AI Exposure Isn’t Cutting Jobs Yet

Updated: 2025.10.17 5D ago 11 sources
An Economic Innovation Group analysis by Sarah Eckhardt and Nathan Goldschlag finds that occupations most exposed to AI are not seeing higher unemployment, labor force exits, or occupation-switching compared to less-exposed jobs. In fact, unemployment has risen more among the least-exposed quintile, and exposed workers are not fleeing to lower-exposure roles. Early claims of AI-driven displacement in U.S. labor markets are not supported by observable trends to date. — This tempers automation panic and redirects policy toward measured, evidence-based responses rather than premature plans for mass displacement.

Sources

Predicting Job Loss?
Alex Tabarrok 2025.10.17 40% relevant
Both temper sensational claims about imminent AI-driven job loss; this article adds that historical occupational forecasts barely beat simple trends, reinforcing skepticism toward near-term, dramatic AI displacement predictions.
Senate Dem Report Finds Almost 100 Million Jobs Could Be Lost To AI
BeauHD 2025.10.07 70% relevant
The Senate HELP staff report claims AI will replace nearly 100 million jobs and cites ChatGPT-generated occupation loss rates (e.g., 89% of fast‑food roles), which contrasts with EIG’s finding that AI‑exposed occupations are not yet seeing higher unemployment or exits.
AI is Not Killing Jobs, Finds New US Study
msmash 2025.10.01 88% relevant
The article summarizes a Yale Budget Lab and Brookings study concluding that generative AI has not caused higher unemployment or large job losses since late 2022, aligning with prior evidence (EIG: Eckhardt/Goldschlag) that AI‑exposed occupations aren’t seeing elevated unemployment, exits, or occupation switching.
Links for 2025-09-26
Alexander Kruel 2025.09.26 70% relevant
The linked radiology piece reports more radiologist jobs and a 48% wage increase despite rapid AI uptake, reinforcing evidence that AI exposure has not produced net job losses in a flagship high‑exposure occupation.
Top Economists Agree That Gen Z's Hiring Nightmare Is Real
BeauHD 2025.09.22 85% relevant
Powell, UBS’s Donovan, and Goldman’s Mei argue the youth hiring slump is mainly a low‑turnover, slow‑growth story and that AI 'may be part of the story' but isn’t the primary driver—echoing evidence that AI‑exposed occupations aren’t seeing outsized unemployment so far.
Is AI making it harder to enter the labor market?
Tyler Cowen 2025.08.26 85% relevant
This new study reports a 13% relative employment decline for 22–25-year-olds in the most AI‑exposed jobs, with effects concentrated where AI automates rather than augments—contradicting the earlier EIG finding of no higher unemployment in AI‑exposed occupations by revealing hidden cohort effects and using payroll‑provider data rather than broad labor-force stats.
How Retrainable are AI-Exposed Workers?
Tyler Cowen 2025.08.25 72% relevant
The paper reports positive earnings returns for AI‑exposed workers who retrain and estimates 25%–40% of occupations are 'AI retrainable,' reinforcing evidence that AI exposure hasn’t yet produced broad labor‑market harm and that adaptation is feasible.
At least five interesting things: Cool research edition (#68)
Noah Smith 2025.08.12 100% relevant
EIG graphs showing unemployment and exits by AI-exposure quintile (1–5) and reduced switching from high-exposure occupations after generative AI’s rollout.
Who will actually profit from the AI boom?
Noah Smith 2025.08.10 60% relevant
Both pieces temper sweeping AI predictions using real-world indicators; here, moderate PE ratios for Nvidia, the big clouds, and AI labs suggest markets don’t foresee runaway profits, paralleling evidence that AI-exposed jobs aren’t showing mass displacement.
Nikolai Yakovenko: the $200 million AI engineer
Razib Khan 2025.07.12 70% relevant
The guests note that nearly three years into the hype cycle there’s still no AGI and no clear 'killer app' transforming knowledge work, aligning with data that AI-exposed occupations have not seen unusual displacement to date.
Making AI Work: Leadership, Lab, and Crowd
Ethan Mollick 2025.05.22 70% relevant
The article notes companies report only small-to-moderate gains and 'no major impact on wages or hours' through end‑2024, aligning with evidence that AI‑exposed occupations aren’t seeing higher unemployment or exits; it adds a mechanism (organizational adoption failure) for the macro null effect.
← Back to All Ideas