in the survey to help participants
distinguish the difference between the intended purpose of survey items.
88
Perception of Skills Changes (PSC), Skills, Education, and Training
Within needing to acquire new skills or education to perform their job, more participants
indicated needing new skills rather than education as a result of automation. This is reflected by
the 54.7 % (n = 35) of participants agreeing or strongly agreeing with needing new skills, versus
31.3% (n = 20) across the same selections for needing to acquire additional education. In
addition, there were zero selections for do not know for needing new skills but were three
participants (4.7%) paired with education (Table 14). For this sample, the results indicate that
more participants believed they would need new skills, but not necessarily additional education.
The lack of support reflected for needing new education, may be due to the fact that the sample
population had already obtained college degrees. A chi-square analysis was run, but no
statistically significant differences emerged, possibly due to the total survey sample size
(N = 64). Table 14 displays the descriptive statistical results of Perception of Skills Changes
(PSC) for skills, education, and training.
89
Table 14
Perception of Skills Changes (PSC), Skills, Education, and Training Descriptive Data
Variable
I will have to acquire new skills to
perform my job as a result of
automation.
I will have to acquire additional
education (certification or college)
to perform my job as a result of
automation.
M 2.48 2.95
SD .908 .933
n % n %
Strongly agree 8 12.5 3 4.7
Agree 27 42.2 17 26.6
Disagree 19 29.7 27 42.2
Strongly disagree 10 15.6 14 21.9
Do not know 0 0 3 4.7
Note. Underlined and italicized words reflect the emphasis used in the survey to help participants
distinguish the difference between the intended purpose of survey items.
90
Perception of Acceptance and Use of Automation
Results across the scale means of the items from the Acceptance and Use of Automation
(AUA) broadly reflected the general acceptance of automation (Table 15). Performance
Expectancy (PE), Perceived Usefulness (U) items probed for participants perspective on
accomplishing tasks more quickly, improving performance, increasing productivity, enhanced
effectiveness, and making the job easier. Performance Expectancy (PE) and Perceived
Usefulness (U) had a mean of 1.74, indicating positive skewedness toward participants
perceiving that automation would be useful to them. Effort Expectancy (EE), Perceived Ease of
Use (EOU), had a mean of 1.84, also indicating positive skewedness and favorable outlook
around items that probed for ease of learning how to operate automation, getting new automation
to perform, becoming skillful in using automation, and a perception that automation would be
easy to use.
As far as Attitude Toward Using Automation (ATUA), Anxiety (ANX), items measured
apprehension, fear of losing information, hesitancy due to uncorrectable mistakes, and being
intimidated to use new automation. Attitude Toward Using Automation (ATUA) was reverse
coded when run through SPSS to enable accurate comparison to other means and standard
deviations. A mean of 2.05 indicates that most participants selected disagree, meaning that they
were not fearful of using new technology.
Facilitating Conditions (FC), Compatibility (C) items probed for whether participants
needed automation to be compatible with most aspects of their work, fit well with the way they
liked to work, and fit into their work style to use it. A mean of 2.20 on a one to four scale
indicates a strong negative skewedness and that most participants did not believe that automation
would have to be compatible for them to begin using it. This result means that most participants
91
indicated that they have the capacity to accept new forms of automation even if it does not fit
with their current ways of working.
Facilitating Conditions (FC), Perceived Behavioral Control (PBC) items asked
participants about their perceived control over using new automation, and whether they felt they
had resources and knowledge necessary, or if having the right resources, opportunities, or
knowledge would make it easy to use automation. A mean of 1.99 on a one to four scale
indicates that more participants agreed that they either had control, or if given the right
resources, knowledge, or opportunity, they would feel they had control over using new
automation. The results indicate that participants had a generally positive perception around their
overall control of using new automation. Table 15 displays the descriptive statistical results from
the survey items for Acceptance and Use of Automation (AUA).
Table 15
Items From Acceptance and Use of Automation (AUA) Descriptive Data
Variable
Performance
expectancy
(PE); perceived
usefulness (U)
Effort
expectancy
(EE); perceived
ease of use
(EOU)
Attitude toward
using
automation
(ATUA);
anxiety (ANX)
Facilitating
conditions (FC);
compatibility
(C)
Facilitating
conditions (FC);
perceived
behavioral
control (PBC)
M 1.74 1.84 2.05 2.20 1.99
SD .578 .517 .595 .513 .467
Note. Scale items only represented participants who selected strongly agree, agree, disagree, or
strongly disagree resulting in a 1 to 4 scale.
92
Responsibility for Providing Training on New Automation
There were two primary questions around responsibility to provide training. The first
question asked participants to use a sliding scale to select between zero and 100% as to whether
they felt the worker or the employer should be responsible for providing training. The second
question asked participants to indicate if they were expecting to receive training in conjunction
with existing or near-term automation that would be put in place. In preparation for analysis,
percentage responses for the first question were grouped into quartiles for question one, and
frequencies and percentage results of participants were put into a table.
As far as the results for worker or employer responsibility for providing training, 86%
(n = 55) of participants reported that their employer should be primarily responsible, with over
half (56.3%; n = 36) indicating their employer was 75 to 100% responsible (Table 16). For the
second question, over half (56.3%; n = 36) of workers indicated that they were expecting to
receive training by their employer for how to use new automation (Table 17). Only (9.4%)
reported that they did not expect to receive any training from their employer for new automation.
The results are a strong indication that the participants in the survey believed that the employer
should be the primary provider of training paired with new automation in the workplace. While
over half of participants reported that they were expecting training in conjunction with
automation, it is noteworthy that over one third (32.8%) of participants were not expecting
(20.3%) or were not aware of any new automation (12.5%). Table 16 displays the descriptive
statistical results from perception of responsibility for providing training on new automation.
Table 17 displays the descriptive statistical results from the percentage of participants who
expected to have automation implemented in their workplace and to receive training for the
automation.
93
Table 16
Perception of Responsibility for Providing Training on New Automation
Quartile percentage n %
0 to 24.99% (worker completely responsible) 1 1.6
25 to 49.99% 5 7.8
50 to 74.99% 19 29.7
75 to 100% (employer completely responsible) 36 56.3
Did not know or preferred not to answer 3 4.7
94
Table 17
Percentage Expecting to Have Automation Implemented and to Receive Training
Response n %
I am expecting to receive training provided by my employer for how
to use new automation
36 56.3
I am not expecting to receive any training from my employer for the
new automation
6 9.4
My current workplace is not implementing any new automation 13 20.3
I am not aware of any new automation being implemented 8 12.5
I do not know 1 1.6
95
Results Research Question Four: Which Tasks and What Percentage of One’s Job Does an
Individual Worker Perceive can be Automated?
The purpose of research question four was to understand workers’ perceptions about the
tasks or duties of their job that they believed could be performed by automation. Survey
questions asked participants to indicate a percentage of tasks or job duties that they believe could
be automated, those that had already been automated, and to self-describe the words or phrases
for the types of automation they believed could be replaced. The results are described in the
subsequent sections.
Percent of Tasks or Duties Deemed Replaceable or Already Replaced
For this portion, participants were asked to provide a percentage on potential tasks or
duties deemed potentially replaceable or already replaced. There were two questions that asked
participants for (a) what they believed could be a reality in the future; and (b) what had already
transpired in the way of tasks or duties already replaced by automation. Survey items asked
participants to indicate a percentage of agreement (e.g., “move the slider to indicate what percent
you believe x or y”) in order to provide stratification of choice on a scale. In preparing for the
analysis, responses were recoded into quartiles using logical brackets to allow for meaningful
groupings.
For the first question, participants were asked for their perception of what could be the
reality of tasks or duties being replaced, rather than their current reality of what had already been
replaced. Forty-six percent (n = 26) of participants reported that they believed 25 to 49.99% of
their daily job-related tasks or duties could be performed with an automated technology, 31.3%
(n = 20) reported zero to 24.99%, 18.8% (n = 12) reported 50 to 74.99%, and 7.8% (n = 5)
reported 75 to 100% could be performed with an automated technology (Table 18). The data
96
indicate that although most participants (71.9%; n = 46) believed that less than 50% of their job
could be automated, 26.6% believed that over 50% of their job could be automated. This is a
potentially significant result given that the sample of participants were professionals with college
degrees, and nearly one third believed over 50% of their daily job could be automated.
For the second question, participants were asked to indicate a percentage of tasks or
duties that had already been replaced. As far as tasks and duties replaced, 68.8% (n = 44) of
participants indicated between zero and 24.99%, 17.2% (n = 11) indicated between 25 and
49.99%, 7.8% (n = 5) indicated between 50 and 74.99%, with zero indicating above 75%, and
6.25% (n = 4) selecting they did not know (Table 18). This is an interesting result when
compared to the fact that over two-thirds of participants (67.2%; n = 43) reported their
perception that at least 25% or more of their job could be automated in the future. The data
indicate that workers not only believe that at least a quarter of their job could be automated, but
also that nearly one third (25%; n =16) of participants indicated that at least 25% of their job had
already been automated. Table 18 displays the descriptive statistical results of participant
responses for percent of tasks or duties deemed replaceable or already replaced.
97
Table 18
Percent of Tasks or Duties Deemed Replaceable or Already Replaced
Variable Quartile percentage n %
Self-selected participant results on a scale of
0% to 100% for the percentage of daily job-
related tasks or duties perceived that could
be performed with an automated technology
0% to 24.99% 20 31.3
25 to 49.99% 26 40.6
50 to 74.99% 12 18.8
75 to 100% 5 7.8
I do not know 1 1.6
Self-selected participant results on a scale of
0% to 100% for the percentage of daily job-
related tasks or duties in their current
workplace that have already been replaced
by automation
0% to 24.99% 44 68.8
25 to 49.99% 11 17.2
50 to 74.99% 5 7.8
I do not know 4 6.25
98
Participant Self-Described Types of Tasks or Duties Deemed Replaceable
Participants were also asked to provide a percentage on tasks or duties already replaced
by automation. In describing tasks or duties deemed replaceable by automation in the future,
71.9% (n = 46) of participants provided a detailed self-described open-ended response, 14.1%
(n = 9) did not foresee any replacement, 12.5% (n = 8) did not know, and only 1.6% (n = 1)
preferred not to answer (Table 19). The resulting 71.9% of participants that were able to describe
potential tasks or duties is substantial given that the question allowed for participants to select do
not know or prefer not to answer.
In preparing the self-described open-ended data, responses were cleaned, coded, and
grouped into logical categories. The cleaning and coding involved listing all responses and
correcting spelling errors and synthesizing descriptions. The listed descriptions were then
evaluated and grouped based on similar functions and then bucketed into categories based on
their related function. The exercise resulted in nine categories and 43 tasks or duties that
participants reported they believed could be performed by automation (Table 20). Based on the
number and variety of self-described tasks or duties, the data indicate that the majority of
participants had an understanding of at least one type of task or duty that could be performed by
automation. This is an important data point in context of the entire survey in that it provides
additional validity in that almost three-quarters of participants self-described types of tasks or
duties they believed could be replaced. The most numerous responses fell under data processing,
communication, and data management respectively. Tables 19 and 20 display the descriptive
statistical results of self-described results of daily tasks or duties deemed replaceable, and self-
described responses for the types of tasks or duties that participants believed could be replaced,
respectively.
99
Table 19
Self-Described Results of Daily Tasks or Duties Deemed Replaceable
Response n %
Participant self-described 46 71.9
I do not foresee any tasks or duties that could be replaced 9 14.1
I do not know 8 12.5
I prefer not to answer 1 1.6
100
Table 20
Self-Described Responses of Types of Tasks or Duties Deemed Replaceable
Category Responses
Application automation Software updates
Communication
Digital chat assistants, phone answering systems, auto sending
emails, merging email data, sending reports, voice to text,
scheduling and notification of meetings, events
Data analysis Managing data sets, resume scanning, trend analysis, predictive
models, budget projection, policy analysis
Data gathering Issue tracking, compiling data, time keeping, collecting
performance data
Data management Reconciliation, storing and retrieval, keeping attendance and
records, timesheets, customer relationship management
Decision making processes Low level repetitive tasks
Data processing Data entry, form completion, review, data scrubbing and
sorting, medical coding, generating reports, creating invoices,
grading papers or exams, purchase requests, recruiting and
HR actions, underwriting, financial statements, conducting
research
Data visualization Mapping current processes
Physical tasks Mailroom sorting, hand calculations, welding and metal
grinders, robotic cranes
101
Results Research Question Five: Is There a Relationship Between an Individual Worker’s
Perception and Understanding of the Impact of Automation, Level of Acceptance of
Change and Level of Acceptance and Use of Automation?
Research question five was designed to determine if there was a relationship between
individual workers’ perception and understanding of the impact of automation, acceptance of
change and acceptance and use of automation. In order to answer the research question, the
following scales were run through SPSS software to test for correlation: Understanding of
Impact of Automation (UIA) (Jobs and Tasks), Perception of Career Changes (PCC) (Role, Job,
Career), Perception of Skills Changes (PSC) (Skills, Education, Training), Acceptance of
Change (AC), and Acceptance and Use of Automation (AUA). There were a total of nine
variables used for the correlation analyses including the three that tested for Understanding of
Impact of Automation (UIA, PCC, PSC), five sub-scales from AUA, and the singular AC scale.
A correlation matrix in this section presents the results from SPSS to include the scale mean,
standard deviation, Cronbach’s alpha, and respective correlation.
There was a total of 36 possible correlated relationships based on the analysis conducted.
Of the 36 possible relationships, there were 15 variables that emerged as correlated, with seven
relationships that were significant at the 0.05 level, and eight that were significant at the 0.01
level (two-tailed) (Table 21). The results indicate that 41.6% (15 out of 36) of the possible
relationships were statistically significant with a 95% confidence level or greater, meaning that
the relationships were not due to chance.
Acceptance of Change (AC) had the largest share of relationships from the correlations,
with five out of eight (62.5%) possible variables reflecting significant correlations. One of the
relationships was between AC and Understanding of Impact of Automation (UIA) (r = .249,
102
p < 0.05). This points to the fact that those who reported having a positive outlook on accepting
change in general also reported having a good understanding of the impact of automation. This is
an important data point as it provides support for the basis of the entire study itself and will be
discussed in depth in chapter five.
AC was also related to Perception of Skills Changes (PSC) (Skills, Education, Training)
with a positive correlation (r = .387, p < 0.01) meaning that those who perceived needing new
skills, education, or training as a result of automation also had a positive outlook on change in
general. As far as AC and Perceived Usefulness (U) there was a positive correlation (r = .305,
p < 0.05), reflecting that those who perceived automation as potentially useful also reported a
higher willingness to embrace change. Workers who perceived automation to be easy to use
(Perceived Ease of Use, PEU) reported high levels of accepting change (AC), with a positive
correlation that reflects the relationship between the two (r = .370, p < 0.01). The last correlation
for the AC scale emerged with Compatibility (C). Acceptance of Change and Compatibility were
positively correlated (r = .297, p < 0.05), indicating that those workers who reported needing
automation to be compatible with their preference for work also reported being able to cope with
change well.
Anxiety (ANX) about using new automation was significantly correlated with four scales.
Participants who reported having a good Understanding of the Impact of Automation (UIA)
reported lower levels of Anxiety (ANX) (r = .272, p < 0.05). This is important in that ANX falls
under a reflection of the general Attitude Toward Using Automation (ATUA) which means that
better understanding of automation could lead to lower anxiety for use. Anxiety and Perception
of Career Changes (PCC) were the only negatively correlated in the entire analysis (r = -.273,
103
p < 0.05). The negative result highlights that workers who perceived needing to make career
changes as a result of automation also reported higher levels of anxiety about using automation.
Another relationship was evident between ANX and Perceived Ease of Use (EOU). The
positive correlation (r = .456, p < 0.01) indicates that workers who reported perceiving
automation as easier to use also reported lower levels of anxiety toward using the automation.
Perceived Behavioral Control (PBC) was the final observed positive correlation to ANX (r =
.391, p < 0.01). With PBC, the results indicate that workers who believe they have more control
over using the technology have lower levels of anxiety toward using it.
Perceived Behavioral Control (PBC) had three statistically significant relationships at the
99% confidence including the details previously mentioned in the relationship to ANX. Along
with lower levels of anxiety, Perceived Behavioral Control (PBC) was positively correlated with
Understanding of Impact of Automation (UIA) (r = .416, p < 0.01), indicating that workers who
reported having an understanding of the impact of automation also had a higher level of
perceived control over using the automation. The remaining resulting relationship for PBC was
between PBC and Perceived Ease of Use (EOU) with results reflecting a positive correlation
(r = .591, p < 0.01). In this case, a higher confidence in control over the automation is related to
higher confidence in automation being easy to use.
There were five observed relationships associated with Perceived Ease of Use (EOU),
including the three relationships related to Acceptance of Change (AC), Anxiety (ANX), and
Perceived Behavioral Control (PBC). Along with those previously described, EOU was
positively correlated with Understanding of Impact of Automation (UIA) (r = .438, p < 0.01),
reflecting that those who reported a good understanding of the impact of automation also
perceived automation as easier to use. Perceived Usefulness (U) and Perceived Ease of Use
104
(EOU) were also positively correlated (r = .317, p < 0.05). In this case, the relationship reflects
that workers who perceived automation as useful also perceived that it was easy to use.
As far as workers who reported the perceived need for changing skills, there were three
positively correlated results including the result described with Acceptance of Change (AC).
Other than AC, Perception of Skills Changes (PSC) (Skills, Education, Training) was also
positively correlated with Perception of Career Changes (PCC) (Role, Job, Career) (r = .489,
p < 0.01). The results indicate a relationship between workers who believed automation would
cause them to have to change roles, job, or career, also indicated they would need new skills,
education, or training. The other observed relationship was a positive correlation between PSC
and Compatibility (C) (r = .246, p < 0.05) which highlights that those who indicated a perceived
need for changing skills also indicated automation needing to be compatible with their general
ways of working.
Table 21
Table of Correlations and Significant Relationships
Variable
n
(items)
M SD
Cronbach’s
alpha (a)
UIA PCC PSC U EOU ANX C PBC
1. Understanding of impact of automation
(UIA) (Jobs & tasks)
4 1.79 .496 .829
2. Perception of career changes (PCC)
(Role, job, career)
3 3.04 .623 .691 -.018
3. Perception of skills changes (PSC)
(Skills, education, training)
2 2.72 .816 .726 .028 .489**
4. Performance expectancy (PE);
Perceived usefulness (U)
5 1.74 .578 .944 .199 -.062 .116
5. Effort expectancy (EE);
Perceived ease of use (EOU)
4 1.84 .517 .895 .438** .191 .195 .317*
6. Attitude toward using automation
(ATUA); Anxiety (ANX)
4 2.05 .595 .842 .272* -.273* -.036 .145 .456**
7. Facilitating conditions (FC);
Compatibility (C)
3 2.20 .513 .717 -.115 .120 .246* .073 -.208 -.216
8. Facilitating conditions (FC);
Perceived behavioral control (PBC)
4 1.99 .468 .838 .416** .060 .007 .179 .591** .391** -.100
9. Acceptance of change (AC) 10 1.97 .367 .781 .249* .068 .387** .305* .370** .128 .297* .238
* Correlation is significant at the 0.05 level (2-tailed).
** Correlation is significant at the 0.01 level (2-tailed).
105
106
Chapter Four Summary
Based on the methodology presented in chapter three, chapter four provided the results
and analysis for this study. To collect data for analysis, a survey was designed to produce valid
and reliable results. The administration of the survey produced 64 valid completions for
documentation and analysis. The resulting sample included a wide range of demographic data
from a host of occupations, varying levels of income and experience, who serve in different
types of roles, representing multiple educational degree types. The survey included questions and
items that probed for demographic information and asked participants to provide their level of
understanding of automation, potential impact, perception of individual responsibility to acquire
new skills, training, or education, and perceived career changes. Scales from valid and reliable
frameworks served as the basis for survey items to probe for acceptance of change, and
acceptance and use of automation.
Most participants reported having encountered automation in their workplace, could
name specific types of automation, and indicated that given the chance, they would make
suggestions for new automation when given the opportunity. The sample reflected a positive
perception about accepting and using new technology, given the right knowledge, resources, and
control over using the automation. Participants also reported having a solid understanding of the
general types of jobs that could be affected by automation as well as how automation would
affect their specific job. Almost 50% of participants believed up to 50% of their job could be
performed by automation, but did not report that automation had already replaced a great number
of tasks or duties.
As far as training and education, participants reported that they believed their employer
was primarily responsible and over 50% were expecting to receive training on new automation in
107
the future. Most participants did not believe automation would cause them to have to make
career changes, but that it would cause them to have to change their job or role. Needing to learn
new skills was selected more frequently than needing new education, but this could have been
due to the highly educated nature reflected by the sample population.
The study revealed over 15 statistically significant correlations (out of 36). Four
relationships emphasized the importance of workers having a good understanding of the impact
of automation as related to their ability to accept change, to maintain control, lower their levels
of anxiety, and to perceive automation as easy to use. Two correlations indicated that workers
who thought they would need additional skills also believed they could better tolerate general
change and that automation would cause them to make career changes. Three relationships
highlighted that those who thought they would have to change jobs or careers reported high
levels of anxiety and those who thought automation would be easier to use also perceived
automation as more useful and thought they would have more control over using it.
Worker willingness to accept change as related to how useful they thought automation
would be and how easy it was to use represented two more relationships. Two more correlations
included the relationship between lower levels of anxiety and higher levels of perceived ease of
use and greater perception of control over using automation. The last two correlations revealed
the relationship between compatibility of automation and willingness to accept change and the
need to learn new skills.
Although the research was designed to probe for differences in individual worker
perceptions and understanding of automation based on demographic background, there were no
significant results. There were also no significant results associated with differences in an
individual worker’s level of accepting change or use of automation based on their demographic
108
background. While the survey results and sample size may not have produced a large data pool
for which to conduct more complex statistical analyses, the large number of significant
correlations provide a basis for rich recommendations and discussion that will be presented in
chapter five.
109
Chapter Five: Recommendations and Discussion
Although no cause and effect can be determined based on this study, the analysis resulted
in correlations that indicated significant relationships which provides evidence for the use of this
study’s theoretical and conceptual frameworks for the support of strategic initiatives around
automation implementation. This chapter provides specific recommendations based on study
results. This chapter also discusses limitations and delimitations of this study and describes the
needs for future research. Lastly, the chapter provides ethical considerations around
implementing automation and balancing the need for profit, productivity, and social good.
Although automation will require a specific implementation plan, as highlighted in the chapter
two literature review, recommendations do not include specifics for implementation planning.
Rather, recommendations in this section provide high level strategies to support potential greater
levels of worker and future worker acceptance and use automation.
Recommendations for Practice
There are three primary recommendations based on the results of the study which are
described in this section. Recommendation one is to use educational interventions to support
knowledge and skill development to improve future acceptance and use of automation. The
second recommendation is to support acceptance and use automation through enhancing
perceived value of automation and worker efficacy. Recommendation three is to develop and
enhance organizational systems and processes to support the acceptance and use of automation.
These recommendations are supported by the research provided in chapter two, the methodology
used to guide this study as described in chapter three, and the results discussed in chapter four.
Grounded in research, this section will provide an overview of the three recommendations
designed to support worker acceptance and use of automation.
110
Recommendation 1: Use Educational Interventions to Support Knowledge and Skill
Development to Improve Future Acceptance and Use of Automation
The correlation results of this study revealed nine (out of 15) significant relationships,
underscoring the importance of educating workers to support the acceptance and use of
automation. Of the correlations, there were four relationships that underscore the importance of
workers having a solid understanding of how automation might impact them and their ability to
accept change, their belief that they could maintain control of their environment, maintaining
lower levels of anxiety in conjunction with automation implementation, and their perception of
how easy automation is to use. The strength of the correlations around the importance of workers
having and understanding the impact of automation was also supported by most study
participants (89% or more) reporting that they had an understanding of both the general types of
jobs that could be affected by automation as well as how their job could be affected by
automation. The level of understanding reported by participants was underscored by the fact that
over 71% of participants also provided a written response for the type of automation they
believed could replace or augment their work. This means that most of the workers in this study
not only had a good understanding of how they would be affected, but provided specific written
examples, which subsequently provides additional validity to the importance of the correlation
results around the need for workers to understand the impact of automation.
Other significant relationships that emerged from this study provide additional support
for the need for a focus on knowledge and skill development. Workers who perceived that
automation would result in skills’ changes also reported higher levels of being able to accept
change and a belief that they would need to potentially change jobs or careers. Additionally,
workers who perceived having to change jobs or careers also reported higher levels of anxiety.
111
Those who perceived automation as easier to use also perceived automation as more useful and
reported that they would have more control over using it.
Focus on Improving Knowledge and Skill Around Automation Earlier in Educational Systems
Given these results, there is an opportunity for both organizations and governmental
entities to institute educational interventions to expose future workers to automation earlier in the
talent pipeline. These educational interventions could start as early as realistically possible, even
at the grade school (K–12) level, given the right conditions. More specifically, future workers
will need to have (a) the knowledge of how automation will impact their job and job tasks, as
well as; (b) an understanding of the types of skills, education, and training they will need to help
support adoption and use. While this recommendation is focused on future workers, many of the
concepts are appropriate for use in today’s organizations with the current workforce. The
following will provide suggestions for the types of educational interventions including
knowledge and associated skill exposure for preparing workers for future automation.
There are two primary types of knowledge: declarative knowledge, which includes
concepts, principles, or facts, and procedural knowledge, which centers around strategies or how
to accomplish something (Sternberg, 2017). Both declarative knowledge and procedural
knowledge will be an important part of the knowledge process to help future and current workers
understand new automation and how it will affect them. Learning new knowledge or skills is
typically supported by one or more of four knowledge and skills’ enhancement approaches
presented by Clark and Estes (2008). The four approaches to enhancing knowledge and skills
include (a) providing information; (b) job aids; (c) providing training; or (d) education. Within
the context of educational interventions, any of the four approaches may be appropriate in
conjunction with the others, or on their own, depending on the educational level or setting.
112
Leaders who are responsible for designing educational programs will need to incorporate each
into the learning design as appropriate.
Leverage Information, Job Aids, Training, and Education to Enhance Knowledge and Skill
Around Automation
Research has shown that learning leads to knowledge, and knowledge in turn helps to
facilitate and reinforce additional knowledge (Sternberg, 2017). The concept of knowledge
building and reinforcement highlights a significant opportunity to consider using enhancing
existing educational systems to gradually expose students to increased automation and
complexity. With regard to knowledge, adults are often unaware or reluctant to admit where they
are lacking specific skills (Clark & Estes, 2008), but with educational programs, students could
build confidence in their skills around automation before entering the workforce. Using
education to facilitate this process would enable continuous evaluation of skill gaps and
gradually increasing exposure to new skills and more complex forms of automation.
As discussed in the literature review, automation comes in many forms and varies in
complexity. Determining the most appropriate approach to enhancing knowledge and skills can
be thought of in terms of complexity. As a part of planning for the use of any of the methods
described in the following sections, educational planning efforts will need to incorporate
evaluation of (a) the type of automation; and (b) the level of complexity of the specific
automation. Automation with the least complexity is likely only going to require minimal
amounts of information, job aids, or training whereas the most disruptive and complex forms of
automation will likely require high amounts of training, paired with job aids, and information.
This could be thought of on a sliding scale with the least complex forms of automation, such as
desktop or computer automation on the left, further increasing in complexity, such as Robotics
113
Process Automation (RPA), digital assistants, Natural Language Processing (NLP), through to
the most complex, such as Artificial Intelligence (AI), Machine Learning (ML), or cognitive
computing on the right side of the continuum. Education related to automation will be
specifically associated with learning an entirely new set of skills or obtaining a certification
through K–12 schooling, college, or a professional program. Assessing which method to use can
be determined using the standards discussed in each section.
Information. Knowledge and skill enhancement via information is appropriate only
when there is no specific need for individuals to practice a given action (Clark & Estes, 2008).
Information used on its own would only be appropriate in cases where the requirement is to
simply provide information that does not require an action of the recipient. This could be
information about industry changes, changes to the educational program itself, or changes to a
current automated process itself. However, because automation has a wide spectrum of types,
complexity, and applications, information could be used to support any stage of the knowledge
or skill enhancing process.
Providing information could be appropriate for any specific type of automation
educational interventions and will likely be used in conjunction with job aids or training. In fact,
it is likely that all other forms of enhancing knowledge or skills for automation will require the
use of information in some form. This is because any exposure to automation training or
education should be paired with a communications plan, which will include strategically planned
types and timing of information. The goal with information is to reduce uncertainty and may be
provided through thoughtful communication (Clark & Estes, 2008). For automation, educational
leaders could determine when information will fulfill the knowledge requirement as a part of
their communications’ plan based on timing of given curriculum. The communication plan
114
should also provide an overview of the actions required to achieve the desired end state, and to
reiterate prior experience or similar concepts for which people may be familiar (Clark & Estes,
2008).
Information would also be appropriate for enhancing knowledge around understanding
general industry trends around automation and what students and workers can expect in the
workplace. A standard refresher of automation-specific information should be included as a
regular part of educational programs. This will continually inform future workers of
technological changes and industry trends, therefor providing a continuous enhanced
understanding of how automation will impact them and how to prepare for it.
Types of automation to use information to enhance knowledge would be any low-level
forms of automation, such as familiar computer desktop or backend automation. These are forms
of automation that require minimal training or additional skill to operate. An example could be
an information sheet that provides an overview of how to complete a very basic task that can be
accomplished in a few steps, such as clicking to install or update a currently used and familiar
system. Some examples of the types of information that could be used include communications
or updates to generate awareness of coming changes and plans, the types of automation that are
being considered for use, how the new automation will (or will not) be compatible with current
systems or processes, how to prepare for changes, and to socialize the need for new skills,
training, or educational courses.
Based on the results of this study, it will also be important to provide information on the
potential impact that can be expected for an individual worker’s job, tasks, and potential career
changes. This should start by providing information to enhance the individual’s understanding of
the impact that automation may have on their future job and tasks. This should also include
115
information for the types of skills, training, or additional educational courses that will be needed
for coming automation. In doing this, future workers will have a better understanding of coming
changes and understand how educational systems or organizations are going to support them
through the process. Improvements in these areas can be made through evaluating the type of
automation that will be implemented and developing a communication and training plan to
ensure future workers understand the pending automation and how they will be personally
responsible for preparing to adopt and use it.
Job Aids. Given that automation will require practice for people to become familiar and
use it, job aids may be appropriate. Job aids typically provide self-help type information that can
provide guidance for how to perform low level tasks (Clark & Estes, 2008). Job aids could be
designed to include information that supplements training that has already been provided through
something like a simple checklist, reminder list, or summary sheet (Clark & Estes, 2008). One
key distinction of job aids is that job aids are only appropriate when tasks do not require guided
practice for complex tasks (Clark & Estes, 2008).
Job aids should be considered to use in conjunction with automation that requires
understanding for how to specifically use new, or updated (upgraded) systems, applications, or
other forms of automation. Job aids could be developed to provide step by step guides to using
automation, how to set up new systems or conduct upgrades, or how to conduct repairs. Job aids
are likely appropriate for all forms of automation, but will vary in the level of detail contained,
depending on the complexity of automation. Desktop (computer) automation, Robotics Process
Automation (RPA), digital assistants, language processing, human augmentation tools or
systems, and even Artificial Intelligence (AI), Machine Learning (ML), and other advanced
forms of automation are some examples of potential candidates for the use of job aids.
116
In most cases, job aids would be used to supplement information, training, or education
that was already provided to workers (Clark & Estes, 2008). As students complete training or
education, they could be provided with job aids which would provide helpful, easily accessed,
and straightforward guidance or tips to remind them of material that was covered in more depth
in prior training. Job aids can also serve in the place of additional training, when people are
already experts, and may only need to understand what has changed or a new approach to
accomplishing a familiar task (Clark & Estes, 2008). In the case of students, job aids may suffice
even in situations of more complex automation for those students who have reached the pinnacle
of their programs, and have a firm understanding and high level of subject matter expertise.
Under information processing system theory, information that is learned and connected
meaningfully with prior knowledge is more readily used as it is reinforced by prior knowledge
(Schraw & McCrudden, 2006). For automation specifically, job aids may serve as a critical piece
for learning as they could be used to help break down complex tasks into more manageable parts
and help workers organize the way they ingest specific pieces of information. In conjunction
with the training aspect described in the next section, job aids could serve to help promote the
proper framing for selecting, organizing, and integrating in the knowledge transfer process and
help to reduce information overload, otherwise referred to as cognitive load (Mayer, 2011).
Training. Training is appropriate when people must acquire the ability to do something
which requires modeling, practice, and feedback (Clark & Estes, 2008). Training is in essence
the combination of information, practice, and feedback (Clark & Estes, 2008). Training is not to
be conflated with personal development, although personal development is important.
Development relates to improving skills or knowledge for the personal growth of the individual,
while training relates to the systematic approach of enhancing an individual’s skills or
117
knowledge with the goal of improving the individual, team, and organization (Aguinis &
Kraiger, 2009).
Training serves as the basis for primary knowledge and skill enhancement for how to use
or repair automated technologies. This is because training is a combination of information and
guided practice, paired with feedback to help provide corrective actions (Clark & Estes, 2008).
Training for automation could take place within educational settings or a workplace, but,
according to Clark and Estes (2008), is not defined by a specific setting. Regardless, training
should be well planned, well communicated, and have clear objectives (Clark & Estes, 2008), in
this way, it will support the more rapid adoption of new automation and begin fostering new
generations of exposed workers.
Almost all forms of automation will require training, either for demonstrating and
practicing use, or for how to program or repair. Schraw and McCrudden (2006) advocated that
practice and application is critical for developing mastery in conjunction with acquiring
component skills and for individuals understanding how those skills can be integrated with
existing processes. Given that most automation today is focused on replacing routine tasks, it is
likely that much of the focus will be on providing training on how to use specific types of
automation. Training should only include topics covering what is necessary to understand how to
use or repair the automation, except in cases where the organization is responsible for the actual
creation of the automation itself, such as building computer code or algorithms, engineering
(computer or mechanical), or constructing the physical or digital automation. Depending on the
type of automation, there may need to be a period of initial training, as well as subsequent
periodic training. Determining additional training needs should be based situations that require
leaning new knowledge or skill, or to significantly enhance existing skill (Clark & Estes, 2008).
118
This study’s results indicated that a majority of participants believed they would have a
greater need for an increase in skill that could come through training rather than additional
education. Additionally, the results indicated that 86% of participants had an expectation that
their employer would be primarily responsible for providing training and that 56.3% of
participants reported that they were already expecting training for how to use new automation.
This could indicate an opportunity for educational leaders to partner with industry to allow for
students to engage in hands on training or internships at appropriate levels of educational
attainment, such as at the end of a program or graduation. This would serve two purposes, to (a)
to increase the participation rate of employers to provide training to their existing workforce; and
(b) to provide employment opportunities and familiarity with workplace settings and norms for
soon to be graduates.
Along with educational and organizational responsibility, future workers will bear some
of the responsibility for preparing themselves for automation. Because nonroutine cognitive
work will be highly valued in the future, workers themselves will need to be continually focused
on upskilling and enhancing their digital dexterity (Mok, 2018). According to Mok (2018),
digital dexterity is described as the cognitive ability and social awareness to use information and
technology in innovative ways (Mok, 2018). Building relationships with industry for student
internship or cross training programs will also allow for students to have earlier exposure to the
idea their long-term success will be partially predicated on continual upskilling across their
career. This underscores the importance of building the educational pipeline through hands on
internships or cross training programs for students.
Education. While education was the primary focus of the overall recommendation, it is
important to understand how it differs from information, job aids, or training. Education is the
119
form of knowledge and skills’ enhancement that is reserved for situations where people need to
be able to handle unanticipated challenges (Clark & Estes, 2008). Education should be thought of
in the context of comprehensive changes and situations where cause and effect may be required
to enable people to identify root causes and work to fix them (Clark & Estes, 2008). For the
context of recommendation one, education is an all-encompassing vehicle that serves to create a
comprehensive understanding of automation. Although education is typically thought of in the
context of being provided through in a school setting, for automation, it may also be performed
in technical certification or internship program that involve a great deal of hands-on training.
Educational programs should include specific knowledge or skill for how to solve
complex problems (Clark & Estes, 2008) or perform upgrades or enhancements. Technological
education, specifically around automation, is typically paired with higher complexity automation
applications which can include algorithmic-based automation, computer code-based automation,
or advanced robotics, to name a few. Machine Learning (ML), Artificial Intelligence (AI),
advanced engineering, language generation, large caches of data processing, or data analytics
related fields, will all require workers with subject matter expertise, or those who have specific
education around such concepts. This will mean that programs should be designed to incorporate
increasing levels of complexity as students achieve mastery of base level concepts. As students
progress through various courses or subjects, they should develop expertise which should allow
them to take on more complex challenges, requiring further education. Developing expertise will
be most appropriate for building knowledge and skill around designing, repairing, or integrating
automation, rather than simply using the automation.
120
Recommendation 2: Support Acceptance and Use Automation Through Enhancing
Perceived Value of Automation and Worker Efficacy
The correlational results of this study revealed four (out of 15) significant relationships
related to the importance of enhancing perceived value of automation and worker efficacy in
order to support their acceptance and use automation. Two of these correlations were directly
related to workers’ willingness to accept change, how useful workers perceived automation and
how easy it would be to use. The results indicate that workers who reported a high willingness to
embrace change also reported greater levels of perceiving automation to be useful and easy to
use. This means that workers who are more likely to embrace change in general may also be
more likely to see the positive aspects and appreciate automation for how it can help them in
their job.
The other two relationships that resulted from the correlations were related to anxiety and
how worker perception of how easy automation would be to use, and how much control they had
over using automation. Workers reported lower levels of anxiety when they felt automation was
both easier to use and that they had more control over using it. This is an important reflexive
relationship in the context of acceptance and use of automation, in that if workers believe they
have control over automation, believe it will be easy to use, and have low anxiety about it, they
may be more motivated to use it.
In conjunction with the results of the correlations, most of the study participants (93.8%)
reported that they would at least sometimes, almost always, or always make a suggestion to
improve things through automation. Additionally, (84.4%) indicated having prior exposure to
automation, which means they had a general understanding that automation is a part of the
workplace. Therefore, even workers who may not have indicated being exposed to prior
121
automation would still make a suggestion to put in new automation to support workplace
improvements. When taken together, these results show that the majority of the sample
population not only reported having a high degree of being exposed to prior automation, but also
had a high degree of positive behavioral intent to make improvements through automation given
the opportunity. Highlighting these results are important as behavioral intent is tied to motivation
and may provide evidence that familiarity with automation can support adoption.
While the previous recommendation described using education to gradually build and
improve skills in future workers, workers will still need some form of motivation to foster
acceptance and use of automation. Motivation is broken into three facets including (a) active
choice, that is an individual’s drive to work towards a goal; (b) persistence, continuing to
persevere through accomplishment; and (c) mental effort, the cognitive effort one devotes to
achieving a goal (Clark & Estes, 2008). This is a critically important concept for automation as it
highlights the importance of understanding potential motivational issues that could hinder
workers’ persistence through the automation adoption period. The following will provide
recommendations for supporting acceptance and use of automation through motivational
constructs including enhancing perceived value of automation and enhancing worker efficacy.
The concept of motivation is often seen as a critical component of the learning process as
it involves an individual’s decision to actively choose to start a task, persist toward a goal, and to
invest mental effort required for ultimate achievement (Clark & Estes, 2008). Motivation at its
core is a willingness to take action, but the likelihood to take action will vary from person to
person based on the nature of specific situations (Ryan & Moller, 2017). There are several kinds
of psychological constructs that are foundational to starting tasks, persisting at them, and
investing the necessary mental effort, but there are two types that are most widely research
122
supported as critical to motivated behaviors. The first type is centered around increasing
perceived value, which in terms of this study translates to acceptance. The second type
encompasses enhancing efficacy, which translates to use, for the purpose of this study. This
section provides a background on value and efficacy and how organizations can use theoretical
concepts to support automation implementation.
Enhance Worker Perceived Value to Support Acceptance of Automation
The concept of perceived value, or acceptance, is supported by Expectancy-Value Theory
(EEVT) which was developed in the 1980s by Eccles-Parsons et al. in 1983 (Wigfield et al.,
2017). Kurt Lewin originally posited that the goals and expectations of individuals contributed to
their task completion, which was further expanded by John Atkinson who developed the
achievement motivation theory which combined individual needs, expectations, and values
(Schunk, 2014). Essentially, EEVT posits that expectancies and values influence a person’s
choice to start and complete a task (Wigfield et al., 2017). Expectancies of individuals often
include perceptions based on historical beliefs, past goals, and memories and may be influenced
by how individuals believe they will be perceived by others (Wigfield et al., 2017).
There are three types of value that are described through EEVT including: attainment
value (or importance to the individual), intrinsic value (or personal enjoyment through
achievement), and utility value (or perceived usefulness to the individual) (Wigfield et al., 2017).
EEVT includes many components that are seen as subjective due to the fact that value is relative
to the perceived value of the individual (Wigfield et al., 2017). This is also the case with the
level of perceived (or actual) effort, or cost, to attain such value, which may include financial,
social, emotional, psychological, or other costs (Wigfield et al., 2017). While organizations
should include plans for how to influence individual belief to bolster both attainment and
123
intrinsic value for individuals around accepting automation, efforts should be primarily focused
on utility value, given that utility value can be extrinsically influenced. Focusing on utility value
helps support initial rational for taking action, in that there is a belief that completing an action
will provide useful value for the participant (Wigfield et al., 2017). Enhancing utility value is
also important in the context that individual achievement values contribute to expectancies and
influence potential outcomes (Wigfield et al., 2017).
As leaders begin to strategize for how to motivate workers to accept new automation,
they can develop comprehensive plans that remind workers of the expected personal costs and
benefits of adoption. Leaders can influence workers by discussing the importance of the effort
and remind them of personal utility value of completing the task (Pintrich, 2003). Efforts around
overall communication can include messaging that details the expected benefits of automation,
specifically around ease of use, and compatibility with what workers are already accustomed to.
Communication on the benefits must also be paired with training, and communication about the
fact that training will provide workers with an understanding of how to use the new automation
and customize it to be compatible with their working style. In following the recommendations
around motivation, leaders can expect results such as higher levels of interest as individuals
believe are an active part of the process and have some control over the outcome (Wigfield et al.,
2017).
Enhance Worker Efficacy to Support Use of Automation
Self-efficacy theory was developed by Bandura (1997) and helps explain how a person’s
belief in their ability to accomplish something influences their desire to persist through both
intrinsic and extrinsic rewards as discussed with Deci and Ryan’s (1985) self-determination
theory (SDT). For this study, self-efficacy is a critical component because training and feedback
124
will be critical components of learning and using new automation. Bandura (1997) expressed the
importance of this in that people who successfully learn new things do so by observing others,
through modeled experiences, through social encouragement, or by practicing tasks themselves.
Social Cognitive Theory (SCT) research has shown that modeling improves self-efficacy,
learning, and performance (Denler et al., 2006). Additionally, people who have high self-efficacy
tend to set higher goals and plan ways to strategically accomplish them (Zimmerman et al.,
2017). More importantly, in the context of accepting and using automation, Zimmerman et al.
(2017) point out that those with high self-efficacy are also prone to select activities or tasks that
they are also confident they can accomplish. In essence, self-efficacy further reenforces the
incentive to persist until achievement (Zimmerman et al., 2017). All of these concepts help
reiterate the need for knowledge to be tied to motivation to improve the likelihood of both
acceptance and use of automation.
Leaders should focus on building organizational commitment and confidence in
achieving common benefits and goals, as well as communicating the consequences of a lack of
action (Clark & Estes, 2008). Enhancing team confidence may be achieved through what
Bandura (2000) described as collective efficacy, where group confidence is improved based on
the collective improvements of the whole group, rather than just one individual. For more
challenging and complex tasks, it may be helpful to structure learning and task accomplishment
in a group setting to increase both collective and individual confidence.
As implementation begins to take place, organizations can consider sharing quick wins or
success stories of how teams are using new automation. Leaders can also communicate messages
about the collective benefits that are being achieved to improve collective efficacy. As leaders
evaluate the effectiveness of motivational strategies, they can look for under-confidence; lacking
125
confidence that they can reach the goal, or overconfidence; making mistakes and blaming others
with no regard for constructive feedback (Clark & Estes, 2008). This will be important in
determining whether to make strategy adjustments across the program lifecycle to and reinforce
desired outcomes.
Recommendation 3: Develop and Enhance Organizational Systems and Processes to
Support the Acceptance and Use of Automation
The correlational results of this study revealed two (out of 15) significant relationships
related to the importance of creating strategic organizational systems and process to support the
acceptance and use automation. The correlations were directly related to a worker’s perception of
how compatible automation would be with their daily routine and (a) to how willing they would
be to accept change; and (b) their perception of needing to learn new skills. The resulting
relationship between willingness to accept change and perceived need for skills changes,
highlights the importance of focusing on effective organizational planning. Because
organizations will have the power to select types of automation, and to develop implementation
plans, it also means they can focus on effective comprehensive planning.
As organizations look to put new automation in place, it will be critical for leaders to
understand what will be required to ensure workers will both accept and use the new automation.
This will require leaders to ensure they develop support channels for new automation by
preparing workers for planned organizational implementation through comprehensive planning.
While the previous recommendations focused on (a) enhancing knowledge and skills through
educational interventions; and (b) leveraging perceived value and worker efficacy; this
recommendation focuses on organizational improvements to support acceptance and use of
automation.
126
Leverage Understanding of Organizational Gaps to Identify Areas of Improvement
In order to properly frame the scope of effort required for implementing automation, it is
important to understand the current state of the workplace. In conjunction with the
recommendations in this chapter, this section advocates for the use of the Clark and Estes (2008)
Knowledge, Motivation, and Organizational (KMO) gap analysis framework as a starting point
ahead of any planning around automation. The Clark and Estes (2008) framework was developed
as a method for helping organizations determine how they could turn evidence-based results into
practical results. The authors posit that using evidence-based results of knowledge, motivation,
and organizational gaps allows leaders to select solutions that will lead to substantial
organizational performance improvements (Clark & Estes, 2008). The framework helps leaders
align solutions with overall organizational goals, identify potential barriers and gaps in
performance, and to select appropriate solutions to make data-driven decisions (Clark & Estes,
2008). The KMO framework is supported by substantial empirical evidence and has been used
for many years by a host of public and private organizations across multiple industries and
settings.
Using a framework such as the Clark and Estes’s (2008) Knowledge, Motivation, and
Organization (KMO) Gap Analysis can help leaders quickly hone in on exactly where specific
focus should be allocated to support successful automation implementation. The framework
effectively helps to determine whether there should be additional focus on knowledge, as in skills
or understanding of what is needed to successfully operate new automation, or motivation, as in
what will energize workers to adopt and use automation, or developing organizational systems
or removing barriers to facilitate successful implementation and use of automation. While the
Clark and Estes’s (2008) KMO model requires evaluating knowledge, motivation, and
127
organizational barriers, the gap analysis can help organizations determine specific efforts around
each area of focus.
Organizational performance gaps are often easier to identify once an organization has
conducted the knowledge and motivation gap analysis (Clark & Estes, 2008). This is because
modifications to knowledge or motivation levels in the organization typically help determine the
changes needed to organizational processes (Clark & Estes, 2008). The best way to understand
the most critical focus areas is by conducting organizational surveys, interviews, or focus groups
designed to remove potential bias and premature conclusions (Clark & Estes, 2008). After the
gap analysis is complete, and the organization can begin looking to see whether people have the
right tools in place, well designed processes, and procedures that all support common
organizational goals (Clark & Estes, 2008).
Document and Standardize Processes and Polices to Support Acceptance of Automation
With automation, organizational performance problems can be addressed by evaluating
current materials and tools needed to support automation integration, documenting and
standardizing processes, and ensuring that organizational polices are updated to reflect the
changes that are expected from new automation. In addition to Clark and Estes’s (2008)
recommendation to provide the right knowledge, skills, and motivational support for to support
the organizational change process, Dixon (1994) recommends the following steps including (a)
development of clear goals, vision, and performance measures; (b) alignment of processes and
structures to overall organizational goals; (c) persistent and candid communication about
progress and planned changes; and (d) involvement of those in management in the continuous
improvement process. Regardless of the chosen approach, organizations will need to make every
effort to ensure selected automation includes consideration for compatibility.
128
The first step of ensuring compatibility starts by documenting and standardizing
processes and polices to work in conjunction with new automation. This will require planning to
ensure automation is also paired with appropriate training to increase worker confidence of
future compatibility in the case of a brand-new process. Organizations will have to conduct
process reviews so they can be documented and then improved prior to automation
implementation. This will not only aid in ensuring compatibility, but can increase worker
confidence that the new automation will provide the benefits as discussed with expectancy value
theory.
Develop Appropriate Outcomes and Objectives to Support Organizational Goals
A large component of automation will be evaluating the changes that will affect
individual workers and their immediate environment. Learning and motivational theory posit that
environmental changes can affect behavior (Daly, 2010; Tuckman, 2006), reiterating the
importance of creating an environment for learning that contributes to strengthening automation
adoption and use. As leaders prepare learning and implementation plans, they should identify the
specific outcomes they are looking to accomplish that will lead to changes in behavior (Daly,
2010). This should include plans for helping to control the environmental influences, that could
affect the desired behavior (Tuckman, 2006), which may include both positive and negative
factors.
Planning should include a focus on helping both the organization and the individual
across the change process of adopting new automation through comprehensive learning,
motivation, and organizational plans. Learning should be broken down into both appropriate
channels (e.g., information, job, aids, training, or education) as well as manageable parts with a
focus on only what is important for communicating specific knowledge. Motivation strategies
129
should focus on internal (intrinsic) and external (extrinsic) factors and to provide both support
and incentives that reward desired behavior. As discussed in previous sections, modeled behavior
will be an important aspect in demonstrating desired behavior and should be included in the
organizational planning efforts as Denler et al. (2006) note the importance of ensuring modeled
behavior comes from credible sources and has functional value.
Automation implementation plans should also include, as Daly (2010) describes, clear
behavioral objectives that can then be measured and evaluated based on demonstrated individual
performance. It is also recommended that organizations encourage individuals to set challenging
but realistically achievable goals and to engage in self-evaluation (Denler et al., 2006). Self-
assessment can be paired with training to enable workers to evaluate their progress and bolster
confidence over time. Materials and activities should be well planned, useful for meeting
learning objectives, helpful in connecting learners to common interests, and based on task for
which they will actually be expected to perform (Pintrich, 2003).
Objectives should be supported by the use of immediate feedback (Tuckman, 2006) with
a stronger emphasis on providing reinforcement of positive behaviors and a lesser emphasis on
correction of unwanted actions (Daly, 2010). Leaders can develop structured feedback
mechanisms to ensure this is taking place, and should also consider the use of surveys or
observation checklists as recommended by Clark and Estes (2008). Additionally, organizations
can include content that provides strategies for time management, motivation, learning, and how
to help control physical and environmental factors to promote better monitoring and management
of individual performance (Dembo & Eaton, 2000).
130
Limitations and Delimitations
Limitations are the factors of a study that are typically outside a locus of control and tend
to be associated with the chosen method, models, constraints, or research design that can impact
the potential output or results of the study (Theofanidis & Fountouki, 2018). This study had four
primary limitations including (a) the inability to control the total number of people who chose to
participate; (b) the self-reported nature of information provided by the participants; (c) self-
selection bias; and (d) participation bias. Specifically, for self-selection and participation bias,
those who participated or did not participate may have potentially altered the type of data
gathered. Additionally, the convenience sampling for the study may have resulted in higher
education levels of participants as observed in the results. This could be attributed to a greater
likelihood that the pool of candidates would have similar educational attainment to me. For this
study, the survey required a general understanding of how to use a computer or mobile device,
access to the internet, and to be a member of LinkedIn or Facebook social media platforms.
Although surveys were broadcast to as many individuals as possible to increase the participation
rate, there was no control over whether recipients choose to complete the survey or not. As far as
the nature of self-reported information, surveys were designed for participants to answer all of
the survey questions, but there was no control over the way in which each participant answered a
given question or their level of truthfulness.
Delimitations are limits or boundaries that are established to ensure feasibility to answer
research questions and meet objectives (Theofanidis & Fountouki, 2018). Delimitations are also
the specified items or concepts that are not covered in a given study in order to adhere to the
chosen conceptual framework and methods (Galvan & Galvan, 2017). For this study, the survey
items were delimited to those selected and adapted from the Unified Theory of Acceptance and
131
Use of Technology and Acceptance of Change (AC). Additionally, Likert-type items were
chosen and ordered as a part of the survey design in order to answer the research questions. As
far as data, only data from participants who fell within the scope of the research design were
used for the data analysis, and surveys from out-of-scope participants were excluded. The survey
participation window was delimited to the cut-off date and participants were not able to take the
survey more than once. To reduce the potential for survey fatigue and encourage completion, the
survey was pilot tested with peers and revised to reduce the average completion time to 10
minutes.
Recommendations for Future Research
While there were specific recommendations that were presented as a part of this study,
additional research will need to be conducted. As discussed in the literature of this study,
technology itself is continually evolving at an increasingly rapid pace, which does not lend itself
to assuaging the fears that organizations and individuals have about automation and the changes
that will be required. The challenges that face current organizations for how to influence workers
to accept and use automation will only continue to increase, requiring continual research,
practice, and application. In order to overcome these challenges, future research will need to
include both academic (traditional and theoretical) as well as industry-based (practical and
application) research. It is also recommended that future research consider incorporating
adjustments to study design, based on observed limitations and results from the design and
execution of this study.
Academic, Industry, and Societal Based Needs for Future Research
Although there has been a general emphasis on research around knowledge and skill
improvements, motivational constructs, and organizational improvements on the academic side,
132
there is room to conduct additional research specifically related to automation. Conceptually, all
three constructs could employ their own academic studies to produce meaningful results. Studies
could include types of information, job aids, training, or education that is most effective for use
in educational settings and for building knowledge and skills around specific types or categories
of automation. With motivation, studies could explore the effectiveness of specific motivational
techniques, theories, or models when paired in various organizational settings and with specific
automation implementation and use. Organizational research could include analyses and testing
of new comprehensive deployment and implementation plans, including potential adaptation and
use of change management theory and application.
On the practical side, research has shown that planning can help organizations realize the
full potential and benefits of automation (Manyika et al., 2017a). While broader macro-level
strategies and considerations for workforce planning were discussed in chapter two, concepts
around what constitutes effective planning for automation could benefit from additional research.
Tasks and jobs that are replaceable will continue to shift which will require organizations to keep
track of both current and future skills needed for the automation of today and tomorrow
(Manyika et al., 2017a). There is also a need for continued consideration about the second and
third order effects of technology itself, such as user addiction, overreliance, and the social and
societal ramifications that result from its use (Brynjolfsson & McAfee, 2014). As workers’ jobs
or tasks are augmented or replaced, there will need to be a means for determining where workers
can focus their time to continue to create value in new ways (Eggers et al., 2017). Organizations
will continue to struggle with balancing automation and human skill needs which will require
new strategies on how to maintain performance and continued innovation (Makridakis, 2017).
133
More broadly across society, research will need to be conducted to understand larger
macro-level implications of automation. This may include rethinking social support systems
across job and transition training, education, and employment assistance (Mok, 2018; Wright &
Schultz, 2018). Policy discussions and research into new ways to provide financial assistance
through things like unconditional basic income (Colombino, 2019), or other tax funded programs
should continue. Additionally, research should take into account as Vermeulen et al. (2018)
discusses, the types of activities that will replace and contribute to fulfillment as a result of
workers having more personal discretionary time and money outside their job.
Future Research Considerations Based on the Design and Results of This Study
As a part of future efforts, researchers should seek a broader pool of participants to
enable additional analysis such as multivariate (MANOVAs) to enable evaluation of multiple
associations, which was not possible in this study due to the limited sample. Having a broader
pool of participants will aid in developing a larger participant profile and the possibility of
developing worker personas to develop targeted strategies to support transition, acceptance, and
use of automation. A larger sample and longer participation period would also enable the
opportunity to expand the demographic profile of participants to enable additional analysis for
targeted segments (populations) of the larger sample.
Although there is value in a larger pool of participants for generalizable research, future
research could also include targeting of specific industries where there is a greater likelihood for
automation. Along these same lines, research could be targeted at a specific organization or
micro segment within a specific industry to allow for additional understanding of a specific
subsect of workers. For example, a study could be conducted with a focus on the call center
industry more broadly (macro), or it could be focused specifically on call center workers (micro).
134
The rationale for either macro or micro would be justified given that there are heavy investments
being made into automated answering services, which are causing disruption in the industry as
well as with those who work in a call center. In either case, the generalizability of the data
collected could be used to help organizations understand commonalities among specific personas
of workers to then tailor automation planning efforts to more specific needs of those worker
groups.
There is also opportunity to make adjustments to the valid and reliable scales used for
this study. Namely, several of the scales used resulted in higher-than-expected Cronbach’s alpha,
with some above .90. While having an internal consistency above .70 is important, Salkind and
Frey (2020) note that producing consistent reliability requires work and may require continued
testing and re-testing to remove unnecessary items. These results indicate the possibility of
conducting additional pilot studies to further pair down the number of items in each of the scales
with the target of between .70 and .80. Making these adjustments will reduce the number of total
items in the overall survey design, reducing survey fatigue, or allowing for additional scales from
other constructs to be added in their place.
Implications for Equity: Focus on Ethics and Balancing Profit, Efficiency, and Social Good
to Ensure the Most Vulnerable Are Considered as a Part of the Automation Strategy
While the economist Milton Friedman (1970) discussed corporate responsibility as
maximizing profit through whatever means necessary, as long as laws are followed, Anderson
(2015) argued that the broader role of ethics should be on achieving a balance of profit,
efficiency, and social good. Ethical decisions around automation are not considerably different
than other ethical business decisions in that leaders’ function as ethical agents who act beyond
legal relationships with their stakeholders, which can result in mutual benefits for well-planned
135
decisions (Anderson, 2015). As with other organizational decisions, leaders will be faced with
moral, ethical, and social implications, but decisions around automation may require the
introduction of external regulations to ensure workers’ fundamental rights are considered and to
help govern ethical issues, particularly for those most vulnerable (Vermeulen et al., 2018; Wright
& Schultz, 2018).
One way of helping to evaluate automation decisions may be through use of stakeholder
and social contracts theory to consider whether short-term financial gains are worth the longer-
term macro effects (i.e., impacts to broader society as a result of individual organizational
decisions and actions) (Wright & Schultz, 2018). Because organizations are already engaged in
social contracts with their stakeholders, they could then leverage automation decisions for
maximizing benefits while improving stakeholder relationships (Wright & Schultz, 2018). To
help evaluate the impact of automation decisions across stakeholders, Gartner (2018)
recommends conducting scenario-based planning, which includes thinking through possible
outcomes and impact that may come from decisions or actions taken (Poitevin, 2018). Scenario-
based planning can also be used to help create workforce development strategies to enhance
readiness, promote interoperability, and to design experiences for employees and customers alike
(Poitevin, 2018). Evaluating the potential impact of automation of decisions across all
stakeholders will not only allow leaders to evaluate their decisions but will help them to assess
mutual benefits that will work for the good of all stakeholders involved (Wright & Schultz,
2018).
While many employers may find balancing their desire for profit, greater productivity,
and the concern for the well-being of workers daunting, society as a whole may suffer without
proper consideration (Lent, 2018; Makridakis, 2017). These decisions will require consideration
136
of what Lent (2018, p. 208) calls “techno-caution,” a term used for considering potential positive
and negative effects of automation decisions. As leaders make difficult decisions, they must also
guard against the temptation of becoming callous and abandoning ethical considerations in the
midst of widespread adoption of automation (Cords & Prettner, 2018). When organizations take
the time to consider all stakeholders in the decision-making process, there can be a realization of
mutual gains (Anderson, 2015). This balance can be achieved through trade-offs between
progress and performance, and that focusing on social good may actually have the unintended
effect of increasing profit (Rangan, 2015).
Conclusion
Although workers and organizations alike may face similar fears about the rapid
advancement of technology and automation in the coming years, in light of historical research,
the coming Fourth Industrial Revolution is likely to play out much like the First Industrial
Revolution. While there is still a lack of clarity about the future, this should be an exciting time
for people and organizations as they work through the challenges and benefits that automation
will bring. What is clear is that organizations will need to take an active role in evaluating and
selecting automation that will not only keep them operational, but will allow them to thrive, and
to support their workers in new and exciting ways. In all of this, organizations should not forget
the importance of the human side of automation, and the intentional effort and investment
building human capital will require. As research has shown, regardless of the level investment
organizations make in automation or technology, in the end, people will still be needed.
As for individuals, ultimately, those who are working for their own best interest are
rewarded through alignment of incentives in a capitalistic system, and small differences, such as
upskilling and a focus on individual development, can lead to huge differentials in a market
137
society (Wheelan, 2019). These differences will continue to separate those who succeed, from
those who fail, insomuch as the technological age continues to produce a larger differentiation of
winners and losers (Wheelan, 2019). As Wheelan (2019) described, an organization or person
only needs to be just a little bit different than another to have massive success. In the long run, it
will be the workers who have the right skills and are familiar with technology who will fair far
better than workers who possess skills that are easily replaced by technology, robots, or
computers (Brynjolfsson & McAfee, 2014).
138
References
Acemoglu, D., & Restrepo, P. (2017). Robots and jobs: Evidence from US labor markets. NBER
working papers 23285. National Bureau of Economic Research, Inc.
https://www.nber.org/papers/w23285.pdf
Acemoglu, D., & Restrepo, P. (2018). Artificial intelligence, automation, and work. NBER
working papers 24196. National Bureau of Economic Research, Inc.
https://www.nber.org/papers/w24196
Aguinis, H., & Kraiger, K. (2009). Benefits of training and development for individuals and
teams, organizations, and society. Annual Review of Psychology, 60, 451–474.
Alexander, P. A., Schallert, D. L., & Reynolds, R. E. (2009). What is learning anyway? A
topographical perspective considered. Educational Psychologist, 44, 176–192.
https://doi.org/10.1080/00461520903029006
Amabile, T. M. (1996). Creativity in context. Westview Press.
Anderson, E. (2015). The business enterprise as an ethical agent. In Rangan, S. (Ed.),
Performance and progress: Essays on capitalism, business, and society (pp.186–201).
Oxford University Press.
Arntz, M., Gregory, T., & Zierahn, U. (2017, October). Revisiting the risk of automation.
Economics Letters, 159, Article 157, 157–160.
http://dx.doi.org/10.1016/j.econlet.2017.07.001
Autor, D. (2015) Why are there still so many jobs? The history and future of workplace
automation. Journal of Economic Perspectives, 29(3), 3–30.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Prentice-Hall.
139
Bandura, A. (1997). Self-efficacy: The exercise of control. Freeman.
Bandura, A. (2000). Exercise of human agency through collective efficacy. Current Directions in
Psychological Science, 9, 75–78. https://doi.org/10.1111/1467-8721.00064
Bessen, J. (2016). Learning by doing: The real connection between innovation, wages, and
wealth. Yale University Press.
Borry, E., & Getha-Taylor, H. (2018). Automation in the public sector: Efficiency at the expense
of equity? Public Integrity, 21(1), 6–21. https://doi.org/10.1080/10999922.2018.1455488
Boxall, P., Hutchison, A., & Wassenaar, B. (2014). How do high-involvement work processes
influence employee outcomes? An examination of the mediating roles of skill utilisation
and intrinsic motivation. The International Journal of Human Resource Management,
26(13), 1–16. https://doi.org/10.1080/09585192.2014.962070
Brookings. (n.d.). About us. Retrieved August 16, 2020 from
https://www.brookings.edu/about-us/
Brown, P., Lloyd, C., & Souto-Otero, M. (2018, November). The prospects for skills and
employment in an age of digital disruptions: A cautionary note (SKOPE Research Paper
No. 127). Center on Skills, Knowledge and Organisational Performance, Cardiff
University. https://files.eric.ed.gov/fulltext/ED593405.pdf
Brynjolfsson, E., & McAfee, A. (2011). Race against the machine: How the digital revolution
is accelerating innovation, driving productivity, and irreversibly transforming
employment and the economy. Digital Frontier Press.
Brynjolfsson, E., & McAfee, A. (2014). The second machine age: work, progress, and
prosperity in a time of brilliant technologies (1st
ed.). W.W. Norton & Company.
140
Cambridge University Press. (n.d.) Reskilling. In Cambridge dictionary. Retrieved February 23,
2020, from https://dictionary.cambridge.org/dictionary/english/reskilling
Chao, G., Hurst, E., & Shockley, R. (2018). The evolution of process automation: Moving
beyond basic robotics to inteligente interactions. IBM Institute for Business Value.
https://www.ibm.com/downloads/cas/QAQMRGVN
Chui, M., George, K., & Miremadi, M. (2017). A CEO action plan for workplace automation.
McKinsey Quarterly, (3), 88–94.
http://search.ebscohost.com.libproxy2.usc.edu/login.aspx?direct=true&db=bth&AN=125
907612&authtype=sso&custid=s8983984
Clark, R. E., & Estes, F. (2008). Turning research into results: A guide to selecting the right
performance solutions. Information Age Publishing, Inc.
Cohn, J. (2013, February 20). The robot will see you now. The Atlantic.
Colombino, U. (2019). Is unconditional basic income a viable alternative to other social welfare
measures? IZA World of Labor 128(2). https://doi.org/10.15185/izawol.128v2.
https://wol.iza.org/articles/is-unconditional-basic-income-viable-alternative-to-other-
social-welfare-measures/long
Collins. (n.d.). Technology. In Collins dictionary.com. Retrieved February 23, 2020, from
https://www.collinsdictionary.com/dictionary/english/technology
Cosman, E. (2016). Automation workforce development: Additional skills
required. InTech, 63(4), 45. http://search.proquest.com/docview/1841076862/
Cords, D., & Prettner, K. (2018). Technological unemployment revisited: Automation
in a search and matching framework. IDEAS Working Paper Series from RePEc.
http://search.proquest.com/docview/2119114221/
141
Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed
methods approaches. Sage Publications.
Daly, E. (2010). Behaviorism. Retrieved 2018, from
http://www.education.com/reference/article/behaviorism/
Davenport, T. H. (2019, Spring). The business value of digital workflows. Workflow Quarterly.
https://workflow.servicenow.com/quarterly/wp-
content/uploads/sites/2/2019/04/Workflow_Quarterly_The_Business_Value_Issue.pdf
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human
behavior. Plenum Press.
Dembo, M., & Eaton, M. J. (2000). Self-regulation of academic learning in middle-level schools.
The Elementary School Journal, 100(5), 473–490.
Denler, H., Wolters, C., & Benzon, M. (2006). Social cognitive theory. Retrieved 2018, from
http://www.education.com/reference/article/social-cognitive-theory/
Di Fabio, A., & Gori, A. (2016). Developing a new instrument for assessing acceptance of
change. Frontiers in Psychology (7), Article 802.
https://doi.org/10.3389/fpsyg.2016.00802
Dixon, R. J. (1994). Business process reengineering: Improving in new strategic directions.
California Management Review. 36, 93–108.
Eccles-Parsons, J. S., Adler, T. F., Futterman, R., Goff, S. B., Kaczala, C. M., Meece, J. L., et al.
(1983). Expectancies, values, and academic behaviors. In J. T. Spence (Ed.), Achievement
and achievement motivation (pp. 105–121). Guilford Press.
142
Eggers, W., Schatsky, D., & Viechnicki, P. (2017). AI-augmented government: Using cognitive
technologies to redesign public sector work. Deloitte University Press.
https://www2.deloitte.com/content/dam/insights/us/articles/3832_AI-augmented-
government/DUP_AI-augmented-government.pdf
Ernst and Young. (2017). Intelligent automation: Reshaping the future of work with robots.
https://www.ey.com/Publication/vwLUAssets/EY_intelligent_automation/$FILE/EY-
intelligent-automation.pdf
EY Global. (2018, September 14). Intelligent automation: how to reshape the future of work.
Retrieved March 30, 2020 from https://www.ey.com/en_lu/tmt/intelligent-automation--
how-to-reshape-the-future-of-work
Fleming, P. (2019). Robots and organization studies: Why robots might not want to
steal your job. Organizational Studies 40(1), 23–37.
https://doi.org/10.1177/0170840618765568
Frey, C., & Osborne, M. (2017). The future of employment: How susceptible are jobs to
computerisation? Technological Forecasting & Social Change, 114(C), 254–280.
https://doi.org/10.1016/j.techfore.2016.08.019
Friedman, M. (1970, September 13). The social responsibility of business is to increase its
profits. New York Times Magazine.
Galvan, J.L., & Galvan, M.C. (2017). Writing literature reviews: A guide for students of the
social and behavioral sciences. (7th ed.). Routledge.
Grant, C., & Osanloo, A. (2015). Understanding, selecting, and integrating at theoretical
framework in dissertation research: Developing a blueprint for your “house”.
Administrative Issues Journal 4(2), 12–26. https://doi.org/10.5929/2014.4.2.9
143
Hawksworth, J., Berriman, R., Cameron, E., & Goel, S. (2019). Will robots really steal our jobs?
An international analysis of the potential long-term impact of automation.
https://www.pwc.com/hu/hu/kiadvanyok/assets/pdf/impact_of_automation_on_jobs.pdf
Hemous, D., & Olsen, M. (2014). The rise of machines: Automation, horizontal
innovation and income inequality. IDEAS Working Paper Series from RePEc.
https://ideas.repec.org/p/cpr/ceprdp/10244.html
Ipsos. (2019). Ipsos – world economic forum: global citizens & automation.
https://www.ipsos.com/sites/default/files/ct/news/documents/2019-02/ipsos-wef_-
_global_consumer_views_on_automation_-_02.20.19.pdf
Keynes, J. (1978). Economic possibilities for our grandchildren. In E. Johnson & D. Moggridge
(Eds.), The collected writings of John Maynard Keynes (pp. 321–332). Royal Economic
Society. (Original work published 1930).
https://doi.org/10.1017/UPO9781139524162.031
Kim, H-J., Hawley, J. D., Cho, D., Hyun Y., & Kim, J-H. (2015). The influence of learning
activity on low-skilled workers’ skill improvement in the South Korean manufacturing
industry, Human Resource Development International, 19(3), 209–228.
https://doi.org/10.1080/13678868.2015.1122895
Kim, Y., Kim, K., & Lee, S. (2017). The rise of technological unemployment and its
implications on the future macroeconomic landscape. Futures, 87, 1–9.
https://doi.org/10.1016/j.futures.2017.01.003
Lankisch, C., Prettner, K., & Prskawetz, A. (2017, June). Robots and the skill premium: An
automation-based explanation of wage inequality. ECON WPS.
https://www.econstor.eu/bitstream/10419/168027/1/896045625.pdf
144
Lee, Y., Lee, J., & Hwang, Y. (2015). Relating motivation to information and communication
technology acceptance: Self-determination theory perspective. Computers in Human
Behavior, 51, 418–428. https://doi.org/10.1016/j.chb.2015.05.021
Lent, R. (2018). Future of work in the digital world: Preparing for instability and opportunity.
Career Development Quarterly, 66(3), 205–219. https://doi.org/10.1002/cdq.12143
Lindborg, H. (2017). Preparing for the revolution. Quality Progress, 50(8), 14–15.
http://search.proquest.com/docview/1938776419/
Lordan, G., & Neumark, D. (2018). People versus machines: The impact of minimum wages on
automatable jobs. Labour Economics, 52(C), 40–53.
https://doi.org/10.1016/j.labeco.2018.03.006
Lund, S., Manyika, J., Hilton Segel, L., Dua, A., Hancock, B., Rutherford, S., & Macon, B.
(2019). The future of work in America: People and places, today and tomorrow.
McKinsey Global Institute. https://www.mckinsey.com/featured-insights/future-of-
work/the-future-of-work-in-america-people-and-places-today-and-tomorrow
Makridakis, S. (2017). The forthcoming Artificial Intelligence (AI) revolution: Its impact on
society and firms. Futures, 90(C), 46–60. https://doi.org/10.1016/j.futures.2017.03.006
Manyika, J., Lund, S., Chui, M., Bughin, J., Woetzel, J., Batra, P., Ko, R., & Sanghvi, S. (2017a,
November). Jobs lost, jobs gained: What the future of work will mean for jobs, skills, and
wages. McKinsey Global Institute.
https://www.mckinsey.com/~/media/mckinsey/featured%20insights/Future%20of%20Or
ganizations/What%20the%20future%20of%20work%20will%20mean%20for%20jobs%
20skills%20and%20wages/MGI-Jobs-Lost-Jobs-Gained-Report-December-6-2017.ashx
145
Manyika, J., Chui, M., Miremadi, M. Bughin, J, George, K., Willmott, P, & Dewhurst, M.
(2017b, January). A future that works: Automation, employment and productivity.
McKinsey Global Institute.
https://www.mckinsey.com/~/media/mckinsey/featured%20insights/Digital%20Disruptio
n/Harnessing%20automation%20for%20a%20future%20that%20works/MGI-A-future-
that-works-Executive-summary.ashx
Mayer, R. E. (2011). Applying the science of learning. Pearson Education.
McClelland, D. C. (1985). Human motivation. Scott Foresman.
McClelland, D. C., Atkinson, J. W., Clark, R. A., & Lowell, E. L. (1976). The achievement
motive. Irvington.
Merriam, S. B., & Simpson, E. L. (2000). A guide to research for educators and trainers of
adults (2nd ed.). Krieger.
Merriam-Webster. (n.d.). Automation. In Merriam-Webster.com dictionary. Retrieved February
23, 2020, from https://www.merriam-webster.com/dictionary/automation
Merriam-Webster. (n.d.). Artificial intelligence. In Merriam-Webster.com dictionary. Retrieved
February 23, 2020, from https://www.merriam-
webster.com/dictionary/artificial%20intelligence
Merriam-Webster. (n.d.). Upskill. In Merriam-Webster.com dictionary. Retrieved
February 23, 2020, https://www.merriam-webster.com/dictionary/upskill
Mitchell, J., Gagne, M., Beaudry, A. & Dyer, L. (2012). The role of perceived organizational
support, distributive justice and motivation in reactions to new information technology.
Computers in Human Behavior, 28(2), 729–738.
146
Mok, L. (2018). How we will work in 2028 demands changes in how we will reward
(ID: G00350647). Gartner database.
Moore, G. (1998). Cramming more components onto integrated circuits. Proceedings of the
IEEE, 86(1), 82–85. https://doi.org/10.1109/JPROC.1998.658762
Muro, M., Maxim, R., & Whiton, J. (2019, January 24). Automation and artificial intelligence:
How machines are affecting people and places. Brookings.
https://www.brookings.edu/research/automation-and-artificial-intelligence-how-
machines-affect-people-and-places/
New International Version Bible. (2011). The NIV Bible. https://www.thenivbible.com (Original
work published 1978)
Nitin, M., Lowes, P., Ronanki, R., Wen, J., & Sharma, S. K. (2017, February) Machine
intelligence: Tech trends 2017. Deloitte University Press.
https://www2.deloitte.com/us/en/insights/focus/tech-trends/2017/moving-beyond-
artificial-intelligence.html..html
Patel, P. (2018). Compelling benefits, common misconceptions: Putting intelligent automation to
work for federal. Accenture. https://www.accenture.com/_acnmedia/pdf-90/accenture-
putting-automation-to-work-for-federal.pdf
Pintrich, P. R. (2003). A motivational science perspective on the role of student motivation in
learning and teaching contexts. Journal of Educational Psychology, 95, 667–686.
https://doi.org/10.1037/0022-0663.95.4.667
Pintrich, P. R., & Schunk, D. (1996). Motivation in education: Theory, research and
applications. Merrill.
147
Poitevin, H. (2018). The future of work and talent: Culture, diversity, technology
(ID: G00354454). Gartner database.
Prettner, K. (2017). A note on the implications of automation for economic growth and the labor
share. Macroeconomic Dynamics, 1–8. https://doi.org/10.1017/S1365100517000098
Rangan, S. (2015). Balancing and trade-offs: Discussion summary. In Rangan, S. (Ed.),
Performance and progress: Essays on capitalism, business, and society (pp. 280–283).
Oxford University Press.
http://www.oxfordscholarship.com.libproxy1.usc.edu/view/10.1093/acprof:oso/97801987
44283.001.0001/acprof-9780198744283-chapter-20.
Rezvani, A., Khosravi, P., & Dong, L. (2017). Motivating users toward continued usage of
information systems: Self-determination theory perspective. Computers in Human
Behavior, 76, 263–275. https://doi.org/10.1016/j.chb.2017.07.032
Robinson, S.B., & Leonard, K.F. (2019). Designing quality survey questions. Sage Publications.
Rocco, T. S., & Plakhotnik, M. S. (2009, March). Literature reviews, conceptual frameworks,
and theoretical frameworks: Terms, functions, and distinctions. Human Resource
Development Review 8(1), 120–130. https://doi.org/10.1177/1534484309332617
Ross, M., & Bateman, N. (2019, November 21). Low-wage work is more pervasive than you
think, and there aren’t enough “good jobs” to go around. Brookings.
Ryan, R. M., & Moller, A. C. (2017). Competence as central, but not sufficient, for high-quality
motivation: A self-determination theory perspective. In A. J. Elliot, C. S. Dweck, & D. S.
Yeager (Eds.), Handbook of competence and motivation (2nd ed., pp. 313–333). The
Guilford Press.
148
Salkind, N. J., & Frey, B. B. (2020). Statistics for people who (think they) hate statistics. Sage
Publications.
Schatsky, D., & Mahidhar, V. (2014, January 22). Intelligent automation: A new era of
innovation. Deloitte. https://www2.deloitte.com/us/en/insights/focus/signals-for-
strategists/intelligent-automation-a-new-era-of-innovation.html
Schraw, G., & McCrudden, M. (2006). Information processing theory. Retrieved 2018, from
http:// www.education.com/reference/article/information-processing-theory/.
Schunk, D. H. (2014). Motivation in education: Theory, research, and applications. Pearson
Education, Inc.
Spreitzer, G., Porath, C., & Gibson, C. (2012). Toward human sustainability: How to enable
more thriving at work. Organizational Dynamics, 41(2), 155–162.
https://doi.org/10.1016/j.orgdyn.2012.01.009
Sternberg, R. J. (1996). What should we ask about intelligence? American Scholar, 65(2), 205–
217.
Sternberg, R. J. (2017). Intelligence and competence in theory and practice. In A. J. Elliot, C. S.
Dweck, & D. S. Yeager (Eds.), Handbook of competence and motivation (2nd ed., pp. 9–
24). The Guilford Press.
Swaine, M. R., & Freiberger, P. A. (2020). Electronic Numerical Integrator and Computer.
Encyclopedia Britannica. Retrieved April 7, 2020 from
https://www.britannica.com/technology/ENIAC
Tay, G., & Aggarwal, A. (2018, March 29). Learn how the future of work is transforming
employee experience globally (ID: G00349068). Gartner database.
149
Theofanidis, D., & Fountouki, A. (2018). Limitations and delimitations in the research process.
Perioperative Nursing, 7(3), 156–164.
https://www.spnj.gr/articlefiles/volume7_issue3/pn_sep_73_155_162b.pdf
Tuckman, B. (2006). Operant conditioning. Education.com Retrieved 2018,
http://www.education.com/reference/article/operant-conditioning/
U.S. Census Bureau. (2017). 2017 North American Industry Classification System (NAICS).
U.S. Census Bureau. Retrieved September 27, 2020, from
https://www.census.gov/eos/www/naics/2017NAICS/2017_NAICS_Manual.pdf
Varian, H., R. (2014). Big data: New tricks for econometrics. Journal of Economic Perspectives,
28(2), 3–28.
Venkatesh, V., Morris, M., Davis, G., & Davis, F. (2003). User acceptance of information
technology: Toward a unified view. MIS Quarterly, 27(3), 425–478.
https://doi.org/10.2307/30036540
Vermeulen, B., Kesselhut, J., Pyka, A., & Saviotti, P. P. (2018). The impact of
automation on employment: Just the usual structural change? Sustainability, 10(5),
Article 1661. https://doi.org/10.3390/su10051661
Webb, M. (2020, January). The impact of artificial intelligence on the labor market. Stanford
University. https://web.stanford.educ/~mww/webb_jmp.pdf
Wheelan, C. (2019). Naked economics: Undressing the dismal science. W.W. Norton & Company.
Wigfield, A., Rosenzweig, E. Q., & Eccles, J. S. (2017). Achievement values: Interactions,
interventions, and future directions. In A. J. Elliot, C. S. Dweck, & D. S. Yeager (Eds.),
Handbook of competence and motivation (2nd ed., pp. 116–134). The Guilford Press.
150
Wisskirchen, G., Thibault Biacabe, B., Bormann, U., Muntz, A., Niehaus, G., Jimenez Soler, G.,
& von Brauchitsch, B. (2017). Artificial intelligence and robotics and their impact on the
workforce. IBA Global Employment Institute.
Wright, S.A, & Schultz, A.E. (2018). The rising tide of artificial intelligence and business
automation: Developing an ethical framework. Business Horizons. 61(6), 823–832.
Zimmerman, B. J., Schunk, D. H., & DiBenedetto M. K. (2017). The role of self-efficacy and
related beliefs in self-regulation of learning and performance. In A. J. Elliot, C. S.
Dweck, & D. S. Yeager (Eds.), Handbook of competence and motivation (2nd ed., pp.
313–333). The Guilford Press.
151
Appendix A: Survey Protocol and Items
Survey Introduction:
Thank you for taking the time to take this survey.
The survey is open to anyone residing within the United States who is 18 or older.
The results of this survey will contribute to an academic study on the future of work which is
designed to help employers understand and consider what workers may need when new
automation is introduced.
The survey should take about 10–15 minutes to complete and has four (4) main sections that ask
questions about:
• Your understanding of automation
• How you view change in general
• How you see automation in the workplace
• And some general demographic questions
All of your responses are completely anonymous and there are no questions that ask for
identifiable information. You will also have the opportunity to select that you "do not know" or
prefer not to answer for any of the questions.
Thank you for taking the time to complete it, please click below to get started.
Section I – Up Front Qualifying and Demographic Questions
Q1. Are you 18 years of age or older?
1. Yes
2. No
Q2. Do you currently live and reside within the United States?
1. Yes
2. No
Q3. When thinking about my gender, I identify as:
1. Female
2. Male
3. I prefer not to answer
4. I prefer to self-describe
152
Q4. My age falls into the following:
1. 18–23
2. 24–29
3. 30–34
4. 35–39
5. 40–44
6. 45–49
7. 50–55
8. 55–64
9. 65+
10. I prefer not to answer
Q5. What is the highest level of school you have completed or the highest degree you
have received?
1. Less than a high school degree
2. High school graduate, diploma, or GED
3. Some college but no degree
4. Associate degree in college (Please state Major or Concentration below)
5. Bachelor's degree in college (Please state Major or Concentration below)
6. Master's degree (Please state Major or Concentration below)
7. Doctoral degree (Please state Major or Concentration below)
8. I prefer not to answer
Q6. Which of the following best describes your length of work experience? (Interval)
1. No formal work experience
2. Less than 1 year
3. More than 1 year, but less than 3 years
4. 3–5 years
5. 6–10 years
6. 11–15 years
7. 16–20 years
8. 20+ years
9. I prefer not to answer
Q7. Where are you currently employed or working?
1. PUBLICLY TRADED company
2. PRIVATE-NOT-FOR-PROFIT, tax-exempt, or charitable organization
3. PRIVATE company or business
4. State or Local GOVERNMENT employee
5. Federal GOVERNMENT employee (Non-Military)
6. Active duty U.S. Military
7. SELF-EMPLOYED in own business, professional practice, or farm
8. Working WITHOUT PAY in family business or farm
9. Does not apply, I am not currently working
10. I Prefer not to answer
153
Q8. Are you a U.S. Military Veteran?
1. Yes
2. No
3. I prefer not to answer
Q9. What is your current occupation?
1. Please describe your occupation below:
2. Does not apply
3. I prefer not to answer
Q10. Which of the following most closely matches your level in your organization?
1. Intern
2. Entry Level (non-manager)
3. Mid-Level (non-manager)
4. Manager
5. Senior Manager
6. Director
7. Vice President
8. Senior Vice President or Partner
9. C level executive (CIO, CTO, COO, CMO, Etc)
10. President or CEO
11. Owner
12. I Prefer not to answer
13. Other – Please describe below:
14. Does not apply
15. I prefer not to answer
Q11. What is the current size of your whole organization or company in terms of the
number of people who work there?
1. Please write in the approximate number of people below:
2. Does not apply
3. I do not know
4. I prefer not to answer
Q12. Which of the following most closely matches your current industry? (specified if
unclear in title)
1. Accommodation and food services (providing customers with lodging and/or
preparing meals, snacks, and beverages for immediate consumption such as
hotels/motels, RV parks, food services/drinking places/restaurant)
2. Administrative and support and waste management services (performing
routine support activities for the day-to-day operations of other organizations such
as office administration, facilities support, employment
services/placement/executive search, document prep services, call
centers/answering/telemarketing, private mail centers, copy centers, collection
agencies/repossession, court reporting/stenotype, travel arrangement,
154
investigation/security guard/armored car, pest control, janitorial, landscaping,
carpet cleaning/upholstery, waste management)
3. Agriculture, forestry, fishing and hunting
4. Arts, entertainment, and recreation (operating or providing services to meet
varied cultural, entertainment, and recreational interests of their patrons such as
performing arts/sports, museums/historical sites, amusement/gambling/recreation
industries)
5. Construction
6. Educational services (providing instruction and training in a wide variety of
subjects such as elementary/secondary schools, colleges/universities, training
industries/computer/technical/trade schools, cosmetology/barber schools, fine
arts, exam preparation)
7. Finance and insurance
8. Health care and social assistance (providing health care and social assistance for
individual such as health practitioners or social workers, ambulance services,
hospitals, nursing/residential care facilities, dental/chiropractic/mental health/
other health and therapy offices, medical labs)
9. Information (transmitting or communicating information such as publishing,
motion picture and sound recording, broadcasting, telecommunications, data
processing/hosting, other information services)
10. Management of companies and enterprises (such as holding of securities of
companies and enterprises, for the purpose of owning controlling interest or
influencing their management decisions, or administering, overseeing, and
managing other establishments of the same company or enterprise and normally
undertaking the strategic or organizational planning and decision-making role of
the company or enterprise)
11. Manufacturing
12. Mining quarrying, and oil and gas extraction
13. Other public services (except public administration) (such as repair and
maintenance, personal and laundry services, personal care, death care, religious
activities, grantmaking, civic, professional organizations, labor unions)
14. Professional, scientific, and technical services (performing professional,
scientific, and technical services for the operations of other organizations such as
legal services, accounting/tax prep/bookkeeping/payroll services,
architectural/engineering services/inspection/surveying, specialized
design/interior/industrial/graphic, computer system design,
management/scientific/technical consulting, scientific research and development,
advertising/public relations, veterinary)
15. Public administration (administration, management, and oversight of public
programs by Federal, State, and local government such as federal/state/local
government agency workers, including police, fire, military occupations)
16. Real estate and rental and leasing
17. Retail trade (merchandise and goods sold to the general public)
18. Transportation and warehousing (air, rail, water, truck, transit and ground
passenger, postal service, couriers and messengers, warehousing activities)
19. Utilities (such as electric, gas, steam, water, and sewage removal)
155
20. Wholesale trade
21. Other – Please specify below:
22. I do not know
23. I prefer not to answer
Section II – General Perception of the Impact of Automation
Section II Introduction
Okay, great we made it through some of the demographic info, lets jump right into the first
section.
This section asks some questions about:
• Your perception and understanding of automation
• Types of jobs you think may be affected
• Skills and education that you think might be needed to remain relevant in the workforce
The term “automation” will be used throughout the survey and refers to any method, system,
application, or device that replaces a task done by a human or makes completing a task easier by
requiring less effort from humans.
Some examples of automation could be as simple as a calculator or microwave, or as complex as
an artificial intelligence that can perform complicated tasks. Some other examples include:
• a computer or smartphone application that converts your spoken words to text on a
computer
• a robotic tool that helps you lift a heavy object
• a chatbot or automated phone answering virtual assistant that helps customers
• self-driving automobile technology that drives a car for you
• software that can scan digital documents to provide specific information you were
looking for
• something that monitors for security issues and alerts of breeches
Let's get started.
Personal Exposure to Prior Automation
Q13. Considering all of the jobs I have had in the past and present:
1. I can think of a time when new automation was implemented in my workplace:
2. I am not aware of new automation being put in place at any workplaces for which
I have worked
3. Prefer not to answer
Behavioral Intent
Q14. If I identified an opportunity to improve things in my workplace through new
automation, I would:
156
1. Always make a suggestion
2. Almost always make a suggestion
3. Sometimes make a suggestion
4. Almost never make a suggestion
5. Never make a suggestion
6. Do not know
7. Prefer not to answer
Understanding Impact on Jobs and Job Tasks (Researcher Created Items)
Q15. Please indicate your level of agreement with the following statements (Matrix style
grouping):
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. In general, I understand the types of jobs that may be affected by automation.
2. I understand how my job may be affected by automation.
3. In general, I understand the duties or tasks within a job that may be replaced by
automation.
4. I understand the duties or tasks within my job that that may be replaced by
automation.
Perception of Career Changes as a Result of Automation (Researcher Created Items)
Q16. Please indicate your level of agreement with the following statements (Matrix style
grouping):
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. Automation will cause me to have to change roles within my workplace.
2. Automation will cause me to have to change jobs.
3. I expect to make career changes due to automation.
Perception of Needed Skills, Education, Training (Researcher Created Items)
4. I will have to acquire new skills to perform my job as a result of automation.
5. I will have to acquire additional education (certification or college) to perform my
job as a result of automation.
157
Perception of Personal or External Responsibility for Acquiring Education and Training
Q17. In the past, when new automation has been implemented in my current or a past
workplace:
1. My employer has provided required training for how to use new automation that
was put in place
2. My employer has not provided any training for how to use new automation that
was put in place
3. I do not know of any training provided by my employer related to how to use new
automation
4. I do not know
5. I prefer not to answer
Q18. When new automation has been implemented in my current or a past workplace,
and my employer has not provided training, I have taken the initiative to seek
additional training, certifications, or education:
1. Always
2. Almost always
3. Sometimes
4. Almost never
5. Never
6. Do not know
7. Prefer not to answer
Q19. Please move the slider toward who you believe should be mostly responsible for
training, certification, or education as a result of automation. A 0 means you are
100% responsible, a 100 means your employer is 100% responsible.
I am primarily
responsible
My employer is primarily
responsible
100% my responsibility 100% my employer
1. Do not know / Prefer not to answer
Q20. Regarding any new automation the is being or will be put in place in my current
workplace:
1. I am expecting to receive training provided by my employer for how to use new
automation
2. I am not expecting to receive any training from my employer for the new
automation
3. No, my current workplace is not implementing any new automation
4. I am not aware of any new automation being implemented
5. Do not know
6. I prefer not to answer
158
Perceived Percentage of and Self-Described Types of Job Tasks Deemed Replaceable
Q21. In your current workplace, on a 0% to 100% scale, what percentage of your daily
job-related tasks or duties do you believe could be performed with an automated
technology? (Move the slider)
0% of tasks or duties 100% of tasks or duties
1. Do not know / Prefer not to answer
Q22. In your current workplace, on a 0% to 100% scale, what percentage of your daily
job-related tasks or duties have already been replaced by automation? (Move the
slider)
0% of tasks or duties 100% of tasks or duties
1. Do not know or Prefer not to answer
Q23. Which of your daily tasks or duties of your work do you believe may be replaced by
automation in the future?
1. Please describe in a few words or statements the tasks or duties you think could
be replaced
2. I do not foresee any tasks or duties that could be replaced
3. I do not know
4. I prefer not to answer
Section III – Acceptance of Change (AC) Items
Section III Introduction
Great job, you are over halfway done.
This section and the next section have most of the survey questions, but shouldn’t take too much
time.
This section will ask questions about how you deal with change in your life in general.
Let's get started.
159
Q24. When thinking about change in general: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
Predisposition for Change
1. When I am faced with a change, I can see things from multiple perspectives.
2. I am able to take most of the opportunities that occur to me.
Support for Change
3. I trust the people close to me when faced with change.
4. When I compare myself with others, I am better able to cope with change.
Change Seeking
5. I am looking for changes in my life, even when things are going well.
6. I normally seek different ways to do the same things in my daily routine.
Q25. When thinking about change in general: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
Positive Reaction to Change
1. I am able to tolerate even the negative aspects of change.
2. I can find the positives in changes that are apparently negative.
Cognitive Flexibility
3. If necessary, it is not difficult for me to change my mind.
4. When I’ve made an important decision, I can change if it involves an
advantage.
Section IV – Unified Theory of Acceptance and Use of Technology (UTAUT) Items
Section IV Introduction
Okay, there are just two more sections.
This section asks about your understanding and comfort with automation.
As a reminder, the term “automation” refers to any method, system, application, or device that
replaces a task done by a human or makes completing a task easier by requiring less effort from
humans.
Let's get started.
160
Facilitating Conditions (FC)
Compatibility (C) – The degree to which an innovation is perceived as being consistent with
existing values, needs, and experiences of potential adopters
Q26. When thinking about automation in the workplace: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. Using new automation would need to be compatible with most aspects of my
work.
2. New automation would need to fit well with the way I like to work.
3. Using new automation would need to fit into my work style.
Performance Expectancy (PE)
Perceived Usefulness (U) – the degree to which a person believes that using a particular system
would enhance his or her job performance
Q27. When thinking about automation in the workplace: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. Using new automation in my job would enable me to accomplish tasks more
quickly.
2. Using new automation would improve my job performance.
3. Using new automation in my job would increase my productivity.
4. Using new automation would enhance my effectiveness on the job.
5. Using new automation would make it easier to do my job.
Effort Expectancy (EE)
Perceived Ease of Use (EOU) – the degree to which a person believes that using a system would
be free of effort
Q28. When thinking about automation in the workplace: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. Learning to operate new automation is generally easy for me.
161
2. I generally find it easy to get new automation to do what I want it to do.
3. It is generally easy for me to become skillful at using new automation.
4. In general, I find new automation easy to use.
Facilitating Conditions (FC)
Perceived Behavioral Control (PBC) – Reflects perceptions of internal and external constraints
on behavior and encompasses self-efficacy, resource facilitating conditions, and technology
facilitating conditions
Q29. When thinking about automation in the workplace: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. I feel I would have control over using the new automation.
2. I feel I would have the resources necessary to use new automation.
3. I feel I would have the knowledge necessary to use new automation.
4. Given the resources, opportunities and knowledge it takes to use new
automation, it would be easy for me to use it.
Attitude Toward Using Automation (ATUA)
Anxiety (ANX)
Q30. When thinking about automation in the workplace: (Matrix style grouping)
1. Strongly agree, 2. Agree, 3. Disagree, 4. Strongly disagree, 5. Do not know, 6.
Prefer not to answer
1. I feel apprehensive about new automation.
2. It scares me to think that I could lose a lot of information using new
automation by making a mistake while using it.
3. I hesitate to use new automation for fear of making mistakes I cannot correct.
4. New automation is somewhat intimidating for me.
Section V Remaining Demographic Survey Questions
Section V Introduction:
Okay, the heavy lifting is done, and this is the final section.
This final section only asks demographic questions that will be important for helping to identify
needs for workers in similar situations.
162
Q31. Are you of Hispanic, Latino, or of Spanish origin?
1. Yes
2. No
3. Do not know
4. Prefer not to answer
Q32. Choose one or more races that you consider yourself to be (Check all that apply)
1. American Indian or Alaska Native
2. Asian
3. Black or African American
4. Native Hawaiian or Other Pacific Islander
5. White
6. Biracial or Multiracial
7. Do not know
8. I prefer not to answer
Q33. When describing my education level, I would most identify with the following
(check all that apply)
1. First in my family to graduate high school
2. First in my family to attend college
3. First in my family to graduate college
4. First in my family to obtain a graduate degree
5. Not the first in my family for any of the options listed
6. I prefer not to answer
Q34. When thinking about technical knowledge, I would most identify with the following
(check all that apply)
1. I have completed a formal technical schooling or training program (i.e.,
technical degree, code school, Information Technology Information Library
(ITIL), Comp TIA Suite, Information Systems Security or Network
Certifications, etc.) – Please describe below:
2. I do not have any technical schooling or certifications.
3. I prefer not to answer
163
Q35. When thinking about where I was born and raised, I would describe myself as:
1. Born and raised in the United States
2. Born in the United States, but raised in another country, having later moved
back to the United States
3. Born and raised outside of the United States, but later moved to the United
States
4. Born outside the Unites States, but raised in the United States
5. I prefer not to answer
Q36. I would consider myself fluent in the English language
1. Yes, and it was my first language
2. Yes, but it was not my first language
3. I am not fluent in English
4. I prefer not to answer
Q37. Which statement best describes your current employment status?
1. Working (paid employee)
2. Working (self-employed)
3. Not working (temporary layoff from a job)
4. Not working (looking for work)
5. Not working (retired)
6. Not working (disabled)
7. Not working (other)—Please describe below
8. I prefer not to answer
Q38. Which of the following best describes your current employment situation?
1. Working full time—40 or more hours per week
2. Working part time—20–39 hours per week
3. Working less than part time—less than 20 hours per week
4. Unemployed or not working
5. I prefer not to answer
Q39. Which of the following best describes how many jobs you currently work?
1. I work only one job
2. I work two jobs
3. I work three or more jobs
4. Does not apply, I am not currently employed or working
5. Does not apply, I am a business owner or self-employed
6. I prefer not to answer
164
Q40. Which of the following best describes the length of time you have spent in your
current occupation?
1. Less than 2 years in the same occupation
2. 2 – 5 years in the same occupation
3. 6 – 10 years in the same occupation
4. 11 – 15 years in the same occupation
5. 16 – 20 years in the same occupation
6. 20+ years in the same occupation
7. I prefer not to answer
Q41. Prior to your most recent occupation, which of the following best describes the
length of time you spent in your most recent previous occupation?
1. I have not worked in another occupation
2. Less than 2 years in the same occupation
3. 2 – 5 years in the same occupation
4. 6 – 10 years in the same occupation
5. 11 – 15 years in the same occupation
6. 16 – 20 years in the same occupation
7. 20+ years in the same occupation
8. I prefer not to answer
Q42. Please indicate the choice below that reflects your personal annual income before
taxes:
1. Less than $10,000
2. $10,000 to $19,999
3. $20,000 to $29,999
4. $30,000 to $39,999
5. $40,000 to $49,999
6. $50,000 to $59,999
7. $60,000 to $69,999
8. $70,000 to $79,999
9. $80,000 to $89,999
10. $90,000 to $99,999
11. $100,000 to $149,999
12. $150,000 or more
13. I prefer not to answer
Q43. What is your zip code?
14. Please enter your zip code below
15. I prefer not to answer
165
End of Survey Message
Thank you for taking the time to complete this survey.
Your responses are very valuable in the conversation for workers may need as companies,
businesses, and organizations implement new automation technologies in the workplace.
Please consider texting or sharing the link you used with a few others who could take the survey.
Thanks again for your help.
166
Appendix B: Survey Protocol Crosswalk
Research question Survey items (by section or type)
RQ1: Are there differences in an individual
worker’s perception and understanding of
automation based on their demographic
background?
Section I—General perception of the impact of
automation
Section IV—Demographic survey questions
RQ2: Are there differences in an individual
worker’s level of acceptance of change
and acceptance and use of automation
based on their demographic background?
Section II—Acceptance of change (AC) scale
Section III—Acceptance and use of automation
(AUA) survey items
Section IV—Demographic survey questions
RQ3: What are the perceptions of individual
workers related to automation?
Section I—General perception of the impact of
automation survey items
Section III—Acceptance and use of automation
(AUA) survey items
RQ4: Which tasks and what percentage of
one’s job does an individual worker
perceive can be automated?
Section I—General perception of the impact of
automation survey items
RQ5: Is there a relationship between an
individual worker’s perception and
understanding of automation, level of
acceptance of change and level of
acceptance and use of automation?
Section I—General perception of the impact of
automation scales
Section II—Acceptance of change (AC) scale
Section III—Acceptance and use of automation
(AUA) survey items
167
Appendix C: 2017 U.S. North American Industrial Classification System Descriptions
Sector Title and descriptions
11 Agriculture, forestry, fishing and hunting (Growing, raising animals, harvesting
timber, fish or other animals from farms, ranges or other natural habitats)
21 Mining quarrying, and oil and gas extraction (Such as coal, ore, liquid
minerals/crude petroleum, gasses, or other mining activity)
22 Utilities (electric, gas, steam, water and sewage removal)
23 Construction (buildings, structures, heavy construction other than buildings,
alterations, reconstruction, installation, maintenance, repairs)
31–33 Manufacturing (all types including mechanical, physical, chemical, or chemical
transformation of materials, substances, or components into new products)
42 Wholesale trade (merchant wholesalers/durable goods, nondurable goods, wholesale
electronic markets/agents/brokers)
44–45 Retail trade (motor vehicle and parts dealers, furniture, electronics and appliances,
building material and garden, food and beverage stores, gas stations,
clothing/accessories, sporting goods/hobby/music instruments/bookstores, general
merchandise, misc. retail, electronic shopping/mail-order houses/vending machine
operators/fuel dealers/other direct sellers)
48–49 Transportation and warehousing (air, rail, water, truck, transit and ground passenger,
pipeline, scenic/sightseeing, support activities, postal service, couriers and
messengers, warehousing and storage)
51 Information (publishing, motion picture and sound recording, broadcasting,
telecommunications, data processing/hosting, other information services)
52 Finance and insurance (monetary authorities-central bank, credit intermediation,
securities/commodity contracts/ other financial investment, insurance carriers,
funds/trusts/other financial vehicles)
53 Real estate and rental and leasing (renting, leasing, or use of tangible or intangible
assets except copyrighted works)
54 Professional, scientific, and technical services (legal services, accounting/tax
prep/bookkeeping/payroll services, architectural/engineering
services/inspection/surveying, specialized design/interior/industrial/graphic,
168
Sector Title and descriptions
computer system design, management/scientific/technical consulting, scientific
research and development, advertising/public relations, veterinary)
55 Management of companies and enterprises (holding of securities of companies and
enterprises, for the purpose of owning controlling interest or influencing their
management decisions, or administering, overseeing, and managing other
establishments of the same company or enterprise and normally undertaking the
strategic or organizational planning and decision-making role of the company or
enterprise)
56 Administrative and support and waste management services (office administration,
facilities support, employment services/placement/executive search, document
prep services, call centers/answering/telemarketing, private mail centers, copy
centers, collection agencies/repossession, court reporting/stenotype, travel
arrangement, investigation/security guard/armored car, pest control, janitorial,
landscaping, carpet cleaning/upholstery, waste management)
61 Educational services (elementary/secondary schools, colleges/universities, training
industries/computer/technical/trade schools, cosmetology/barber schools, fine arts,
exam prep,
62 Health care and social assistance (ambulance services, hospitals, nursing/residential
care facilities, social assistance, dental/chiropractic/mental health/ other health and
therapy offices, medical labs)
71 Arts, entertainment, and recreation (performing arts/sports, museums/historical sites,
amusement/gambling/recreation industries)
72 Accommodation and food services (hotels/motels, RV parks, food services/drinking
places/restaurant)
81 Other public services, except public administration (repair and maintenance,
personal and laundry services, religions/grantmaking/civic/professional
organizations, labor unions)
92 Public administration (federal/state/local government agency workers, including
police, fire, military occupations)
Abstract (if available)
Abstract
Automation of human tasks has begun to fuel a massive workforce transition labeled the “Fourth Industrial Revolution”, because it will fundamentally change the dynamic of how people work, and the types of jobs they will do (Vermeulen et al., 2018). The purpose of the study was to explore individual worker demographics and perspectives about the potential impact of automation and how those perspectives relate to the individual’s willingness to accept change, and accept and use automation in order to assess potential opportunities for better planned organizational change. The study employed a survey designed to produce results of workers’ level of understanding of the impact of automation, their willingness to accept change, and accept and use automation. The study resulted in multiple correlational relationships indicating the importance of workers having an understanding of automation, how it will impact them in the future, and how organizations can better prepare workers for changes that come with automation. Based on this study, the following is recommended, to (a) use educational interventions to support knowledge and skill development to improve future acceptance and use of automation; (b) support acceptance and use automation through enhancing perceived value of automation and worker efficacy; and (c) develop and enhance organizational systems and processes to support the acceptance and use of automation.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Emotional intelligence self-perceptions in non-clinical leaders: an examination into a healthcare organization
PDF
Your soft skills are showing: organizational efforts to develop soft skills
PDF
An improvement study of leading a sustainable electric utility future through organizational change effectiveness
PDF
Industry 4.0 impacts on U.S. food industry executive self-directed learning
PDF
Improving the representation of female executives in a large utility provider: a modified KMO framework
PDF
Women in executive leadership: a study of the gender diversity gap
PDF
Perceptions of U.S. Army first-line supervisors on first-term soldier attrition
PDF
An examination of the impact of diversity initiatives and their supporting roles on organizational culture: an experiential study from the perspective of diversity personnel
PDF
The lack of gender diversity in executive leadership ranks within higher education
PDF
Labor versus management perspectives on remote, in-person, and hybrid work: an examination of incongruences, challenges, and strategies for optimal workplace collaboration, motivation, and performance
PDF
COVID-19's impact on Christian cross-cultural workers
PDF
Evaluating technical support provided in online education for students with disabilities
PDF
Stress and burnout: a qualitative study of the influences on female leaders' decision to leave
PDF
Influencing and motivating employee engagement: an exploratory study on employee engagement in organizational injury-prevention programs
PDF
Evaluating the relationship between communication, trust, and retention at Tombstone Police Department
PDF
Barriers to practice: primary care patient experience and quality health outcomes in the Veterans Health Administration
PDF
Influences on the use of restorative practices in a United Kingdom junior school: an evaluation study
PDF
The impact of advanced technologies on the workplace and the workforce: an evaluation study
PDF
The Army’s process to evaluate costs versus benefits: a case study on the change of command ceremonies
PDF
Hail to the chief: an exploration of female chief executives’ successes
Asset Metadata
Creator
Ahlgren, Daniel L.
(author)
Core Title
Preparing for the future of work: exploring worker perceptions of the impact of automation
School
Rossier School of Education
Degree
Doctor of Education
Degree Program
Organizational Change and Leadership (On Line)
Degree Conferral Date
2021-08
Publication Date
07/31/2021
Defense Date
07/15/2021
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
automation,digital dexterity,future of work,human augmentation,intelligent automation,OAI-PMH Harvest,re-skilling,Technology,up-skilling
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Seli, Helena (
committee chair
), Lever, Scott (
committee member
), Phillips, Jennifer (
committee member
)
Creator Email
dahlgren@usc.edu,mrahlgren@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC15670636
Unique identifier
UC15670636
Legacy Identifier
etd-AhlgrenDan-9952
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Ahlgren, Daniel L.
Type
texts
Source
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
automation
digital dexterity
future of work
human augmentation
intelligent automation
re-skilling
up-skilling