A UCLA study finds no (yet) collapse in youth employment feared from AI

By The Washington Post · September 28, 2026.

Two professors at the University of California, Los Angeles (UCLA) have published a working paper —an academic study not yet peer-reviewed— that, according to the newspaper itself, is one of the first serious attempts to measure the effect of artificial intelligence on hiring in the university class of 2026. Their conclusion, for now: there is no rise in unemployment among recent graduates that can be attributed to AI, despite fears that the technology would devastate entry-level hiring.

It has to be said honestly: the available material on this paper is very limited. It includes no concrete figures, no sample size, no author names and no detail on the exact method used to isolate AI's effect from the rest of the labor-market variables. With that caveat up front, the finding —the absence of a visible shock in recent-graduate unemployment— deserves to be taken seriously, but also kept in perspective.

Broadly speaking, the fear that AI would hollow out the first rung of the professional ladder has been part of the US labor debate for more than a year now: the logic is that much of the work done today by junior analysts, newly graduated programmers or administrative assistants consists of routine, well-defined tasks, precisely the kind of task where today's language models perform best. That this devastation has not yet shown up in the statistics does not mean the pressure is absent: it may be showing up more quietly, as a slowdown in hiring or a raising of the entry bar, rather than as mass, visible layoffs.

Our reading is that this data point fits what we have been observing sector by sector: AI's impact on employment does not arrive as a sudden, uniform collapse, but as a gradual, uneven recomposition that first erodes the most routine tasks within each profession before touching total hiring volume. For the class of 2026, the absence of an unemployment spike is good news in the short term, but it does not settle the underlying question: whether companies will keep needing as many entry-level positions as before, or whether the career ladder is turning into something different, with fewer initial rungs and more demanded from day one. That is precisely the hard transition we acknowledge bluntly when we argue that the abundance AI may bring in the long run will not arrive without friction in the short term.