It used to be pretty easy to figure out how to get a white-collar job. After high school, you got a job that wasn’t very exciting, like writing routine legal memos, coding small features, or collecting research for someone three levels above you. You learned by doing. That’s always been the case. To get where you are now, you had to do the boring work at the bottom. That bottom is becoming less and less accessible.
It’s not as easy to pin down what’s happening with entry-level jobs right now as the headline makes it seem. Some of it is worry about the economy—companies aren’t hiring as many people as they did two years ago, and the uncertainty surrounding tariff policy isn’t making anyone feel good about growth. This year, the unemployment rate for college graduates hit 5.8%, which was the highest level since 2021. The rate of underemployment rose above 40%. Economists are quick to point out that not all of that is due to AI, but those numbers are still scary. Some of it is just what a cooling economy does when it cools down.
But there’s more going on than meets the eye, and it would be wrong to blame trade policy and market jitters for the whole trend. Finding legal precedents, writing baseline code, sorting through data, and writing first drafts used to be the tasks that defined entry-level work. These are the exact tasks that AI is getting better at. For a generation of college graduates who thought they would learn on the job, the job may have been changed in some ways before they got there.
Last month, Dario Amodei, CEO of Anthropic, said that AI could eliminate half of entry-level jobs in the United States in just five years. That shape is very interesting, and it’s worth taking a seat with it. Even if it turns out to be closer to a third or a quarter, the disruption would still be big. A law firm that used to need four junior associates to prepare a case might only need two now, along with a subscription to Claude. The math isn’t vague. It’s starting to show up in how people are hired.
A professor at the University of Pennsylvania named Lynn Wu, who studies work and technology, said, “The biggest problem is that the career ladder is being broken.” She’s not wrong, and the way she said it—not as a guess but as an observation—makes it sound like this is no longer just a theory. The whole ladder isn’t broken, but the first rung is shaking in ways it wasn’t a few years ago.
It’s strange that young people entering the job market right now are probably the smartest people in history when it comes to AI. They learned to use these tools as kids. But that familiarity hasn’t yet given them a clear advantage in a market that’s still trying to figure out what it wants from new employees. Expectations change more quickly than job descriptions. AI skills are something that companies want, but the exact role is still being shaped in real time.
There are good reasons to be cautiously optimistic, but you have to be patient. MIT researchers who looked into changes in the workforce found that technology doesn’t tend to get rid of the need for human judgment; instead, it makes it more important. Roles that are based on empathy and presence, like early childhood teachers, health aides, and therapists, have stayed the same or even grown. And most analysts think that entry-level jobs won’t go away completely, but will change shape. Soon, the person who used to spend eight hours putting together case files might only have to do that for two hours and have to spend six hours on something that needs better judgment.
How long you’ve been waiting for a callback will probably affect whether that change feels like a chance or a burden.
