AI and early-career hiring: What did the studies measure?

Compare what Statistics Canada, a U.S. Census working paper and Stanford–ADP measured about AI exposure, hiring and early-career employment.

A hiring rate tells you how often people enter jobs. An employment count tells you how many jobs are filled. These three studies use different age groups and comparison groups, and they estimate how exposed jobs are to AI. Their results show where hiring slowed and employment fell behind. Determining how much AI contributed would require evidence about when and how employers used it.

Three studies, three measures

Statistics Canada: Hiring and departures

The Labour Force Survey analysis compared six-month averages for employees aged 15 to 29 in Canada’s ten provinces. From the six months ending in November 2022 to those ending in August 2026, the hiring rate went from 4.0% to 3.3% in occupations with low potential AI exposure and from 2.5% to 1.9% in coding-intensive occupations. The coding decline began earlier in 2022. The study’s measure of departures includes quits and layoffs.

U.S. Census working paper: Adjusted employment

The paper used U.S. administrative records to track employment, hires and departures for workers aged 22 to 24, grouped by industry and state. It reported a 12% decline in adjusted employment over the ten quarters after ChatGPT launched for the fifth of groups most exposed to AI. Fewer hires accounted for most of that decline. Some employment trends began during the pandemic, which makes it harder to isolate the role of generative AI.

Stanford–ADP: A gap relative to less-exposed peers

The August 2026 revision used U.S. ADP payroll records through June 2026 and compared occupations for workers aged 22 to 25. Employment in the two most exposed fifths was 19% below where it would have been had it grown at the pace of the three less-exposed fifths. That gap mainly reflected fewer hires. The authors describe the estimate as descriptive. Earlier headlines of 13% and 16% used firm-adjusted estimates, so comparing them directly with 19% would mix methods.

How to read the next hiring headline

Ask who was counted, what the researchers measured and what they used as a comparison. A drop in the hiring rate, a decline in adjusted employment and a gap relative to another group answer different questions.