Stanford Lab Finds 22-25 Workers Lose Jobs in Top 20% AI-Exposed Roles Post-ChatGPT
Updated
Updated · digitaleconomy.stanford.edu · Jun 10
Stanford Lab Finds 22-25 Workers Lose Jobs in Top 20% AI-Exposed Roles Post-ChatGPT
3 articles · Updated · digitaleconomy.stanford.edu · Jun 10
Summary
Stanford’s updated Canaries Dashboard shows workers aged 22-25 saw noticeable employment declines in the two most AI-exposed occupation groups after ChatGPT’s November 2022 launch, while the other three exposure groups still posted growth.
ADP-backed data suggest the divergence is concentrated at the start of careers: the pattern weakens for ages 26-30, is muted up to 34, and largely disappears for older workers.
Software developers and customer service representatives—both in the highest-exposure group—show substantial declines for younger workers, while less-exposed jobs such as home health aides show gains and stock clerks show little age-related pattern.
The dashboard also finds employment is weaker where AI use is more automation-heavy rather than augmentation-heavy, though Stanford says the results show correlation in a large but non-representative sample, not proof that AI caused the declines.
The monthly tracker draws on a balanced sample of 25,000 ADP client firms covering 4.6 million workers and more than 730 occupations, aiming to provide an early signal of broader labor-market shifts.