AI does not make jobs disappear overnight. Yet the first serious data shows one place where it is already being felt: hiring of beginners.
Three studies overlap: the Coface study on automatable tasks, the French Treasury note of June 30, 2026 and Stanford’s US payroll data.
What does the Coface study on AI and employment really measure?
Published on April 1, 2026 with the Observatory of threatened and emerging jobs, the study breaks each job down into tasks. When more than 30% of a position’s tasks can be automated by generative AI, Coface classes it in the transformation zone. Today, this concerns 3.8% of French jobs. In two to five years, if agents run entire workflows, the share would rise to 16.3%.
Several outlets drew “5 million jobs under threat” from it, sometimes read as “destroyed”. The report of the National Assembly’s social affairs committee points out that these studies measure the exposure of tasks to automation, not job cuts.
Why are 22-to-25-year-olds the first affected?
Stanford researchers tracked millions of US payslips. Since late 2022, employment of 22-to-25-year-olds in the jobs most exposed to AI has fallen by 13 to 16% compared with other workers, according to the study’s version. Older employees in the same jobs remain stable or are growing.
Trésor-Éco note no. 391 points the same way, with more caution: young people’s entry into the labour market “seems to be slowing”. The most commonly put-forward hypothesis is simple. The tasks given to beginners (summaries, first drafts, research) are also the ones AI can already do.
| Figure | What it measures |
|---|---|
| 3.8% | French jobs already weakened by generative AI (Coface/OEM, April 2026) |
| 16.3% | Share exposed in 2 to 5 years, if AI agents become widespread (Coface/OEM) |
| 13 to 16% | Relative fall in employment of 22-to-25-year-olds in exposed jobs, United States (Stanford) |
| 62% | French people who see AI as a risk to employment (2025 digital barometer, cited by the Treasury) |
What these figures do not prove
- The net effect of AI on employment remains uncertain, according to the Treasury, which also notes that AI is sometimes used as a pretext for layoffs decided for other reasons.
- What the data shows today is a slowdown in junior hiring, with no general rise in unemployment in exposed jobs.
- A University of Chicago study of 25,000 Danish workers in 11 occupations found no notable effect on wages or hours worked.
What to do when looking for a first job?
A junior who knows how to hand basic tasks to AI, then check the result, brings something other than execution. Recruiters are starting to say so: we detailed their view on degrees, and the tested prompts for rewriting your CV with AI. On the company side, the underlying question is raised by AI-related layoff announcements: who trains tomorrow’s seniors if no beginners are hired any more?
Our take
I don’t believe in mass destruction of jobs, I believe in the transformation of jobs. This transformation has an entry cost, and beginners are the first to pay it. A company that stops hiring juniors deprives itself of its future experts. Learning early to work with AI, under human control, remains in my view the best asset for a young graduate.
Sources
- French Treasury (Direction générale du Trésor), “Artificial intelligence: what effects on employment?” (primary source, June 30, 2026)
- National Assembly, report of the social affairs committee
- Le Monde Informatique, on the Coface/OEM study
- Fortune, on the Stanford research
- TechRadar, on the University of Chicago study
À découvrir aussi sur AI Sunday
Chaîne YouTubeL'IA au travail, en vidéoDernière vidéo : Notre place demain avec l'IA : on doit s'inquiéter ?Voir la chaîne Prof de langues IAParle anglais ou portugais avec un prof IABientôt : un prof vocal qui t'explique en français et te corrige.Rejoindre la liste 
