In 2025, leaders in artificial intelligence made bold predictions about jobs. Anthropic chief executive Dario Amodei said half of entry-level white-collar jobs could disappear. OpenAI's Sam Altman said certain job categories would end completely.
Companies began citing AI when announcing layoffs. Workers organized in response. Students rethought their career plans.
A year later, that wave of job losses has not appeared. AI tools have kept advancing quickly, but the job market has not moved at the same pace.
Jobs Are Changing, Not Disappearing
A Stanford Institute for Economic Policy Research report looked at unemployment since 2022. Workers most exposed to AI saw unemployment rise by 0.77 percentage points. Workers least exposed saw a slightly bigger rise, 0.85 points.
Recent college graduates have a higher unemployment rate than the national average. Researchers say AI may play a role, but remote work trends and past overhiring likely matter too.
Erika McEntarfer, a Stanford Institute fellow, said the pattern looks similar to past tech shifts. She noted the computer revolution took decades to fully change the workforce.
Government job data is also slow to update. This makes it hard to measure AI's real effect right now.
Even without mass layoffs, hiring habits are shifting. About 74% of employers say AI skills are an advantage or a requirement, according to ZipRecruiter. Half expect new hires to already use AI tools well on day one.
ZipRecruiter economist Nicole Bachaud said the bigger issue is matching skills, not a shrinking number of jobs. Companies are adding and cutting roles within the same departments as they figure out what they need.
Stanford's Nicholas Bloom calls this turbulence. AI removes some jobs while creating others focused on building and fixing AI systems.
NYU's Robert Seamans sorts AI's impact into three groups: jobs lost, jobs created, and jobs changed. He said the third group is by far the largest.
Long Hours Inside AI Companies
Some tech leaders have said AI would soon shrink the work week. A Google engineering director predicted a four-day week by 2025. OpenAI later urged companies to try the same idea.
A former OpenAI employee told the BBC the company never tested a four-day week itself. The person described long weekend hours and frequent tense performance reviews.
Workers at OpenAI and Anthropic told the BBC that busy periods, called sprints, can top 90 hours a week. Both companies did not respond to requests for comment.
At Meta, some staff say they were moved onto AI teams without being asked, a practice they call being drafted. One employee said refusing meant quitting.
A UC Berkeley study followed tech workers using AI tools for eight months. It found they worked faster, took on more tasks, and worked longer hours overall.
MIT scholar Neil Thompson said time saved by AI tools often gets used on checking AI output instead of resting. Former Google employee Amin Shali left his job in May, citing AI related strain on his team.
Shali said his sleep and health improved after leaving. He said AI was expected to give people more free time, but for many workers, the opposite has happened so far.
