Something has broken in the way careers begin. For decades, college graduates in their early twenties enjoyed a lower unemployment rate than the workforce as a whole. Since late 2018, according to the New York Federal Reserve, that has flipped: young graduates now often face higher joblessness than everyone else. A June report from the World Economic Forum and PwC put a sharper number on it. Employment for US workers aged 22 to 25 in the roles most exposed to AI has fallen 16 percent since late 2022. The bottom rung of the career ladder is being sawn off. The fight is over who is holding the saw.
The obvious suspect is artificial intelligence. Much of the work traditionally handed to juniors, drafting documents, gathering research, cleaning up data, is exactly the kind of task a capable model now does in seconds. Mark Ma of the University of Pittsburgh found that companies advertising AI in their job titles or descriptions cut overall hiring, with the steepest drops in early-career roles. "They are getting rid of these junior positions," he said, "because the junior-level work could be more easily replaced by AI." It is a tidy, intuitive story, and it may well be partly true.
But a competing explanation has been gaining ground, and it points somewhere less expected: remote work. Peter John Lambert of the London School of Economics and a colleague examined hiring data across several countries before and after both ChatGPT and the pandemic-era shift to working from home. Their conclusion was that the entry-level share of new hires began falling before generative AI tools were good enough to matter. The mechanism is mundane. Junior employees learn by watching, by absorbing the informal cues of an office, and remote setups quietly starve them of that. New York Fed researchers studying a Fortune 500 company found feedback "tapers off dramatically" when colleagues are separated, and that the loss hits younger workers hardest.
The two theories are genuinely hard to untangle, because AI and remote work haunt the same jobs: desk roles that can be done, or automated, from anywhere. Stanford's Nicholas Bloom, who has studied remote work for years, is refreshingly honest about it. Remote work, AI, pandemic learning losses, and a broad tech hiring slowdown are all plausible causes, he says, and current data simply cannot tell us which is doing the damage or by how much. That uncertainty is not academic hair-splitting. If AI is the cause, the pipeline problem is structural and permanent. If remote work is the cause, it is a management choice that companies like Revolut, now requiring its interns back in the office three days a week, can reverse.
What almost no one disputes is the danger of getting it wrong. Entry-level jobs are not just employment; they are, as one WEF contributor put it, an "engine of renewal," the training ground where future managers and specialists are actually made. Automate them away without building a replacement, and firms may win a short-term efficiency while quietly draining the talent they will need in a decade. The report's warning is blunt: the question is no longer whether early-career work will change, but whether companies can still produce tomorrow's leaders once they stop hiring today's beginners.