Predicted AI job apocalypse hasn't arrived, Guardian finds
TL;DR
- Anthropic CEO Dario Amodei predicted in May 2025 that half of entry-level white-collar jobs would vanish; a year on, that carnage has not shown up.
- Stanford finds unemployment among AI-exposed workers rose 0.77 percentage points since ChatGPT, less than the 0.85 point rise for least-exposed workers.
- Recent graduate unemployment sits at 5.6 percent versus 4.2 nationally, and 74 percent of employers now view AI skills as advantageous or required.
A year after two of the loudest CEOs in AI warned that white-collar work was about to be gutted, the payroll data still isn't showing it. The Guardian reports that since ChatGPT launched in 2022, unemployment among the workers most exposed to AI has risen by 0.77 percentage points, which is actually smaller than the 0.85 point rise for the workers least exposed, per a Stanford Institute for Economic Policy Research analysis cited in the piece.
That is a striking gap between forecast and reality. In May 2025, Anthropic CEO Dario Amodei predicted that half of entry-level white-collar jobs would vanish. A month later, OpenAI's Sam Altman said whole job categories were on the way out. What the data actually shows a year on is a labor market where exposed occupations are, per one researcher quoted by the Guardian, 'largely stable.'
The uglier picture sits at the bottom of the ladder. The article notes recent-graduate unemployment running at 5.6 percent versus a 4.2 percent national average, and a ZipRecruiter survey it cites finds 74 percent of employers now view AI skills as advantageous or required, with about half expecting candidates to arrive already fluent. NYU Stern's Robert Seamans frames the biggest effect so far as job change rather than job loss, while Bolt.new CEO Eric Simons is quoted saying three-person teams can now ship what used to take thirty to forty people.
A few caveats. The Stanford numbers only capture what shows up as unemployment, so any attrition that appears as slower hiring rather than layoffs is invisible in them. MIT's Paul Osterman is quoted warning that roughly 35 percent of the U.S. workforce already sits in easily-replaceable roles and that AI will push more work into contract and freelance status, exactly the kind of shift payroll unemployment misses. The Guardian piece also never puts a number on how many junior roles have been quietly removed from hiring plans rather than filled and cut.
For operators, the read is that the panic timeline was wrong but the ground under entry-level work is genuinely moving, and the leverage right now sits with anyone who can walk into a job already fluent in the tools.
Shared on Bluesky by 2 AI experts
-
Spoke to Varsha Bansal for The Guardian about how it’s not just displacement that we should be talking about in terms of AI’s impact on work, it’s also how AI can change jobs for the worse AND how workers need power and …
View on Bluesky →
Originally reported by theguardian.com
Read the original article →Original headline: AI was supposed to destroy jobs. Where’s the carnage?