ADP Research reports employment declines among young workers in AI-exposed jobs

ADP Research reports employment declines among young workers in AI-exposed jobs

ADP Research’s post was published August twenty-sixth, twenty twenty-five. It reports that employment fell for early-career workers in jobs it describes as highly exposed to AI, while employment for older workers in those jobs grew. The post draws on a paper by Dr. Erik Brynjolfsson and his team at the Stanford Digital Economy Lab, who used ADP payroll data to examine occupations, tasks and employment. For workers aged 22 to 25 in highly AI-exposed jobs, employment fell 6 percent between late 2022 and July 2025; among workers aged 30 and older, it grew 6 to 13 percent. The post also contrasts declines among young software developers and customer service workers with employment growth across age groups for health aides, whose jobs have little AI exposure. The post distinguishes AI that automates tasks from AI that augments work, and says early-career workers in roles with easily automated tasks are most vulnerable. Its closing argument is that this research can help employers and policymakers transition young workers toward careers and skills that stand to benefit from AI-driven productivity growth. The post reports employment patterns around the release of AI tools; it does not establish that AI caused those changes. It does not detail the basis for the specific employment estimates beyond its use of ADP payroll data.

Publication: a41468bf-d282-86a4-d483-3fcf76147456

attribution — pass: The reading attributes the post to ADP Research and says its findings come from Brynjolfsson's Stanford Digital Economy Lab paper using ADP payroll data. The closing argument is attributed as the post's own view. This is one piece, not a listing.

Dr. Erik Brynjolfsson and his team of researchers at the Stanford Digital Economy Lab add dramatic layers of additional information

summary — uncertain: The figures and the automate/augment split are accurate, and AI exposure is defined beside the reading. However, the reading drops the post's headline thesis that AI is affecting employment. It replaces it with a caveat that the post does not establish causation, which flattens the author's conviction. Smallest repair: state that the post asserts AI is affecting employment. Then present the causal gap as an open question. The phrase 'productivity boom ahead' was also softened to 'productivity growth'.

Employment for early career software developers and customer service workers fell dramatically after the release of AI tools

date — uncertain: The stated date of August 26, 2025 matches originalPublished at day precision, and there is no hindsight framing. However, the page notes an Oct. 6, 2025 correction to the customer service employment data. The reading cites the customer service decline of nearly 11 percent from the November 2022 peak, which likely reflects that later correction rather than the original post. The reading does not disclose this.

This post was updated Oct. 6, 2025, to correct employment data on career customer service workers.

support — pass: Each claim's excerpt appears verbatim and supports it: the 6 percent decline for ages 22 to 25, the 6 to 13 percent growth for workers 30 and older, the declines for software developers and customer service workers, the vulnerability of easily automated roles, and the policy argument. The question is a genuine open issue and is not leading. No unsupported benefits or costs are introduced.

In jobs with high AI exposure, employment for 22- to 25-year-olds fell 6 percent between late 2022 and July 2025.

media — pass: No media is selected, and no verified media candidates were supplied, so nothing is misidentified.