The Industry of Ideas: Measuring How Artificial Intelligence Changes Labor Markets

By Julia Lane

American Enterprise Institute

June 09, 2023

  • Federal investments in new and emerging technologies—such as in artificial intelligence—have transformed the labor market. New “idea industries” that don’t fit neatly into traditional measures of industries and scientific fields have emerged.
  • This report describes a new, rapidly implementable, conceptual, and empirical approach to tracing how ideas move from investments in research to the marketplace and developing early warning indicators of potential workforce and education impacts.
  • This report proposes a new evidence-based foundation to support US national growth strategies and ensure investments have the greatest chance of success for workers and employers.

Read the PDF.


The launch of ChatGPT has captured the world’s imagination about the potential of artificial intelligence (AI): In just its first two months, over 30 million people used the tool, and roughly five million visited it per day. As Nobel laureate Daniel Kahneman said, “Clearly AI is going to win. . . . How people adjust is a fascinating problem.”1

How AI will transform businesses, workers, and jobs is not just fascinating but also a looming practical problem. ChatGPT alone could affect the jobs of 80 percent of workers to some degree and almost 20 percent to a large degree.2 The massive change in technology investments through targeted legislation such as the CHIPS and Science Act3 will necessitate that American training and education infrastructure be significantly more nimble to realize the promised rewards of quality jobs.

The costly lessons of the past, including “deaths of despair,”4 make clear that vulnerable workers displaced by AI—or by other critical and emerging technologies—should not be relegated to the unemployment slag heap; rather, these workers can find work in quality jobs if the right training opportunities are available. Firms should be able to find the right workers to respond to changing conditions and pay them well. American labor, education, and training institutions must be armed with evidence to respond to rapidly changing needs.5 Academic literature6 and practical guidance will be sorely needed to answer many practical questions.

On the demand side, those questions include:

  • Which sectors of the economy are at the cutting edge of AI?
  • How are AI capabilities changing jobs?
  • Does AI increase inequality or impede access to services?
  • What new career trajectories does AI create?
  • Are AI startup firms more likely to grow and survive than are other startup firms?

We need evidence-based answers to these questions so we know which research and development (R&D) investments are generating the intended outcomes and which are failing.

On the supply side, questions include:

  • Will the future demand for repetitive mental tasks such as coding and searching disappear?
  • Will new jobs be created using different skills even for workers in white-collar, high-skill jobs?
  • How can workforce training be flexible enough to respond to changing demand and ensure that federal investments in AI generate quality employment and economic growth?7

We must address these questions so we know which training programs are working, and which are not, for different types of workers and regions of the country.8

The immediate challenge is that new measurement frameworks are needed. AI—like many other emerging technologies—is not categorized in any standard way.9 As a result, American decision makers cannot rely on standard statistics about the levels and trends of the AI workforce or the resulting education needs. They lack valid information for informed decision-making.

This report proposes a new, rapidly implementable, conceptual, and empirical approach. New technologies are based on ideas, which produce value. So tracking new “idea industries” requires tracking how ideas move from the original science and technology investments in research universities into the marketplace10 and developing early warning indicators of potential workforce and education impacts. AI is a classic example: It began with grant funding and then took decades to grow into an economic and social game changer.11

Figure 1 provides a visual overview of the framework and data that can be used to trace that journey from idea to job impact. Idea creation—in this case, the cuttingedge research in AI—can be identified through studying funded grants, the resulting papers and patents, and the associated authors and grant recipients, as well as the words they use to describe their work, as shown in Figure 1, Panel A.12 This information, combined with university data, can then be used to determine all inputs used to perform AI research—not only labor (including from graduate students, postdocs, and clinical researchers)13 but also the high-tech capital equipment, materials, and services that high-tech vendors supply to those grants, as described in Figure 1, Panel B. Tracking the next step of the journey—the transmission of ideas into the business world—requires identifying the private-sector AI-intensive firms: the firms that hire the AI-skilled scientists and researchers and the high-tech AI vendors, as shown in Figure 1, Panel C.

The next step is characterizing the quality of jobs in these AI-intensive firms. That requires tracing the earnings (and earnings growth) and the duration of jobs for workers in the firms, as described in Figure 1, Panel D. The final step, as shown in Figure 1, Panel E, is to characterize what educational credentials are needed in the AI-intensive workforce so educational institutions can be positioned to provide appropriate training.14

The following section unpacks the framework in more detail, focusing on the available authoritative data that could be compiled to answer these questions. The concluding section describes an implementation road map.

Read the full report.


  1. Ajay Agrawal, Joshua Gans, and Avi Goldfarb, “ChatGPT and How AI Disrupts Industries,” Harvard Business Review, December 12, 2022,
  2. Tyna Eloundou et al., “GPTs Are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models” (working paper, University of Pennsylvania, Philadelphia, PA, March 27, 2023),
  3. Mark Muro et al., “Breaking Down an $80 Billion Surge in Place-Based Industrial Policy,” Brookings Institution, December 15, 2022,; Matt Mazewski and Christian Flores, Economic Impacts of the CHIPS for America Act, Data for Progress, May 2022,; and Steven Brint, “The US ‘CHIPS and Science’ Act Launches Industrial Policy as Counter to China,” International Higher Education, no. 113 (Winter 2023): 9–10,!/action/getPdfOfArticle/articleID/3613/productID/29/filename/article-id-3613.pdf.
  4. Morgan R. Frank et al., “Toward Understanding the Impact of Artificial Intelligence on Labor,” Proceedings of the National Academy of Sciences 116, no. 14 (April 2, 2019): 6531–39,
  5. Julia Lane, Reimagining Labor Market Information: A National Collaborative for Local Workforce Information, American Enterprise Institute, March 9, 2023,
  6. Frank et al., “Toward Understanding the Impact of Artificial Intelligence on Labor”; and Daron Acemoglu et al., “Artificial Intelligence and Jobs: Evidence from Online Vacancies,” Journal of Labor Economics 40, no. S1 (April 2022): S293–S340,
  7. Tim Kautz et al., “Fostering and Measuring Skills: Improving Cognitive and Non-Cognitive Skills to Promote Lifetime Success” (working paper, National Bureau of Economic Research, Cambridge, MA, April 2015),; and National Artificial Intelligence Research Resource Task Force, Strengthening and Democratizing the U.S. Artificial Intelligence Innovation Ecosystem: An Implementation Plan for a National Artificial Intelligence Research Resource, White House Office of Science and Technology Policy, National Artificial Intelligence Initiative Office, January 2023,
  8. Ben Wodecki, “WAICF ’23: Yann LeCun Sets Us Straight on Generative AI,” AI Business, February 10, 2023,
  9. Industries are typically based on what or how goods and services are produced. See Robert E. Yuskavage, Converting Historical Industry Time Series Data from SIC to NAICS, US Department of Commerce, Bureau of Economic Analysis, November 5, 2007, 12–13,; and Maureen A. Haver, “The Statistics Corner: The NAICS Is Coming. Will We Be Ready?,” Business Economics 32, no. 4 (October 1997): 63–65.
  10. Jason Owen-Smith, Research Universities and the Public Good: Discovery for an Uncertain Future (Stanford, CA: Stanford University Press, 2018).
  11. National Research Council, Computer Science and Telecommunications Board, Funding a Revolution: Government Support for Computing Research (Washington, DC: National Academies Press, 1999),; and James Moor, “The Dartmouth College Artificial Intelligence
    Conference: The Next Fifty Years,” AI Magazine 27, no. 4 (Winter 2006): 87–91,
  12. National Research Council, Funding a Revolution; and Moor, “The Dartmouth College Artificial Intelligence Conference.”
  13. Similarly, Google was started by Larry Page, “a graduate student supported by an NSF [National Science Foundation] digital library project at Stanford University.” See National Science Foundation, “On the Origins of Google,” August 17, 2004,
  14. Kautz et al., “Fostering and Measuring Skills.”