Every few weeks a headline announces that AI has come for the graduate job. The claim grips people because it names a specific, sympathetic victim, the twenty two year old with a fresh degree and no way in. It is worth slowing down to ask what the evidence actually shows, because this is one of those fears that is neither a hoax nor the catastrophe the loudest version claims.
Start with what looks real. A Stanford Digital Economy Lab study found roughly a 13 percent relative decline in employment for workers aged 22 to 25 in the occupations most exposed to generative AI, measured against older workers in the same fields. In software specifically, employment for developers aged 22 to 25 has fallen close to 20 percent since 2024, while developers over 30 kept gaining jobs. Those are not small effects, and they point the same way across more than one dataset.
What the numbers do not say
Here is where the sweeping version overreaches. The data shows AI suppressing new hiring far more than it shows AI firing people who already have jobs. Economists at Goldman Sachs describe employers using these tools to avoid adding headcount rather than to cut existing staff. That is a genuine harm to people trying to break into a field, but it is a different thing from a wave of layoffs, and the two keep getting blended together in the retelling.
There is also a timing problem the panic skips over. Hiring for junior roles slowed during a stretch of high interest rates and post pandemic over hiring, both of which would have cooled entry level demand with or without AI. Separating the machine's effect from the wider economy is genuinely hard, and any honest account has to admit the signal and the noise are tangled together.
The longer arithmetic
Zoom out and the picture gets murkier still. The World Economic Forum's most recent jobs outlook projects 92 million roles displaced by 2030 against 170 million created, a net gain on paper. Forecasts that far out deserve heavy skepticism, and a net positive is cold comfort to someone whose particular entry ramp just vanished. Still, it is a useful corrective to the idea that the job count only ever moves in one direction.
So where does that leave it
The fear is not baseless, which is what makes it stubborn. Young workers in exposed fields really are having a harder time getting hired, and pretending otherwise helps no one. What the evidence does not support is the leap from that to a general collapse of entry level work. The measured version is troubling enough on its own, and it is the one worth arguing about. For a related look at how firms are actually deploying these tools, see our piece on Microsoft telling its own engineers to ease off the AI.
Commentarii · 0