Remote work weakens an AI explanation for junior hiring losses

What the study measured

Lambert and Schindler compare 243 million resume-derived new hires and 407 million online vacancy postings across the U.S., U.K., Canada and Australia, 2017–2025. Their outcomes are the junior share of hires and the share of ads asking for no more than three years of experience. Neither is a count of AI-caused layoffs or unemployment.

In separate models, a two-standard-deviation rise in either GenAI or work-from-home exposure predicts roughly a five-percentage-point lower junior share of hires by 2025. In models that include both, the GenAI coefficient shrinks sharply and is often statistically indistinguishable from zero; remote-work exposure remains predictive. The occupation exposure measures are strongly correlated: 0.77 by rank. An additional comparison using remote or hybrid job ads in 2021–2022 also predicts lower junior shares later.

What changed in the answer

The study raises a concrete alternative to calling every exposed-job hiring decline an AI effect: remote work may alter supervision and junior recruitment in many of the same occupations. The remote-work association is not a randomized causal result, and an insignificant GenAI coefficient is not proof of zero AI effect.

French administrative records show a later gap in permanent hiring of 21–29-year-olds through March 2026; this four-country study does not include France or that period. One result cannot erase the other. Both leave the number of AI-caused lost jobs unresolved.

What would settle more of it

Follow comparable employers' actual AI use and remote-work practice together, then count junior and senior hires, departures and employment over time. Until then, ask whether a headline used an AI-exposure score or observed adoption—and whether it considered remote work.