For two years, the people building AI told us it was coming for our jobs. This month, the same people started saying it will create them.
Speaking in Paris, Jeff Bezos said AI would cause “a labour shortage”, not mass unemployment, and would unlock almost endless demand for builders and entrepreneurs. Days earlier, Sam Altman said he was “delighted to be wrong” about one of his biggest fears: that AI would rapidly wipe out white-collar work.
As Gizmodo neatly put it, the chief executives have stopped publicly threatening to replace us.
The script has flipped
This is a sharp turn. Not long ago, Altman warned that “whole categories of jobs” would disappear. Other bosses raced to outdo each other on how much of the workforce their models would soon make redundant, even as graduates walked into the worst entry-level market in years. Now the mood music is all opportunity and human-AI partnership.
Why the change, and why now
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
The timing is doing a lot of work. OpenAI and Anthropic are heading towards enormous public listings, and you do not sell a jobs apocalypse to retail investors. A worker backlash and circling regulators have made doom talk costly, while optimism plays far better on a conference stage.
It is worth remembering who is speaking, too. Bezos now runs Prometheus, a $41bn AI company. “AI creates jobs” is not just a prediction from him. It is a pitch.
The data is messier than either slogan
Look past the framing and the truth is duller than apocalypse or utopia. A new PwC study found AI splitting the labour market in two: it rewards firms that use it to enhance their people, and leaves behind those that use it only to cut costs. That is not collapse. It is redistribution.
The losses, though, are real
The tech sector shed hundreds of thousands of roles in 2025, and graduate unemployment has stayed stubbornly high. The people caught in the gap are real as well. The retailer Rainbow warned its fashion models that “fewer people will be needed”, then the models watched AI-generated versions of themselves appear on the company’s website.
Watch what they build, not what they say
Both versions of the story are partly public relations. The doom narrative flattered the technology, our models are so powerful they threaten your livelihood, and even gave cover for layoffs that were really about spending. Mark Zuckerberg admitted Meta’s cuts were driven by capital expenditure, not AI productivity. The new optimism flatters the IPO.
The honest position is the boring one.
Nobody yet knows how this lands, the effect so far is uneven, and the safest guide is what these companies actually do with their own workforces, not what their founders say from a stage. The bosses have changed their tune. The layoff notices have not.
Until the second catches up with the first, treat the optimism exactly as you should have treated the doom: as a forecast from people with something to sell.