Memorandum
- From
- Delaney Sawyer via Fortune | FORTUNE
- Date
- Filed
- Business·4 min to read
- Re
Survey: Only 3% of U.S. Workers Say They Lost a Job to AI
ReSurvey: Only 3% of U.S. Workers Say They Lost a Job to AI
A new survey of 1,250 U.S. workers finds that four years after ChatGPT's debut, AI has caused job losses for only about 3% of respondents, while 6% gained new AI-related jobs and 9% received promotions tied to AI.
Four years after ChatGPT burst onto the scene, artificial intelligence has not yet triggered the wave of job losses — or gains — that many predicted, according to a new survey of U.S. workers. The survey, conducted by YouGov for sociologist Jeffrey C. Dixon of the College of the Holy Cross, found that only about 3% of employed Americans said they had lost a job due to AI since 2023.
The findings, based on responses from 1,250 employed U.S. adults surveyed between July 30 and Aug. 4, 2026, paint a more nuanced picture than either AI pessimists or optimists might expect. While 3% reported job losses tied to AI, roughly 6% said they had landed a job that did not exist before AI, such as an AI data labeler or AI compliance specialist. About 9% indicated they had received a promotion or advancement they believed was related to their AI skills.
Dixon, who researches AI and commissioned the survey as part of an ongoing study that has not yet been peer reviewed, said the results surprised him. “For AI pessimists like me, it’s surprising that so few of the surveyed workers said they had lost jobs due to AI,” he wrote in the study, which was republished by Fortune.
The survey asked workers three specific questions about AI’s role in their job changes since 2023: whether they had lost a job due to AI automating their previous tasks, whether they had landed a new job that did not exist before AI, and whether they had received a promotion related to their AI skills. Approximately 95% said no to the first question, 90% said no to the second, and 88% said no to the third. Between 3% and 4% were “not sure” they had experienced any of these changes.
These results align with other research in the U.S. and abroad suggesting that AI is not yet having widespread effects on the labor market. Dixon notes that job changes of any kind are relatively rare in the U.S., so the small percentages must be viewed in that broader context. The most significant impact so far appears to be the promotion channel, with workers’ AI skills potentially helping them climb the career ladder — though it is too early to know whether that advantage will last.
Dixon also points to the rise of agentic AI, a newer form of AI that can act autonomously and automate jobs to a greater extent, which may render some current AI expertise less valuable in the future. The survey began in July 2026, when the official unemployment rate was 4.1%, but the workforce was shrinking and AI was reportedly a leading reason for many layoffs.
The study has limitations: it relies on workers’ self-reported experiences, which can be difficult to attribute definitively to AI, and it excludes unemployed individuals. But Dixon says his approach moves beyond many predictions made by economists and analyses produced by AI companies by asking workers themselves how they are experiencing AI. Unlike some research that focuses only on job loss, his study also examines job gains and promotions.
Dixon is currently analyzing additional survey data to answer questions such as who is getting promoted, how workers are using AI on the job, and how they think AI will affect their future opportunities.
