Brookings Publishes Study on Early Labor Market Change Under Generative AI

October 6, 2026, 15:30Source: Brookings (AI)

The Brookings Institution has published "Still waters, rapid currents: Early labor market transformation under generative AI," by economists Anders Humlum and Emilie Vestergaard. The page lists the research the authors build on, but the available text does not set out their findings.

Key points

Brookings published "Still waters, rapid currents: Early labor market transformation under generative AI."
The authors are Anders Humlum (University of Chicago) and Emilie Vestergaard (University of Copenhagen).
A working paper with the same title is listed as NBER Working Paper No. 33777 (2026).
The cited literature covers AI productivity, entry-level jobs, and earlier technology shifts such as computerization.
The available text contains the references but not the study's findings.

Who wrote it and where it comes from

The article is credited to Anders Humlum, Associate Professor of Economics at the University of Chicago, and Emilie Vestergaard, a PhD student at the University of Copenhagen. Brookings files it under Artificial Intelligence, the U.S. economy, and labor and unemployment, within its Economic Studies program.

The reference list points to an NBER working paper by the same authors with the same title: Humlum, A., and Vestergaard, E. (2026), "Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI," NBER Working Paper No. 33777. The two researchers also co-authored a 2025 paper in the Proceedings of the National Academy of Sciences titled "The unequal adoption of ChatGPT exacerbates existing inequalities among workers."

The text available for this report consists mainly of the bibliography and page navigation. It does not include the study's data, conclusions or policy recommendations, so this report does not describe them.

The research the piece draws on

The references show how broad the surrounding debate is. They include work on the macroeconomic effects of AI, such as Daron Acemoglu's "The Simple Macroeconomics of AI" (2025), and on automation and work, such as Acemoglu and Restrepo (2018).

Several entries concern how generative AI affects workers directly:

Noy and Zhang (2023), on the productivity effects of generative AI, published in Science.
Brynjolfsson, Li and Raymond (2025), "Generative AI at Work," in The Quarterly Journal of Economics.
Eloundou and co-authors (2024), on the labor market impact potential of large language models.

Others look at early-career and entry-level jobs. They include "Canaries in the Coal Mine?" from Stanford's Digital Economy Lab, "Looking for the Ladder: Is AI Impacting Entry-Level Jobs?" from the Economic Innovation Group, and a 2026 paper titled "AI-Exposed Jobs Deteriorated before ChatGPT."

The list also reaches back to earlier technology shifts, including Robert Solow's 1987 "We'd Better Watch Out," Paul David's 1990 paper on the dynamo and the computer, and studies of how computers changed the wage structure. It also cites an Axios interview with Dario Amodei, titled "Behind the Curtain: A White-Collar Bloodbath."

Related Brookings work and what to watch

The page links to other recent Brookings material on AI and work. One is "Workforce policy for the age of AI," by Guy Ben-Ishai and Neil C. Thompson, dated September 15, 2026. Another is a past event, "AI and economic mobility: Opportunities and challenges," held June 10, 2026 at the Brookings Institution in Washington, D.C.

Brookings states that each of its publications represents the sole views of its authors. The acknowledgements credit Aidan Conley for research and Ranga Krishnamurthy, Sanjay Patnaik, Robert Seamans and Mike Wiley for editorial support.

For readers following the debate over AI and jobs, the piece is one to read in full. The title and reference list show the authors are examining early labor market effects of generative AI against both recent evidence and earlier technology transitions. The specific results are in the full Brookings article and the NBER working paper No. 33777.