The first rung did not disappear. It moved.

At the end of June 2026, Fortune checked back in with the Stanford economists behind the most cited labor study of the AI era, and the message was blunt: the effect is not going away. The study, built on payroll data from millions of workers, found that early-career workers aged 22 to 25 in the most AI-exposed occupations have seen a 13 percent relative decline in employment since generative AI spread, concentrated in occupations where AI automates work rather than augments it.

Here is the twist: entry-level hiring overall is not collapsing. NACE's spring update projects hiring for the class of 2026 up 5.6 percent, revised upward from a flat 1.6 percent in the fall. Employers still want young people. What is thinning is a specific kind of young job: execute codified tasks under supervision, the classic first rung.

What the first rung was actually made of

The old deal was simple. You arrived with book knowledge, spent two years doing supervised execution (drafts, reconciliations, tickets, research memos), and absorbed judgment by osmosis. The Stanford researchers point out why that rung is exactly where AI bites first: models are strongest at codified knowledge, the very thing formal education certifies. The execution reps that used to season a junior are now the cheapest thing in the building.

The market is already pricing this in. Graduates with internship or co-op experience were hired within three months at 81.6 percent; those without, at 40.7 percent. Experience has become the entry currency precisely because it is evidence of something AI does not certify: judgment exercised on real work with real stakes.

Rebuild the rung on direction, not execution

For employers, the answer is not to stop hiring juniors; it is to redesign what a junior does. Give them direction reps early: own a small outcome end to end, run the agents that do the execution, check the outputs, answer for the result. That was the senior's apprenticeship model; it now has to start on day one. It is also, not coincidentally, how you grow the people who can manage the agent roster you are deploying anyway.

For graduates, the portfolio beats the tool list. Nobody serious hires for prompt tricks. Show small, finished, directed outcomes: a process you automated and supervised, a piece of work where machines did the labor and you did the judgment. That is the blank collar entry path, and it is teachable; it is the reason we made the curriculum at blankcollar.university free to anyone starting out.

Ladders do not vanish. They get rebuilt by the people willing to climb differently, and by the employers honest enough to admit the old bottom rung is gone.