Employment : AI Already Slowing Youth Hiring

Employment : AI Already Slowing Youth Hiring
3 min read

The hiring of 21- to 29-year-olds on permanent contracts is declining more sharply in professions exposed to artificial intelligence, according to a study by the Banque de France published in its Bloc-notes Éco. Since the launch of ChatGPT in late 2022, each new generative AI model has revived fears of massive job losses. Nearly four years later, data is beginning to provide a first glimpse, and the first visible effect in France appears to be hitting the youngest.

To reach this conclusion, five economists analyzed all hires and terminations of permanent contracts reported between January 2019 and March 2026. They cross-referenced this data with an index measuring each profession’s exposure to AI, and then with the responses of around 6,000 companies to the Banque de France’s business survey.

Young People First

In March 2026, permanent contract hiring of 21-29 year-olds in the fifth of professions most exposed is about 28% below its November 2022 level. In the least exposed fifth, the decline is limited to about 8%. The gap narrows among 30-39 year-olds and 40-49 year-olds, then almost disappears from age 50.

Departures are also decreasing, but to a lesser extent. Among 21-29 year-olds, terminations of permanent contracts are about 23% below their November 2022 level in the most exposed professions, compared to 6% in the least exposed. With hiring falling faster than departures, teams are renewing more slowly. In companies reporting intensive use of AI, permanent contract hiring of young people is also declining more than elsewhere, without their total employment diverging significantly from that of other companies.

Initial studies conducted abroad do not always reach the same conclusions. In Denmark, a rapid reorganization of tasks is mainly observed, with no detectable average effect on hours worked or income.

Developers on the Front Line

In their working paper, the authors look at the case of developers and computer engineers, who are heavily represented among the most exposed professions. Among 21-29 year-olds, their permanent contract hiring in early 2026 is about 42% below its November 2022 level, compared to about 11% in less exposed professions, a gap of nearly 31 points. Unlike the overall exposed professions, this decline is not limited to the young: it diminishes with age but affects all age groups.

In Norway, Hernæs and Kostøl note a comparable drop-off, with youth employment for developers down about 18%, while overall employment in the most exposed professions does not decline significantly among the young there. The authors acknowledge, however, that for this profession, they cannot distinguish the role of AI from that of the specific economic conditions in the IT sector, marked by the reorganizations that followed the pandemic.

An Effect Difficult to Isolate

The authors refrain from attributing these trends to AI alone. Several factors muddy the reading of their results. The economic context first: the ECB raised its deposit rate from −0.5% to 4% between July 2022 and September 2023, a tightening that, according to the study, may have weighed on hiring in the same professions, as did the post-Covid correction in the tech sector. Remote work, which largely concerns the same jobs, could also slow the recruitment of juniors. Finally, apprenticeships have increased: the number of apprentices rose from about 430,000 in 2017 to over one million by the end of 2023, which has changed the pathways to employment without alone explaining the differences between professions.

Companies that intensively use AI were already hiring, according to the study, in more exposed professions before the arrival of ChatGPT. Their use of AI is moreover only measured at the beginning of 2026, while recruitment gaps appear several quarters earlier. The study therefore establishes an association, without isolating the specific effect of AI. The slowdown in junior recruitment could nevertheless weigh in the longer term, as several recent theoretical works show that automating tasks assigned to beginners can reduce learning by doing and the accumulation of skills.