Key takeaways
In August 2025, Stanford economists looked at ADP payroll records covering millions of workers. They found that employees aged 22 to 25 in the most AI-exposed jobs had seen a 13 percent relative fall in employment since late 2022. Older workers in the same roles held steady or grew by 6 to 9 percent.
In software engineering and customer service, entry-level employment fell by about 20 percent.
The public debate runs on the fear of one great wave of layoffs. The data shows something narrower. Jobs at the bottom of the ladder quietly never get posted.
Why the bottom rung?
The effect held after controlling for conditions inside each firm. It showed up in headcount, not pay.
Large models are trained on exactly the written-down, book-learned knowledge a new graduate brings. So they compete most directly with the least experienced. Senior people are protected by hard-won judgement that isn’t written down anywhere.
Bank of America has noted that unemployment among recent graduates has begun to exceed the overall rate.
| Category | Entry-level employment fall (22-25) |
|---|---|
| Exposed roles | 13% |
| Software & CS | 20% |
Cite or embed this
AI is also running the hiring
AI is thinning out entry-level roles. It also decides who gets the ones that are left. Résumé screeners, candidate rankers and interview analysers sit between applicants and recruiters.
New York City regulated them first. Local Law 144 requires an annual independent bias audit, a published summary, and advance notice to candidates with a way to opt out.
In late 2025 the city’s own Comptroller audited the enforcement and found it ineffective. Complaints were misrouted, and posted audits got only a quick look.
The pressure that actually bites comes from elsewhere:
Put together, the first job gets harder to get, because AI screens for it. And it gets rarer, because AI does the work it used to contain.
There’s an upside too. Where AI helps with the work, employment grew. Starting salaries for entry-level AI roles rose about 12 percent from 2024 to 2025. And 23 percent of employees already use generative AI every week.
The sales development rep, the classic first sales job, is already being rebuilt around using AI.
Running a team? Notice that the training pipeline is thinning at the bottom. The entry-level work that used to teach judgement is now automated. So decide on purpose how a new hire will gain the judgement AI can’t yet fake.
Frequently asked questions
Is AI causing mass layoffs?
So far the evidence points to a focused hit, not a broad wave. A Stanford study found a 13 percent relative fall in employment for workers aged 22 to 25 in the most AI-exposed jobs since late 2022. Older workers in the same roles were stable or grew. The effect shows up more in hiring and headcount than in firings or pay.
Which jobs are most affected?
Entry-level roles in fields like software engineering and customer service, where early-career employment fell by about 20 percent. These jobs lean on written-down knowledge that overlaps with what large language models do well.
Is it legal to use AI to screen job candidates?
It's regulated, not banned. New York City's Local Law 144 requires annual bias audits, public summaries and candidate notice. The EU AI Act treats employment decisions as high-risk. US anti-discrimination law still applies. Enforcement varies, and New York City's own audit found its law was being enforced ineffectively.
Is there any upside for workers?
Yes. Employment grew in roles where AI helps with the work instead of replacing it. Workers who use the tools are gaining. Starting pay for entry-level AI roles rose about 12 percent from 2024 to 2025. The advantage is moving toward people who can use AI well.
Three-minute lesson on this What AI actually is Marketing course, basic. Free, no sign-up.