The AI Jobs Apocalypse Got Postponed — and the Data Says So

The scariest story about AI just hit a snag: the data

For a couple of years now, the default assumption about AI has been apocalyptic. The robots come, the jobs go, and half of us end up retraining as artisanal candlemakers. It's a great story. It's also, according to the earliest evidence, not the one that's actually happening.

The Economist puts it bluntly: the jobs apocalypse is postponed, and an AI jobs boom is here. The initial effects of the technology on employment, they report, look positive — not neutral, positive. That's a big deal, because almost the entire public conversation, especially in the US, has been built around fear of job loss. Noah Smith makes the same point when he surveys what Americans are actually worried about with AI: the national economic debate is 'mostly about fear of job loss.'

So who's right, the doomers or the data? For now, the data. And 'for now' is doing real work in that sentence, which we'll get to.

The jobs apocalypse is postponed — and 'postponed' is doing a lot of work in that sentence.

Why this matters if you have a job

The Economist's read is that the early effects look positive — AI is showing up alongside hiring, not mass layoffs. The plain-English takeaway for a worker: right now the evidence points to AI making people more productive rather than replacing them wholesale.

The honest caveat, and it's a big one: 'initial effects' means we're early. Early positive signals can curdle. The candlemaker scenario isn't disproven — it's just not what's on the scoreboard yet. So the useful move isn't panic, and it isn't complacency either. It's becoming the person who knows how to run the AI, rather than betting your career on it never touching your desk.

What to watch: whether the labor data stays positive as the tech matures. If the hiring numbers hold as AI gets more capable, the 'augmented, not replaced' story firms up. If they roll over, that's your early warning.

The bigger picture: it's a race, and America's still winning it

Zoom out and the more consequential story isn't 'robots vs. your job.' It's the US versus China. Smith argues America is still beating China in the AI race: US companies have better models, more compute, and far more revenue.

His sharpest point is about how the scoreboard gets misread. When a Chinese model makes headlines — Z.ai's GLM-5.3 reportedly edged Anthropic's Mythos 5 on one narrow benchmark for spotting software flaws — people treat it as China catching up. But Smith notes the released American models aren't even America's best. Anthropic says it has stronger internal models it held back over safety concerns, and on most other cyber measures Mythos still ran well ahead. In plain terms: China is chasing what the US already shipped, not what the US actually has.

Why should you care? Smith flags one real risk to the US lead that has nothing to do with chips: pushing Chinese AI talent out of the country. Talent policy, he argues, could do more damage to America's edge than any single Chinese model release. That's the thing to watch — not just who ships the flashiest model, but whether the US keeps the people building them.