This is an explainer; it does not predict where the labor market goes next.

Is artificial intelligence taking people's jobs? The honest answer in mid-2026 is: the data points both ways, and it's getting harder to tell. Plenty of companies are cutting roles and citing AI; young workers are finding entry-level jobs scarce. Yet headline unemployment remains low, and some firms leaning hardest into AI are adding staff. Here's why the picture is so muddy — and why it matters for workers, employers and investors.

The case that AI is hitting jobs

The most striking evidence is at the entry level. A widely cited analysis by Harvard economists, drawing on tens of millions of résumés, found that entry-level hiring at companies adopting generative AI has dropped sharply since 2023 — by some measures around 80% — concentrated in tech and knowledge work, as Forbes reported. Young workers feel it: recent graduates have faced higher unemployment than the workforce as a whole, CBS News noted, unusual for a group that normally has an easier time finding work.

Employers are also naming AI directly in restructuring announcements, where they once would have said "cost-cutting." Tens of thousands of job cuts this year have been attributed to AI, by some tallies — a language shift that has stoked fears the displacement is already underway.

The case for skepticism

But step back to the whole economy and the signal fades. Researchers at the AI firm Anthropic found no detectable rise in aggregate unemployment for the workers most exposed to AI since ChatGPT launched in late 2022, in their own labor-market study. If AI were broadly destroying jobs, joblessness among exposed workers should be climbing measurably. So far, it isn't.

And some of the heaviest AI adopters are hiring. One analysis cited by TechCrunch found firms using AI most intensively actually grew headcount — though, crucially, those tend to be well-funded, fast-growing tech companies already inclined to expand, so it's unclear whether AI causes the hiring or just comes along with it. Government statisticians, for their part, treat AI like past technologies: a force that reshapes work over years and decades, not quarters.

Why the data is so hard to read

The core problem is attribution. When a company trims junior roles, how much is AI, versus higher interest rates squeezing budgets, versus a hangover from the over-hiring of 2021–22, versus offshoring that was already happening? The numbers can't easily separate these.

Companies also have an incentive to credit AI — it sounds forward-looking, while "we hired too many people" or "demand is soft" does not. So some "AI layoffs" may be ordinary cost-cutting in a new costume.

And averages hide concentration. AI could be barely denting total employment while still choking off the entry-level pipeline in specific high-skill fields — leaving young people unable to get a first rung even as the overall jobless rate looks fine. That's a serious problem for individuals that wouldn't show up as a headline crisis.

What it means

The uncertainty itself has consequences. Firms that can afford to deploy AI and retrain staff may pull ahead of those that can't. Workers in exposed roles — junior engineers, analysts, support staff — face pressure to upskill and adapt faster, while those in hands-on or less-exposed jobs may see little change for now.

Economists stay split: some, like Apollo's chief economist, say there's "zero evidence" AI is killing jobs and bet it will create them; others read the soft entry-level data as an early warning. Boursel takes no side and offers no advice. The honest takeaway is that AI's effect on jobs is real, uneven and still unfolding — and far messier than either "AI is a job killer" or "AI creates more than it destroys" lets on. Anyone selling you certainty, in either direction, is ahead of the evidence.