Yale tracker finds no clear broad AI disruption, flags unemployment uncertainty

What changed in this update

Yale's Budget Lab updated its tracker on October 14 with September Current Population Survey microdata. Occupational composition, employment and wages did not provide clear evidence of large-scale AI-related disruption. However, unemployment growth among AI-exposed workers was elevated relative to comparable unexposed workers in the third quarter. The researchers describe those estimates as volatile and caution that other measures do not corroborate the signal.

Sources: The Budget Lab at Yale — Tracking the Impact of AI on the Labor Market; The Budget Lab at Yale — AI Labor Market Tracker: October 14 update

How the comparison works

The linked May 7 methods compare occupations in the highest third of AI exposure with a weighted group from the lowest third, matching earlier outcome patterns. The analysis examines divergence after late 2022. Exposure describes potential task effects, not verified adoption or a probability of losing a job.

The approach can miss offsetting gains and losses within groups. Narrow subgroups are harder to study with this survey, while assigning occupations to unemployed people presents difficulties. Other simultaneous economic changes can also complicate attribution. Failure to detect broad disruption does not prove that AI has no effects.

Sources: The Budget Lab at Yale

Our interpretation: keep two levels of evidence in view

This update is useful precisely because a national pattern and a difficult individual experience can coexist. Someone facing fewer interviews does not need to dismiss that experience because a broad indicator looks stable. Equally, a team losing colleagues does not establish what is happening across the economy. Moving between those levels requires additional evidence.

For a career decision, we would keep the national finding beside a short record of local changes: which vacancies remain open, which responsibilities have changed, and what employers actually say they need. Avoid assigning every change to AI before considering demand, budgets or a reorganization. The purpose is to identify information that could change a decision, not to turn a tracker into reassurance or alarm.