AI eats leisure time, makes employees work more, study finds (www.theregister.com)

A new paper by economists Wei Jiang, Junyoung Park, Rachel (Jiqiu) Xiao and Shen Zhang finds that generative AI—exemplified by ChatGPT—has been associated with longer workweeks and less leisure for workers in occupations more exposed to the technology. Using U.S. Bureau of Labor Statistics American Time Use Survey data (2004–2023), which records 15‑minute activity intervals, the authors link occupations to AI exposure and exploit the arrival of ChatGPT in 2022 to measure changes. They report that an interquartile increase in generative‑AI exposure corresponds to about +3.15 hours of work and −3.20 hours of leisure per week in 2022–23. The results challenge the expectation that AI will shorten workdays: productivity gains appear to accrue to firms and consumers more than to employees. The paper also finds higher wages but lower job satisfaction for AI‑exposed workers and a post‑pandemic increase in hours among remote workers subject to AI surveillance—suggesting monitoring tools amplify effort. Technically, the study combines occupation‑level AI complementarity measures with time‑use microdata to identify distributional effects, highlighting how labor market bargaining power and product‑market competition channel AI rents away from workers—raising policy questions about workplace monitoring, compensation, and the equitable sharing of AI productivity gains.