AI will have a major impact on labor markets. Here’s how the US can prepare

Markets are the killer app for efficiently organizing unfathomably complex human activities to deliver innovation and prosperity. They can also shift suddenly, creating winners and losers, even as broad measures of economic health rise.

We have some recent experience with this dichotomy. Turn-of-the-century automation breakthroughs and a globalized trade regime increased general prosperity — but they also left lasting and concentrated negative consequences for some workers and communities, especially in the industrial heartland. While it’s too early to say for certain, new technologies like artificial intelligence may cause similar disruptions.

Getting ahead of those shifts to capitalize on evolving opportunities and avoid reliving the painful lessons will require better labor market information. A recent American Enterprise Institute (AEI) paper argues we need new data foundations that provide timely, accurate insights about the ways regional and local economies are adapting to technology-driven employment and skills changes.

This is no simple task. The U.S. economy employs 161 million workers in 8.1 million establishments. Billions of independent individual and business decisions connect workers and firms. The result is the most productive workforce and economy the world has ever known.

Those labor markets depend on information. Inaccurate — or even just significantly lagged — data impairs decision-making, frustrates workers and employers, and creates inefficiencies that bog down the economy. Ensuring prosperity and opportunity during a period of high-velocity technological change requires new types of economic data to help all players make the best possible decisions.

Our nation already invests substantially in understanding national labor markets. Federal statistical agencies collect vast amounts of information that agency experts use to create a national economic portrait, including the supply and demand for workers and skills.

As a recent National Academies report on AI and the future of work noted, “substantial knowledge gaps and challenges remain” in the measurement and analysis of AI on jobs, skills and employment. Many of the nation’s analytic systems and reports are not geared to produce the forward-looking “headlight” measures decision-makers need to navigate ambiguous, rapidly evolving technological change. Federal statistical data usually tell us what happened, sometimes years ago, at a generalized national level. What we really need is prospective regional information to help employers, workers, and educational and training institutions make critical decisions with an eye on the future.

The 2007 introduction of the iPhone gives us some idea of the challenges we face. From app developers to Uber drivers to novel digital and gig work, new jobs emerged almost overnight requiring new skills and defying existing job classifications. AI may have even larger effects, creating new types of work and making obsolete some of the job categories we use to understand the labor market. These effects will require extensive changes to education and training, including substantial upskilling for incumbent workers, to help our businesses and workers prepare for the opportunities and changes AI will bring.

Without forward-looking data on emerging technologies, our painfully outdated federal labor market information (LMI) systems could become even less useful as a reference point for policy and practice. We are driving fast in the dark, and we need better headlights.

Building even a partial understanding of where we’re headed will require new types of data-integrated platforms and innovative tools to examine how AI may affect work and skills. A joint AEI-Stanford University-NYU project is proposing a nonpartisan, nonprofit institute similar in structure to the Frontier Model Forum, a D.C.-based nonprofit working on AI safety. The institute would be unique in its dual focus of integrating government, education and business workforce and skills data, as well as bringing in experts from business, government, and academia to develop new, data-validated analytical frameworks, measures, and programs.

This new institute would provide a bottom-up approach that assesses the impact AI may have on the nation’s labor markets and creates regionally tuned “headlight” reports on emerging job and skill requirements. The new information would transform our understanding of where we’re headed and help all stakeholders make better decisions about how workers, businesses and educators can prepare.

Through supportive partnerships with state LMI agencies, the new institute would help local experts develop and apply tailored information about AI’s impact on specific regions, industries, and jobs. Given the pace of technological change, no information or forecast will be perfect. But, the analysis provided through the institute would give directional guidance about future in-demand skills and support creation of regional testbed programs that can get ahead of the automation-skills curve for incumbent and future workers.

We need the innovations that AI will bring to boost employment and wage opportunities for millions of Americans and keep our economy competitive and growing. At the same time, shining some light on the future will help give “fair warning” on how AI is likely to affect jobs, skills and workers. In this way, we can help all our communities reap the benefits of technological change while avoiding some of the pain that inevitably follows in its wake.

Brent Orrell is a senior fellow at the American Enterprise Institute. Suzette Kent is CEO of Kent Advisory and served as the U.S. Federal Chief Information Officer from 2018 to 2020. Jason Owen-Smith is the executive director of the Institute for Research on Innovation and Science (IRIS), and a professor of sociology and research professor in the Institute for Social Research at the University of Michigan.