Jobs in a Generative AI Future: An Examination of the Impact of Generative AI on the Top White Collar Jobs and Society

Photo sourced from Midjourney with prompt, “Photorealistic image of two hands — one human, one robotic — touching each other. The perspective is zoomed out to see the arms of both hands up to the elbows. The background is a futuristic office scene.”

This paper was written for the Global Institute for the Advancement of Emerging Technology and Innovation. A PDF version of the paper is available via a link at the bottom of this page.

Executive Summary

Numerous articles, blog posts and other writeups sound the alarm of how generative AI is going to change work and jobs. Writers hypothesize that jobs will be eliminated and work, as we know it, is destined to change. It is difficult to argue with their conclusions given what we have seen of this new technology but how did they get there and, perhaps more importantly, what is the ultimate impact of the changes wrought by advancements in artificial intelligence?

In this paper, we set out to examine the likelihood of individual jobs being replaced by generative artificial intelligence (AI). Generally, automation affects blue collar jobs most rapidly. Generative AI will impact white collar jobs, on the other hand, in ways we have not seen previously. As a result, we sought to examine the top ten white collar jobs by income.

To uncover these jobs, we visited numerous websites that listed “top jobs” and then selected the most common ten jobs. We identified representative job descriptions and assessed the responsibilities associated with each role based on a set of five criteria. We used those criteria to assign a score to each job which allowed us to rank the jobs by how likely they are to be eliminated as a result of generative AI.

Our work suggests that the jobs we reviewed will not be eliminated in the near term but, as we might expect, they will be radically altered. We believe the changes to come will affect more than just jobs and will transform multiple components of our job pipeline — primary education, secondary education, college and, finally, work. This approach leads us to go beyond simply analyzing how specific jobs will be impacted to outlining changes we believe are forthcoming for governments, employers, colleges and universities, and K-12 education as a result of generative AI’s transformational influence. We conclude the paper with recommendations to be considered in an effort to get ahead of the forthcoming transformation.

Introduction

Much ado has been made over the impact of AI-driven automation on the global labor market. Just as robotics revolutionized manufacturing by automating the labor of human workers in factories, artificial intelligence, and, in particular, generative AI will transform white collar work. In fact, as referenced by the Wall Street Journal, research from the University of Pennsylvania and OpenAI suggests that almost all work will change as a result of generative AI.

Naturally, many questions arise from knowing this magnitude of change is on the horizon. Among the entities asking the most difficult questions are those involved in the labor pipeline: corporations responsible for hiring qualified employees, colleges and universities that provide education and training for future workers, secondary schools which create the foundational knowledge most workers build upon and, of course, governments tasked with understanding the needs of both its citizens and the commercial enterprises that hire those citizens.

Each of those groups has questions related to serving the needs of their specific constituencies. When distilled to their most basic insights, the questions become the same:

• What jobs will be eliminated and which will be radically changed?
• What new jobs will exist?

In this paper, we seek to provide insight on the first question with a focus on white collar jobs. Americans in the top 1–25% (earning between about $86,000 and and $548,000 annually) of income strata generate over approximately one-third of the individual taxes received by the IRS. That income generates almost $800 billion in tax revenue and is a significant portion of what America collects from its citizens.

The jobs people have to generate that income and the associated tax revenue are many. Yet, we know there are some lucrative careers that increase the likelihood of placing people in that heady group of top 1–25% of earners. Those jobs — physician, attorney, accountant, engineering, etc — are our focus in this paper.

Approach

It is our intention to more firmly understand how generative AI will change the jobs most likely to provide our economy with the high incomes upon which it depends. Our question going into this project was, how will AI impact the careers in the top ten high income career paths. With that question in mind, we began this project by asking Google for the “top white collar jobs”.

The results of the query led to multiple sources with varied jobs. We chose to select the roles that seemed to be the most consistent among the sources. Those roles are:

• Accountant
• Architect
• Attorney
• Civil Engineer
• Dentist
• Financial Manager
• Management Consultant
• Marketing Manager
• Physician
• Software Developer

These jobs represent a range of roles — all of which have many specialties. For this report, we focus on the generic representation of the job. When we reference a specialty (e.g., civil engineer), it is our belief that the role, as described, is representative of the non-specialty variant of the profession.

Once we identified the roles, we turned to LinkedIn Jobs to find job descriptions that would describe these roles. We used the following criteria to identify job descriptions:

• Descriptions must be detailed and have enough information about the job’s expectations to be representative of the role at almost any other company
• Cannot have been created by a very small (<10 employees) business and, in fact, our preference was 50+
• Responsibilities are clearly delineated by bullets and/or can be easily delineated by our team with bullets

Given those criteria, we were able to identify appropriate job descriptions associated with each of the specified roles. The job descriptions we selected are based in a diversity of states and metropolitan areas and we consider them to be a reflection of a broad set of requirements.

Before reviewing the specific requirements of each job description, we established a set of criteria we would use to measure the likelihood of a specific job being replaced by artificial intelligence. To identify these criteria, we, ourselves, turned to artificial intelligence. We asked ChatGPT which factors contribute to jobs being automated and it provided us with a comprehensive list. Of those, we determined the following factors were best suited for our use case:

• Capability of Technology Replacement — How capable will generative AI be at performing this task within the next five years?
• Repeatability of a Process — How unique is each execution of the task? Task repeatability often is a hallmark of a task that can be easily automated
• Amount of Human-to-Human Contact Required — How much human-to-human contact is required to successfully complete this task? Activities requiring human interactions may be augmented by AI but they’re not likely to be replaced by AI in the near future
• Speed and Efficiency of Human Actor — How fast and efficient is the human worker at performing this task? Activities at which the human worker performs slowly (especially as a result of human limitations rather than being an unproductive worker) may be optimal for augmenting with AI
• Likelihood of Regulatory Intervention — How likely is some regulatory or rule-enforcing body to influence how this work is performed? Regulatory intervention can both support and impede the use of AI. In this paper, our position is greater regulatory likelihood incentivizes the usage of AI as a way to mitigate risks associated with human error

For each bulleted job requirement in each job requisition, we assessed each of the criteria on a five point scale. The scales for each are below:

• Capability of Technology Replacement

1 — Low capability, not likely to perform within the next 5 years

5 — High capability, very likely to perform within the next 5 years

• Repeatability of a Process

1 — Not repeatable, each execution of this task is sufficiently different as to be unique

5 — Very repeatable, each execution of this task is the same or similar to the one before it

• Amount of Human-to-Human Engagement Required

1 — Human-to-human engagement is very necessary and appreciated for this task

5 — No human-to-human engagement, human engagement is not necessary or appreciated with this task

• Speed and Efficiency of Human Actor

1 — Fast and efficient, the human worker is able to execute this task quickly and efficiently

5 — Low Speed and Efficiency, the human worker is slow and inefficient at performing this task

• Likelihood of Regulatory Intervention

1 — Low likelihood of regulatory intervention, this industry is not regulated, the task is not impacted by regulation and is not likely to be regulated

5 — High likelihood of regulatory intervention, the industry is regulated, the task is regulated and greater scrutiny is likely as a result of automation

These assessments are subjective. It is, admittedly, challenging to gauge each role in each job description with objectivity using these scales. This is especially true with regard to Capability of Technology Replacement where we make educated guesses about what’s possible within five years based on an extrapolation of what’s possible today and forecasts of what will be possible.

All conditions notwithstanding, it is these assessments that drive our approach. We use these ratings to provide each job with a score that represents how likely it is to be eliminated by generative artificial intelligence. This is achieved by averaging the rating for each bulleted job requirement in each category. Once we obtained the criteria average for each job, we summed the score from all criteria to create the score.

The score rankings are as follows:

• 5–9 — Low likelihood of job being eliminated as a result of artificial intelligence driven automation though specific requirements may be susceptible to change
• 10–14 — Some likelihood of being eliminated as a result of artificial intelligence driven automation. More likely is AI taking on some tasks that can be automated with the human worker assuming responsibility for more hands-on and interpersonal work
• 15–19 — Moderate likelihood of the job being eliminated as a result of artificial intelligence driven automation. At the least, significant requirements of this role will be automated perhaps changing the responsibilities of the role even if not eliminated
• 20–25 — High likelihood of the job disappearing as a result of artificial intelligence driven automation. This is a job for which most of its responsibilities can — and perhaps should — be automated

Going through this process yielded the results demonstrated in Table 1.

Table 1 — Scores of Top Ten White Collar Jobs (listed alphabetically)

The scatterplot below provides a simpler representation of the scores and an understanding of how each role ranks relative to the others.

Graph 1

The results of our analysis are a bit surprising to us. We expected the Accountant role to be very susceptible to the influence of generative AI. The role is one in which, while the numbers may change, the processes and systems used to work are repeatable, require little human interaction, can likely be done more quickly by computers and are subject to regulation through both Generally Accepted Accounting Principles (GAAP) rules and IRS requirements. The bigger surprises were how highly attorneys and dentists ranked.

In fact, dentists and attorneys are similar in multiple ways. Both professions are highly educated, licensed by governmental institutions, and have rules imposed upon them by both government and non-governmental organizations. Additionally, both roles, when excluding specialities, perform repetitive tasks that may differ somewhat on a case-by-case basis but are largely similar enough as to not be considered unique.

Key Trends

Trend: All Jobs at Risk

It becomes clear from a review of Graph 1 that none of the jobs we assessed are highly likely to be replaced by generative AI in the near future based on our analysis. All of the jobs, however, fall into the 10–15 and 15–20 quartiles. This suggests a heightened likelihood of each of the jobs we assessed being significantly affected by generative AI. In fact, based on our analysis of the job requirements for each role, it is our position that each role has individual requirements that lend themselves to being automated — even if the job itself is not eliminated — and those changes will fundamentally alter the way the work is conducted for those jobs.

This point has been made by many other articles and papers. It is worth repeating, however, because education and income will not protect people in these roles. These roles are as susceptible to the coming changes as any other role in our society. Some of them will be even more so affected because generative AI has the potential to make it economically impractical to pay highly compensated human workers to perform tasks the AI can execute at much lower costs.

Trend: “Unique” Tasks May be Repetitive

One of the elements that stands out in our assessments is how activities that might be perceived as one-offs and unique can be repetitive — and, thus — automatable. The accounting professional, for example, will face significant disruption from generative AI. One reason for this is, while many individuals and companies will have varying revenue/income and expenses, the analysis of those exchanges is consistent from one entity to another. Once the rules are established, the accountant turns to calculations — at which computers excel.

The same is true for dentistry where each human has a different mouth but , generally, the analysis of the mouth is the same (e.g., more or less similar number of teeth that are supposed to demonstrate common properties). AI will excel at identifying issues with teeth through imaging earlier than the human dentist might be able to make a diagnosis. Additionally, if a robot can be used to assemble parts in a factory with precision and be used in hospital operating rooms, it can also be used to perform common dentistry tasks.

Both dentists and accountants are well-trained and highly skilled workers. Their training and experience are often what make them successful at their professions. In both cases, as we will find with many roles, technology will perform some of the most common tasks of both professions with greater speed and improved efficiency.

Trend: Regulatory Risk Lends Itself to the Use of Generative Artificial Intelligence

In our assessment, the category of ‘Likelihood of Regulatory Intervention’ is a proxy for multiple types of regulation that influence the performance of the role and the tasks associated with the job, including:

• State or Federal government oversight
• Governmental licensure
• Standards body policies (e.g., Generally Accepted Accounting Principles, Financial Industry Regulatory Authority, etc.)
• Enforced continuing education credits

Accountants and physicians, for example, face significant regulatory intervention while software developers, unless working in a regulated industry, face very little.

We conducted a regression analysis of our final scores in an effort to keep ourselves honest about our results. We used addition to determine our final scores so our variables were, as expected, closely aligned with the final result. More important was how significant each variable was in the final results. In our assessment, the most statistically significant contributor to the final result as demonstrated by the t-stat, a measure of the weight of the factor, was Likelihood of Regulatory Intervention.

Regulatory oversight means mistakes have consequences. Consequences impact the bottom line. We do not use a crystal ball so it is difficult to say absolutely, but it is likely that jobs facing regulatory risk will be among the most likely to be impacted by generative AI. Moreover, the less human-to-human interaction the jobs require, the more likely they are to be regulated as demonstrated by our second most statistically significant factor, Human-to-Human Engagement.

Trend: Roles Requiring Creation of New Artifacts are Less Likely to be Highly Automated (for Now)

One of the most interesting conclusions of this paper is that roles like software developer and marketing manager seem less likely, in the near term at least, to be eliminated by generative AI. This is counterintuitive. Many believe the profession of software development will be eliminated in the not too distant future. This will happen, pundits forecast, as a result of how effectively generative AI creates computer code. Given a scenario, tools like ChatGPT create highly efficient code. The same is true for graphic designers. Generative AI based systems like MidJourney, DALL-E and Stable Diffusion create graphical images across a range of styles (e.g., illustrations, diagrams, photographs, etc.) based on human provided prompts. It would seem creators’ days might be numbered — and they may very well be in the long term. What will protect creators in the near term is that their jobs are about more than the end product.

For most professional creative roles, there are multiple steps to be undertaken before creating the finished product. There are team members to exchange ideas with, clients to appease, and fellow creators to workshop ideas with. There is an ecosystem within the creation process that will not be immediately obviated by generative AI’s ability to create code, images or any other artifact. In fact, we may find that having these tools at their disposal augments the proficiency of creators just as it will for workers in other professions.

Impact Analysis

All of this analysis begs the question, why should people care now? The truth is, no one fully understands the potential consequences AI, and generative AI, specifically, will have on the workplace — and how long before the true impact is fully realized. After all, it was less than a year ago (as this paper was being written) that ChatGPT was made available and, thus, introducing laypeople to the concept of generative AI. What can we truly know a short time later?

Well, what we know for sure is that very few parts of our way of life will remain unaffected by AI (you’ll probably be safe if you live off the grid in a cabin in the woods, for example). Extrapolation allows us to make some predictions about the ramifications of this new technology across various sectors of our economy. So, even at the very start of this long journey, it is possible to forecast what comes next. We acknowledge that some of what we propose will happen and some of it will not but our outlook represents a best, educated guess about what we can expect.

Below, we suggest what we think leaders of various segments of the job pipeline to expect from how AI impacts their organizations:

Employers

• Employees will become more specialized as AI takes on common tasks — first in regulated industries and roles as a means to reduce risk, then more rapidly in other industries
• “Work” will be creation and human-to-human focused as technology takes on calculations and data analysis
• The “work day” and the “work week” may become more flexible as technology more quickly and efficiently executes many of the most frequently performed activities
• Data stewardship (i.e., data maintenance, integrity, model training, model maintenance, etc.) to become strategic core competencies distinct and separate from “Technology” — likely becoming a part of the C-Suite
• Tiers of seniority and requisite training will need to be rethought as the work of entry and junior level employees is taken on by artificial intelligence (e.g, how does a company acquire senior level workers if there aren’t enough junior employees in the pipeline)

Colleges and Universities

• Baccalaureate level curriculum will need to be more hands-on and real-world experience focused than they are now so that graduates leave school prepared to enter the workforce at a higher level than they would traditionally
• Curricula change to emphasize infusion of creativity and abstract thinking across every discipline and course, redirecting energy from rote learning to blue ocean thinking and strategic viewpoints
• Data understanding becomes a part of most curricula as we learn to value what data is, how we generate it is and how we use it
• Possibility exists that the duration of baccalaureate programs increases from four years to six or seven years to provide a longer timeline to train students and ensure they’re more mature upon graduation; could also lead to making college more affordable on a per annum basis
• Emphasis must be placed on determining how artificial intelligence (and other technologies) will affect specific disciplines. Even seemingly non-techy disciplines like art history and sociology can be affected by the rapid pace of technology transformation. That knowledge will both prepare the university to adjust its curriculum and help it forecast demand for individual domains
• Fewer students may enter college as the demand for workers in specific fields decreases until a homeostasis is reached between knowledge worker supply and demand
• Combination of generative artificial intelligence, reduced requirement of college education and greater access to alternative education resources (e.g., skill specific schools, MOOCs, etc) increases the likelihood of disruption of university business model
• Universities will likely rely more on professors who conduct research, develop products and create content rather than teaching. Teaching is often not a core competency at the university level so universities may find it advantageous to share teaching resources with each other while placing additional emphasis on the generation of ideas to sustain the institution

K-12 Schools

• Primary and secondary schools are preparing students for a future that will be drastically different from almost any period in the past century
• Standardized tests will become a reflection of the idea that while students need to understand basic concepts, technology will adeptly perform many of the activities where those concepts are most likely to be used
• Schools will start the challenging task of migrating curricula to Problem Based Learning, Project-Based Learning, Case Based Learning or even hybrid methodologies more likely to promote critical and abstract thinking as well exploration than traditional teaching methods — especially at the secondary level
• As fewer students attend college, secondary schools may find it necessary to return to introducing students to vocations as part of the high school curriculum
• Infusion of data gathering throughout the learning experience will allow more real-time communication between teachers, students and parents. Ultimately, data will identify trends and behaviors in students that provide insights into learning that can be applied to segments of the student population
• The possibility exists, with enough data, for learning insights for students to be significant enough to measure performance for students at the individual, school, district and state level on a rolling basis so as not to rely on periodical standardized tests for gauging achievement
• Automated, dynamic, adaptive lessons will result in the reordering of grade levels. Students may be sorted into age ranges and then placed in classrooms based on learning styles rather than strict grades. This flexibility will allow students to progress through school based on their ability rather than being having to abide by more rigid grade level standards

Governments

• Regulatory environment will have to change with regard to how generative AI is regulated and what it will be allowed to do (e.g., how will AI change dental procedures, what will AI be allowed to do, who will need to oversee procedures — a dentist or a less educated but qualified assistant) — including an official definition of Responsible AI
• The explosion of data will warrant explicit privacy rules covering both individual citizens and institutions
• Specific laws in criminal justice codes covering the theft, manipulation and improper access of data, data models and data methodologies
• Incentivizing the use of human workers — New jobs may be created as a result of AI but the impact generative AI has on the job market as the potential to be severe and governments may have to consider either incentivizing corporations to hire human workers or creating a universal basic income
• Reconsideration of productivity and fiscal metrics as indicators of economic health. Government measures like the unemployment rate and productivity, for example, may need to be reconsidered wholly or, at least, reformulated to allow for changing employment dynamics

Recommendations

It would be dishonest to offer avenues to pursue as if there was certainty associated with them. We stand at a moment where every path is unknown and somewhat difficult to forecast. It is possible, however, to specify courses of action that can prepare organizations of various types to adapt to the revolutionary transformation that is before us.

Make Data a Priority

We will soon live in a model-driven society. Whether models are used to understand customer behavior, create adaptive learning modules for students or optimize the creation of widgets on the factory floor, artificial intelligence models will be critical to the way life works. Data will be the differentiating factor for most businesses in the near future and, in fact, will be as important a factor in the way decisions are made as is the cost of goods and services sold (COGS).

Preparing for that inevitability means getting a handle on your data today. Every digital action conducted within your business generates data. The applications your employees use on their devices generates data as does the actions your customers take across your web presence. Maintaining your business as a competitive entity means using that data prudently. Among the ways you achieve that are clearly defining what data is to your organization.

Many companies take various types of data generated by the organization and data and put it into a database-like repository called a data lake. Unfortunately, data lakes can become swamps if not built carefully. This means, defining what data is to your organization, create data governance policies that explicitly define what data is, who has access to it and how it is managed, and establishing data maturity goals and milestones that acknowledge where your company is today and where you plan on going in the future.

Revisit three and five year strategic plans

When the Internet began to become popular, there were large companies that ignored it. Because it was new, the Internet seemed faddish. Obviously, it was not a fad and it has become integral to the way almost every organization operates. The same will be true for artificial intelligence. It is not a question of if your organization will begin using artificial intelligence strategically but when.

The long range plans your organization has developed likely do not include the data and artificial intelligence elements that will rapidly alter how business and the society work. In fact, a recent survey from Gartner, the noted technology research company, indicates that only 21% of the CEOs they surveyed believe AI will “significantly impact their business over the next three years.” Unfortunately, the other 79%, from our perspective, underestimate how important AI — and generative AI, specifically — will be to their businesses.

Among the expectations organizations should have about the strategic impact of generative AI on their businesses are:

• The products and services you offer, your sales pipeline and your distribution channels will all be affected by artificial intelligence as AI supports the organization’s ability to tailor engagements both just-in-time and predictively
• You will partner with other organizations across your ecosystem — and your customers’ journeys — to gain better access to data and the relevant models in a co-dependent way as data becomes key to insights
• Your organization’s core competencies will be aligned with your AI capabilities as you become more adroit at integrating AI into your systems and processes
• As data becomes more integrated into your systems, your company will make more informed, speedier decisions — allowing you to respond to the needs of the marketplace more effectively

Create AI War Room to Consider Generative AI

Generative AI is not merely a possibility. It is going to happen. What organizations have to determine is how they will respond to the eventuality. Thankfully, we’ve been here before.

When the Internet appeared, there were many questions about what the technology meant for corporate strategy. Famed Harvard Business School professor Michael Porter and author wrote in 2001 that:

We need to move away from the rhetoric about “Internet industries,” “e-business strategies,” and a “new economy” and see the Internet for what it is: an enabling technology — a powerful set of tools that can be used, wisely or unwisely, in almost any industry and as part of almost any strategy. We need to ask fundamental questions: Who will capture the economic benefits that the Internet creates? Will all the value end up going to customers, or will companies be able to reap a share of it? What will be the Internet’s impact on industry structure? Will it expand or shrink the pool of profits? And what will be its impact on strategy? Will the Internet bolster or erode the ability of companies to gain sustainable advantages over their competitors?

That sounds familiar, yes? We face the same questions about AI. This time, however, we have the Internet in our rearview mirror to look back on. One learning from that experience is that as told my MIT Sloan Management Review:

Previous research by Capgemini Consulting and MIT’s Center for Digital Business found that companies that invest in important new technologies and manage them well are more profitable than their industry peers. Respondents to our survey corroborate this view — they overwhelmingly believe that failure to effectively conduct digital transformation will harm their company’s ability to compete.

To prioritize understanding the strategic impact of generative AI on your business, create a war room focused on the technology. The assigned team should be a cross-functional group of staff dedicated to understanding the technology’s capabilities, assessing expectations of its possibilities within three to five years and gauging where within your organization it can be piloted and, ultimately, deployed to become a transformational part of your business.

Experiment with Generative AI

In tandem with the war room, it may benefit your organization to create a skunkworks that develops an organizational competency with the technology. Often skunkworks are considered semi-autonomous research and development units within the organization. In this situation, the recommendation would be to plan the effort with the intention of the team becoming a center of excellence (CoE).

In the foreseeable future, a tremendous amount of “noise” will occur in the generational AI space. This is even more so true when adding traditional predictive and descriptive AI. The players are far from established and more marketplace entrants will emerge with promises of the wonders of the technology. By experimenting early in an organized and deliberate way, you increase the likelihood of being able to distinguish real solutions from snake oil. Additionally, you are better able to develop processes and institutional knowledge that can be disseminated across the company. Ultimately, that preparation should allow the organization to attain a competitive advantage as it successfully deploys the technology more effectively than its competitors.

Weigh Regulatory Risks

One of the societal learnings of the dawn of the internet age is the importance of regulation. The internet largely remains unregulated but that lack of oversight has resulted in questionable practices that security and privacy advocates are committed to not revisiting with AI.

AI presents numerous challenges to privacy, security and intellectual property rights. Many questions remain unanswered about how to enforce ethical and fair use of the bountiful amounts of data that exists to train the models upon which AI depends. This data includes information that people reasonably expect will remain private and confidential. Moreover, some of the data/content has well-established intellectual property ownership that is currently muddying ownership and remuneration rights for both training AI models and the end results of those models’ calculations.

The government will have to weigh in on these challenges. Those decisions will affect how organizations acquire, access, maintain, manage, secure and use all manner of data. How will these decisions affect your business? Your organization will need to work with your industry/trade associations as well as your state and federal legislators to ensure the regulations that are instituted both support ethical use of data and allow organizations the figurative range of motion they need to use AI effectively.

Consider a New Definition of “Work”

From elementary education to college, a significant amount of our learning occurs by rote memorization. Many educators realize there are challenges with this approach to education but it has the benefit of being highly measurable through tests (rather than outputs). This approach carries over, currently, into the work we do.

Much of the work performed in enterprises is routine. In fact, for many back-office tasks, 50–75 percent of the tasks undertaken can be automated. Artificial intelligence is going to execute those activities more efficiently than most human workers are able to accomplish today. It will not benefit organizations to have as many human employees performing those tasks.

The impact will be that what we define as “work” will change. A fundamental conclusion of this report is that roles requiring the creation of or management of created artifacts will be somewhat buffered from the full impact of AI on the workforce. That is not to say those jobs will not be impacted. As of this writing, both the Writers Guild of America and the Screen Actors Guild are both on strike in an effort to protect their industries from the changes they expect generative AI to bring to Hollywood. There will be change.

We expect “work” will transform from the automatable tasks associated with many jobs to being more solutions oriented. In fact, we believe the roles of tomorrow will be concentrated in areas where critical and abstract thinking lead to solutions that move organizations forward. Where AI uses data to predict and describe behaviors, we expect human “work” will be assessing the accuracy of AI’s insights and then using that information to develop customer-focused new products and solutions. Much of the staff who might support that kind of process today will, likely, not exist. Organizations will have to begin considering what comprises “work” in their organization and the types of resources they will need to facilitate that transformation in a model-driven world.

Conclusion

Each new generation of technology changes our way of life. Sometimes that change is evolutionary and other times it’s revolutionary. Can you imagine what it was like when electricity first coursed through wires to provide power to private homes? What was it like to have light in one’s home whenever you wanted it? That obviously revolutionary technology profoundly changed society and helped usher us away from an agrarian society straight into the industrial revolution. It changed how we live in ways even scientists of the period could not have forecasted when those first electrons traveled through those wires. We face a similar moment now.

The promise of generative AI is a fantastical one. It seemingly brings us just a bit closer to general artificial intelligence (conscious AI) though that reality remains distant. In the meantime, what we have at our disposal now is sufficient to upend an economic and workplace status quo we take for granted.

Our analysis shows that white collar jobs will be significantly impacted by generative artificial intelligence. It does not appear that any of the jobs we assessed will be immediately replaced by AI. All of the jobs are nuanced and have specific aspects that would be difficult for the current generation of AI to do proficiently. That notwithstanding, there are tasks associated with each role — based on the job descriptions we used — that we believe could be — will be — automated.

The automation curve will be especially acute in jobs that face some sort of regulation. The automation likelihood becomes even greater when the tasks do not require human-to-human interaction. Tasks like calculating tax forms, creating legal briefs, and assessing patient dental x-rays do not require human-to-human engagement and make the relevant roles — accountant, attorney, dentist, respectively — more susceptible to automation than other roles we assessed.

It is our expectation that responding to the coming changes requires action across the areas of society that we outlined. Governments will need to make determinations about how data is used and how to regulate these new technologies. Employers will find themselves tasked with balancing automation with the societal need of employment. Colleges and universities will face both headwinds and opportunities as AI changes expectations for graduates’ careers. Primary and secondary schools will need to determine how to best prepare the next generation to survive in a world where the job market — and work, itself — potentially looks wildly different from what we know today.

The time is now to begin examining how your organization begins coming to terms with the change that is before us.

References

York, Erika. “Summary of the Latest Federal Income Tax Data, 2023 Update.“ Tax Foundation. January 26, 2023, https://taxfoundation.org/data/all/federal/summary-latest-federal-income-tax-data-2023-update/

“Gartner Survey Finds CEOs Cite AI as the Top Disruptive Technology Impacting Industries.” Gartner, May 17, 2023, https://www.gartner.com/en/newsroom/press-releases/2023-05-17-gartner-survey-finds-ceos-cite-ai-as-the-top-disruptive-technology-impacting-industries

Edelman, David C. and Abraham, Mark. “Generative AI Will Change Your Business. Here’s How to Adapt.” Harvard Business Review, April 12, 2023 https://hbr.org/2023/04/generative-ai-will-change-your-business-heres-how-to-adapt

Degnan, Oliver and Munoz, J. Mark, “5 Ways To An Effective AI Corporate Strategy.” California Management Review, Feb 5, 2020. https://cmr.berkeley.edu/2020/02/effective-ai-corporate-strategy/

Porter, Michael. “Strategy and the Internet.” Harvard Business Review, March 2001, https://hbr.org/2001/03/strategy-and-the-internet

Fitzgerald, Michael et al. “Embracing Digital Technology: A New Strategic Imperative.” Mit Sloan Management Review, October 7, 2013, https://sloanreview.mit.edu/projects/embracing-digital-technology/

Heath, Nick. “How to automate the enterprise: Your guide to getting started.” ZDNet, November 1, 2016, https://www.zdnet.com/article/how-to-automate-the-enterprise-your-guide-to-getting-started/

The Global Institute for the Advancement of Emerging Technology and Innovation (GIAETI) is a non-profit organization committed to innovation and the ethical usage of emerging technologies that can benefit society. We work with business, academic and government leaders to ensure they have the information they need to use new technologies effectively.

To learn more about GIAETI, visit www.giaeti.org or email us at info@giaeti.org

You can download this paper in PDF format here.