A Stanford University research paper published in August 2025, led by economist Erik Brynjolfsson, analyzed high-frequency ADP payroll data across millions of US workers and arrived at a conclusion that runs counter to the prevailing panic: generative AI’s overall impact on aggregate employment has been relatively modest. The study, titled “Canaries in the Coal Mine?”, found that while AI is reshaping parts of the labor market, the wholesale destruction of white-collar work remains more narrative than reality.
Young workers are absorbing the shock
The Stanford research found that employment among early-career workers aged 22 to 25 in AI-exposed occupations declined by 13 to 16% since late 2022. Software developers in those early-career cohorts got hit even harder, facing employment reductions of nearly 20%. These are roles where AI automates tasks rather than augments the person doing them.
Workers in jobs where AI complements human activity saw stable or even increased employment growth. Multiple analyses using Current Population Survey data corroborate the Stanford findings, showing only modest changes in hiring trends for AI-exposed roles across the broader workforce.
Tech layoffs tell a messier story
