Stanford study finds AI adoption linked to falling employment for young workers
Software developers aged 22 to 25 saw the sharpest declines; the authors found no comparable fall in wages, meaning employers cut headcount rather than pay.
- Culture & impact
- Major
Economists Erik Brynjolfsson, Bharat Chandar and Ruyu Chen published a working paper, “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence,” using high-frequency payroll data from ADP, the largest US payroll processor. It found that early-career workers aged 22 to 25 in occupations most exposed to generative AI — including software development, customer service and other roles susceptible to automation by large language models — had experienced a 16% relative decline in employment since late 2022, even after controlling for firm-level shocks such as layoffs affecting a whole company regardless of role. Employment among older, more experienced workers in the same occupations, and among young workers in less AI-exposed occupations, remained stable or continued to grow over the same period.
The authors distinguished occupations where AI was more likely to automate tasks — replacing what a worker did — from those where it was more likely to augment a worker’s output, and found the employment decline concentrated in the former. They reported no comparable fall in wages for affected workers, which they read as evidence that employers were adjusting headcount rather than pay in response to AI adoption, and that the effect held up when the analysis excluded technology-sector firms and remote-only roles, arguing against alternative explanations such as post-pandemic remote-work normalisation or a tech-sector-specific downturn.
The paper’s authors likened young workers in exposed occupations to canaries in a coal mine — the first group to register a labour-market effect that might later spread more broadly — and the study was widely cited in the following months as among the first evidence, using large-scale administrative rather than survey data, that generative AI adoption was already showing up as reduced entry-level hiring rather than only as a subject of employer surveys and economic forecasting.