Every June, we stage a ritual that borders on the sacred. Across the country caps fly, cameras flash, and degrees are handed out with an implicit promise: work hard, get educated and the economy will meet you halfway.
That promise is starting to crack. Nearly 43% of recent college graduates are underemployed, working in jobs that don’t require their degree and are facing the worst entry-level job market since the pandemic. Unemployment among new grads now exceeds the national average. Employers are quietly redefining “entry-level” to mean two or three years of experience for roles that once provided it.
The ladder is still there. But this June, it’s shorter than it once was.
The research is clear that entry-level jobs are shrinking fastest in sectors built on routine cognitive work, and holding steady or growing in sectors built on physical, interpersonal or in-person tasks. This may seem obvious but the implications are significant.
Entry-level hiring has fallen sharply in technology, finance and professional services — the very sectors college graduates have been told to pursue. Job postings for entry-level roles have dropped by as much as 30%–35% in recent years, with particularly steep declines in software development, data analysis and administrative support. In AI-exposed fields like coding and customer service, junior job listings have fallen by 13% in just a few years.
These roles are built on tasks AI can now perform: writing code, generating reports, analyzing data, answering routine questions. Companies still need the work done. They just no longer need humans to do it.
The deeper problem is that entry-level jobs were never just jobs. They were a training system. A junior analyst didn’t just produce models — she learned how to think. A junior developer didn’t just write code — he learned how to debug, structure and build. These roles created value, but they also built capability. AI is extraordinarily good at the first function. It is terrible at the second.
A system that produces a working model in seconds is a productivity breakthrough and an invisible tax on the labor market that would have trained the worker who used to build it. From the outside, this looks like efficiency. Inside the economy, it’s more like trading tomorrow’s workforce for today’s productivity gains.
There is an inherent risk firms are not pricing in. By pulling back on entry-level hiring today, they are reducing the supply of experienced workers they will need tomorrow. The mid-career talent pipeline doesn’t build itself. And for the employee, the waiting compounds. Early career earnings shape lifetime earnings. Skills build on skills. Networks form early. When the first step comes at 26 instead of 22, the damage shows up later in wage gaps and stalled mobility.
There are early signs of what a response could look like. At University of California San Diego, recent convenings have brought employers, educators and workforce leaders together around a shared challenge: rebuilding the pathway from education to employment in an AI-driven economy. The solutions are practical — paid, project-based internships, apprenticeship-style rotations across firms, employer-led training cohorts, and tighter alignment between curriculum and the tools used on the job. None fully replace the traditional entry-level role. But together, they begin to recreate what we’ve lost: a structured pathway for people to build capability before the market assumes it.
Taking this seriously means treating the transition from education to work as infrastructure. If the market is producing fewer entry-level opportunities, we need to build new ones through apprenticeships, employer partnerships and targeted incentives for firms to invest in early-career talent. It also means rethinking education itself: not just training students to perform tasks AI can execute, but preparing them to work alongside these systems, directing them, evaluating them and knowing when they are wrong.
The caps will still fly this June. The question is whether there will be something solid for graduates to land on.
Shapiro, Ph.D., is an economic sociologist focusing on the future of work and social and economic mobility at UC San Diego. He lives in San Diego.