When AI replaces, not augments: a data-driven look at youth employment trends

I’ve been saying for years that AI will improve how we work. At this point in the process however, the data suggests that among the first casualties will be those entering the career market for the first time.

“Canaries in the Coal Mine? Six facts about the recent employment effects of artificial intelligence”, a recent study by Erik Brynjolfsson (whose work I have followed for more than three decades), along with Bharat Chandar and Ruyu Chen based on analysis of monthly payroll microdata from ADP (the largest payroll service provider in the United States), documents an uncomfortable fact: in the occupations most exposed to AI, the employment of people aged between 22 and 25 has fallen significantly since the spread of generative models, even taking into account variables at the company level. As Azeem Azhar points out, it is possibly one of the most important papers in our industry in recent years.

In aggregate terms, the central result shows a relative drop of 13% for the youngest in the most exposed positions. In addition, between the end of 2022 and July 2025, young people aged 22 to 25 in the two quintiles with the highest exposure suffered a decrease of 6%, while those aged 35 to 49 grew by more than 9%. The adjustments occur mainly through employment, not through wages, which…