“Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence,” released by the Stanford Digital Economy Lab on November 13, 2025, was among the first large-scale studies to find AI-linked effects in actual US employment data rather than in surveys or lab experiments. It was written by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen.
The study draws on high-frequency, individual-level payroll records from ADP, the largest US payroll processor, covering the period since generative AI came into wide use in late 2022. Its central finding is that early-career workers aged 22 to 25 in the occupations most exposed to AI experienced a substantial relative decline in employment - reported as on the order of 13 to 16 percent depending on the specification - even after accounting for firm-level factors. Employment for older, more experienced workers in the same occupations, and for workers in less-exposed fields, held steady or kept growing.
The authors frame the young as the “canaries in the coal mine”: the first group where AI’s labor-market effect becomes visible. Two of their six facts are that the adjustment shows up in headcount rather than wages, and that declines concentrate in jobs where AI is more likely to automate work than to augment it. The lab later published follow-up analysis weighing alternative explanations such as interest rates and the timing of the shift. The paper became a touchstone in debates over whether AI is already reshaping the bottom rung of the career ladder, and is often read alongside the Anthropic Economic Index, which tracks how much real work people hand to AI.