For years, the debate about AI and jobs was mostly speculation: forecasts, exposure scores, and arguments about which professions would be “replaced.” The evidence is now starting to come in, and so far it points somewhere specific. It isn’t experienced professionals who are feeling it first. It’s people trying to get their first job.
What the data shows
The most rigorous evidence so far comes from economists at Stanford’s Digital Economy Lab. In a paper titled “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence,” Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen use payroll data from ADP, one of the largest payroll processors in the US, covering millions of American workers.
In the latest revision of the paper, updated in August 2026 with data through June 2026, the headline finding is stark: employment of workers aged 22 to 25 in AI-exposed occupations is now 19% below where it would be if it had kept pace with their peers in less-exposed jobs.
Two details make the finding more persuasive than a typical headline statistic.
It’s specific to young workers. More experienced workers in the same occupations haven’t seen the same decline. If AI were simply shrinking whole professions, you’d expect everyone in them to be affected.
It depends on how AI is used. The declines are concentrated in occupations where AI mostly substitutes for human tasks. Where AI mainly complements workers, helping them do more rather than doing the work for them, employment has been flat or rising, especially for experienced workers.
This is one study built on one company’s payroll data, so it’s a strong signal rather than a final verdict. But it’s the clearest evidence we have so far, and it’s consistent with news reporting that AI’s effect on jobs is showing up mostly as weaker hiring for junior roles rather than as layoffs.
Why the bottom rung goes first
When you think about what entry-level work actually consists of, the pattern makes sense.
Junior work is the most automatable work
A lot of first jobs are made of well-defined tasks: summarizing documents, drafting standard emails, cleaning data, writing basic code, answering routine questions. These are exactly the tasks today’s AI models handle well. Senior work tends to involve judgment, relationships, accountability, and context that’s hard to write down, all of which current AI handles much less well.
Experience is what makes AI useful
Here’s the twist I find most important. AI makes experienced people more productive partly because they can tell when the AI is wrong. A senior engineer can review AI-generated code in seconds. A junior engineer may not spot the bug. So AI tends to complement experience and substitute for inexperience, which is exactly the split the Stanford data shows.
It shows up as fewer hires, not layoffs
Most of this isn’t dramatic. Companies don’t need to fire anyone. They just hire one junior analyst instead of three, or none at all, and let AI tools absorb the work. That’s why the effect is easy to miss in headline unemployment numbers, and why it shows up most clearly among young people entering the workforce.
The long-term problem nobody wants to own
If this continues, it creates a pipeline problem that I think employers are underestimating.
Today’s senior professionals learned their craft by doing junior work for years. The tedious tasks were also the training. If companies stop hiring juniors because AI can do the junior tasks, where do the next decade’s senior people come from?
Each company acting alone has a reason to hire fewer juniors. Collectively, an industry that does that is eating its own future. That’s a classic coordination problem, and I don’t expect it to resolve itself.
What this means if you’re early in your career
None of this means entry-level paths are closed. It means the bar has moved. If I were starting out today, I’d focus on a few things:
- Get very good at using AI, and at checking it. The useful junior employee in 2026 isn’t the one who avoids AI. It’s the one who uses it to work faster and can catch its mistakes. Our guide on how to fact-check an AI answer is a good start.
- Aim for the complementary side. Look for roles that involve judgment, customers, coordination, or responsibility, where AI helps you rather than replaces the task.
- Show proof, not just credentials. Real projects, shipped work, and examples of problems you solved carry more weight when routine tasks are cheap to automate.
- Learn how agents work. As AI agents take on more multi-step work, people who understand how to set them up, supervise them, and fix them will be in demand. Our explainer on what an AI agent is covers the basics.
What this means if you’re hiring
As someone who runs a company, here’s how I think about it.
AI changes what a junior role should be, not whether you need one. The best version of an entry-level job now looks less like “do the repetitive tasks” and more like “supervise the AI doing the repetitive tasks, handle the exceptions, and learn the judgment calls.” That’s a better job, and it builds the senior people you’ll need in five years.
It’s also worth being honest about the trade-off. Cutting junior hiring saves money this year. It costs you your future managers and specialists. Companies that keep investing in early-career talent, and redesign those roles around AI, will have a real advantage when everyone else runs short of experienced people. It’s the same discipline I argue for in getting AI agents past the pilot stage: treat AI as an operational change with owners, not a cost-cutting switch.
My take
The early evidence says AI’s first real labor-market effect is narrowing the entrance, not emptying the building. That’s less dramatic than mass layoffs, and in some ways more worrying, because it’s quiet and it compounds. The fix isn’t to slow down AI. It’s for employers, schools, and young workers to redesign the first rung of the ladder on purpose, before it disappears by default.
Sources
- Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, Stanford Digital Economy Lab, Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence, revised August 12, 2026
- CBS News, AI layoffs and hiring for entry-level workers
