Friday, September 4, 2026
Technology News
Trusted, Accurate, and Validated
Home / Business  / Artificial Intelligence  / AI Is Quietly Closing the Career On-Ramp: Stanford’s Jobs Gap for the Young Hits 19%

AI Is Quietly Closing the Career On-Ramp: Stanford’s Jobs Gap for the Young Hits 19%

AI Is Quietly Closing the Career On-Ramp: Stanford’s Jobs Gap for the Young Hits 19%

TRENDING STORY

By the NewsCenter Desk

Published Tuesday, Aug. 25, 2026 · 12:32 p.m. ET · developing

The AI jobs apocalypse hasn’t arrived. Something quieter has. Employment for workers aged 22 to 25 in the most AI-exposed occupations now sits 19 percent below where it would be had it kept pace with their less-exposed peers, according to the August update of Stanford’s “Canaries in the Coal Mine” study, 140 pages of payroll data that reached the wider tech press through Ars Technica on Monday.

Experienced workers show no comparable gap. That asymmetry is the whole story.

The authors, Stanford economists Erik Brynjolfsson, Bharat Chandar and Ruyu Chen, tracked ADP payroll records covering millions of workers through June. Since late 2022, employment of 22-to-25-year-olds in the two most exposed fifths of occupations fell about 11 percent while the same age group grew about 10 percent everywhere else. Not through layoffs: separations fell, and pay trends show little difference. Companies simply pulled back on hiring at the bottom.

The declining jobs cluster where AI can do the task outright, a pattern the authors say is consistent with automation substituting for labor: accountants and auditors, receptionists. Where it assists instead, chief executives, registered nurses, employment holds or grows. The study draws that line from the Anthropic Economic Index, which maps anonymized Claude conversations onto government occupation data.

One number circulating this week needs correcting. Coverage frames the gap as widening from 13 percent last year to 19 now; those are two different yardsticks. On the paper’s own consistent measure, the move is 15 percent to 19 in about a year.

The authors call all of this “early, descriptive indicators—canaries in the coal mine—rather than causal estimates,” and note some of the divergence predates ChatGPT. Careful, and still worried: the effects are “real, persistent and widening,” Brynjolfsson told The Washington Post’s WP Intelligence last week, describing a labor market that is “quietly closing the on-ramp” for people starting their careers even as overall employment holds.

DEEPER COVERAGE

‣ The Stanford Digital Economy Lab’s own summary of the six facts. Read · Aug. 12

‣ Google’s own usage study, mapping 15 million Gemini interactions, reads the workplace calmer. Read · July 23

THE NEWSCENTER.IO TAKE · Aug. 25, 2026

Everyone is watching for the wrong apocalypse. Mass layoffs would show up in every chart there is; this shows up in almost none of them — employment grows, separations fall, pay trends barely move, and the market quietly stops making room for 22-year-olds.

If you run a company, ask yourself one question: when did you last hire someone entry-level, and where do you think your next generation of senior people comes from once everybody poaches instead of trains?

Top illustration: NewsCenter.io.

Disclosure: NewsCenter’s newsroom systems run on Claude, made by Anthropic, whose Economic Index is among the study’s data sources.