🔍 Read the full analysis: After An AI Layoff, What Happens To Workers? Lessons From 40 Years Of Tech Shocks on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A Goldman Sachs note, reported by Fortune, estimates that workers likely displaced by AI since 2025 took about a month longer to find work than other laid-off workers. Goldman’s public research on earlier technology-related job losses points to lasting pay and earnings costs, though the new AI estimates rely on an indirect measure and remain preliminary.
A Goldman Sachs note dated October 29 estimates that workers likely displaced by AI since 2025 took about one month longer to find another job than other laid-off workers and were roughly 10 percentage points more likely to remain unemployed a year later. Fortune reported the findings; the note is not public, and Goldman’s method identifies likely AI displacement indirectly rather than through workers’ own reports. The results matter because they suggest that the cost of AI-related job loss may continue well beyond the layoff itself.
Goldman economist Pierfrancesco Mei based the reported estimates on Labor Department data covering more than 15,000 people who lost jobs between 2016 and 2025. Since there is no government survey asking whether AI caused a person’s job loss, the researchers combined an occupation’s exposure to AI, the speed of AI adoption in its industry, and whether the position was eliminated. They classified the top 30% of the resulting scores as “likely AI-displaced.”
According to Fortune’s account of the note, the longer job search was found among workers who lost jobs since 2025. Workers aged 35 and older took about five weeks longer to return to full-time work than comparable laid-off workers; those under 35 took about three weeks longer. Goldman also reported that the likely AI-displaced group was more likely to leave its former occupation and industry. Some reemployed workers, particularly older ones, moved into roles with similar or higher AI exposure.
The new estimates sit alongside Goldman’s public research on technology-related job losses over roughly four decades. That work found displaced workers took about a month longer to find work, lost more than 3% in real pay when reemployed, and had nearly 10 percentage points less real earnings growth over the following decade than workers who were never laid off. In that historical analysis, workers who retrained within three years had about two percentage points more real wage growth over ten years and a roughly 10-point lower chance of becoming unemployed again.
Infographic 01 · What happens after the pink slip
The first AI layoffs look like 40 years of tech layoffs
A longer search, a smaller paycheck and slower growth for years. Goldman’s new estimate for AI-exposed workers lines up with its long-run history.
The first AI wave, since 2025
Likely AI-displaced vs. other laid-off workers, US. Note not public; figures as reported by Fortune.
Forty years of tech shocks
Workers displaced from technology-disrupted occupations, US.
“Comparable in magnitude” to earlier waves of technological change, whose damage was “long-lasting.”Goldman Sachs economist Pierfrancesco Mei, as quoted by Fortune
How Goldman spots “AI-displaced” workers: no survey asks about AI, so it scores each job loss by the occupation’s AI exposure, the industry’s AI adoption and whether the position was eliminated, and treats the top 30% as likely AI-displaced (more than 15,000 workers, 2016–2025). An estimate, not a count.
Sources: Fortune, 30 Oct 2026 (reporting the Goldman note of 29 Oct 2026) · Goldman Sachs, Top of Mind #149, 25 Jun 2026, p. 15 · Goldman Sachs Insights, 11 Sep 2026. pp = percentage points.
Powered by Thorsten Meyer AI
The Costs Continue After Layoffs
The findings shift attention from how many jobs are cut to what happens to workers afterward: how long they search, whether they return to full-time work, and what they earn when they do. If AI-related displacement follows the historical pattern Goldman describes, a worker can be rehired and still face a lasting reduction in real pay or slower earnings growth.
The reported age differences also point to an uneven recovery. Older workers in the note took longer to regain full-time employment, while government figures cited in the source show a similar gap among long-tenured workers of different ages. Those figures cover all reasons for displacement, not AI-related layoffs specifically, so they provide background rather than proof of an AI effect.
For employers and policymakers, the distinction matters: a layoff count alone does not measure the full economic impact. The historical research links retraining with better outcomes, but it does not establish that training alone caused the improvement or guarantee that a particular program will help a displaced worker.
Products Worth Considering
Ledger Nano X Bluetooth Crypto Wallet
Ledger Stax Crypto Wallet with Touchscreen
Apple MacBook Air 13-inch with M5 chip, 16GB memory, 1TB SSD, and 12MP Center Stage camera (Midnight)
As an affiliate, we earn on qualifying purchases.
How Goldman Built the Comparison
Goldman’s recent estimate is an attempt to identify AI-related displacement in a dataset that does not ask workers why they lost their jobs. Researchers combined AI exposure, industry adoption, and position elimination into a score, then treated the highest-scoring 30% as likely AI-displaced. Goldman reportedly found no significant difference between that group and other laid-off workers for job losses before 2023, when AI adoption began rising. That check supports the timing of the comparison, but it does not turn the score into a direct record of why each job ended.
The longer historical comparison comes from Goldman research published in June 2026 and summarized in September. Mei and colleagues Jessica Rindels and David Mericle tracked workers across more than 300 occupations using National Longitudinal Surveys following over 20,000 people through their careers. The work examines earlier waves of technological change; it is evidence about past displacement, not a direct forecast of what AI will do to every occupation.
Separately, Bureau of Labor Statistics data in the source show that among long-tenured workers who regained full-time work, about 49% earned as much as before or more, down from about 62% two years earlier. The statistic describes reemployed long-tenured workers overall and should not be read as an AI-specific outcome.
““comparable in magnitude””
— Pierfrancesco Mei, Goldman Sachs economist, as quoted in Fortune
What the AI Estimates Cannot Show
The Goldman note is not publicly available, so the recent figures in this report are based on Fortune’s account and have not been independently checked against the original analysis. The underlying method also estimates AI displacement rather than confirming that AI caused each individual layoff. That makes the findings suggestive, not a definitive count of jobs lost to AI.
The source material does not provide the number of people in the likely AI-displaced group, detailed results by region or education, or enough information to assess every feature of the statistical method. It also remains unclear how outcomes will change as AI adoption spreads, how many workers will move into jobs with different AI exposure, and whether the reported gaps will persist over longer follow-up periods. The one-year unemployment comparison is not a measure of lifetime employment or earnings.
Tracking Reemployment and Pay
The next evidence to watch is whether future labor-market data show that longer job searches and lower reemployment rates continue for workers in roles highly exposed to AI. More detail on the Goldman note, including its sample and results by occupation, age, education, and region, would help readers judge how broadly the findings apply.
Longer follow-up is also needed to establish whether the reported differences translate into lasting wage and earnings losses for workers displaced since 2025. Goldman’s historical research suggests that retraining is associated with better outcomes, but the available information does not identify which programs work best or whether the same relationship will hold for AI-related displacement. For now, the clearest conclusion is limited: the early estimates point to a longer recovery for some likely AI-displaced workers, while the scale and duration of that effect remain uncertain.
Key Questions
What did Goldman Sachs find about workers likely displaced by AI?
According to Fortune’s report on a Goldman client note, workers classified as likely AI-displaced since 2025 took about one month longer to find new work than other laid-off workers and were roughly 10 percentage points more likely to remain unemployed after a year.
How did Goldman identify likely AI-related job losses?
Goldman reportedly combined occupation-level AI exposure, the rate of AI adoption in the industry, and whether a position was eliminated. The top 30% of scores were classified as likely AI-displaced. This is an indirect estimate, not a record of workers confirming AI caused their layoffs.
Are the reported figures confirmed by a public Goldman report?
No. The note was sent to clients and is not public. The figures for the recent AI comparison come from Fortune’s reporting; the source material says other coverage repeated Fortune’s numbers.
Goldman’s public research found that workers displaced by earlier technology changes took longer to find work and experienced lower real pay after reemployment, as well as slower real earnings growth over the following decade. Those historical results do not establish the eventual outcome for workers displaced by AI.
Does retraining guarantee a better outcome?
No. Goldman’s historical analysis linked retraining within three years to stronger wage growth and a lower chance of later unemployment. That association does not prove retraining caused the improvement, and the source does not identify a program that guarantees success.
Source: ThorstenMeyerAI.com
