Evaluating the Impact of AI on the Labor Market

evidence

Comparing the pace of change in the United States occupational and industry mix in the 33 months following ChatGPT's release against equivalent periods during the diffusion of computers and of the internet, the Budget Lab found occupational dissimilarity, industry dissimilarity and exposure measures flat or within historical ranges, and no sign that measures of AI exposure, automation or augmentation were related to changes in employment or unemployment to date; the authors note that historical workplace technological disruption has unfolded over decades and that 33 months is a short window. V55 verification basis: budgetlab.yale.edu was not fetched. The findings and framing were carried in search results attributing them to the Budget Lab's own publications; the October 2025 date of the principal analysis comes from secondary reporting and was not confirmed against the publication itself.

A evidence is not a place. Drawing it on a map would assert something about the world that no stored fact supports.