Research 2026-08-19

The gap between young workers in AI-exposed jobs and everyone else their age has widened to 19 percent

Stanford's Digital Economy Lab published an updated version of its 'Canaries in the Coal Mine' study on 12 August, extending its payroll data through June 2026. Employment among 22- to 25-year-olds in the occupations most exposed to AI now sits about 19 percent below where it would be had it kept pace with workers the same age in less-exposed occupations. That gap was 15 percent when the paper first appeared with data through July 2025. In absolute terms, employment for young workers in the two most-exposed quintiles fell roughly 11 percent between November 2022 and June 2026, while it grew about 10 percent in the least-exposed. The authors — Erik Brynjolfsson, Bharat Chandar and Ruyu Chen — work from ADP payroll records rather than surveys, which is why the series can be refreshed as new months arrive. The new version adds the distinction that does most of the explanatory work: the declines concentrate in occupations built on codified knowledge, the kind written down in manuals and textbooks, while roles that run on tacit knowledge show experienced workers gaining. The authors are explicit that these are descriptive patterns rather than causal estimates, that other explanations exist, and that the effect in their sample is larger than national benchmarks.

Why it mattersAlmost every claim about AI and jobs is either a forecast or an anecdote. This is neither: it is the same measurement, repeated on the same payroll data, and the number has moved in one direction over a year. It also says something more specific than 'AI takes jobs' — there is still no sign of broad displacement across the economy, and experienced workers are holding up. What is thinning out is the rung of the ladder where you were paid to learn the written-down part of a profession by doing it, because that is the part a model already has. If you are choosing a degree or hiring juniors, the split between codified and tacit knowledge is the one worth acting on.

✓ Verified · 3 sources

Machine learning read the shape of sick brain cells and picked out nine already-approved drugs that calmed them down
2026-08-21 Given four hours and a GPU to improve the way AI is trained, the best agent scored 0.25 out of 1 — and most never tried
2026-08-21 The agent invented a second person to vouch for its code. A 24-year-old in Texas refused to believe either of them.
2026-08-21 Pew put half a million web pages through a detector: a third of everything published since ChatGPT carries its marks
2026-08-21 The FDA has cleared 1,357 AI medical devices. Three of them have been tested on whether patients live longer or better.
2026-08-20

The gap between young workers in AI-exposed jobs and everyone else their age has widened to 19 percent

Related stories