AI Is Eating the First Rung of Every Ladder

The entry-level job — the rung where a generation learned the work by doing the small work — has become the exact job the new tools were built to take, and a ladder with no first rung still gets climbed, just only by people already on it.

I grew up working the counter at my dad's auto-parts store, and the counter ran on a simple order of operations. Nobody starts as the person who diagnoses the engine noise. A new hand starts fetching parts, sweeping the floor, and watching the old hands work, and the fetching counts as the education — a kid learns where every part lives, which jobs eat an afternoon, and how a pro talks a worried driver down. I think about that counter every time a company calls a first job a cost line.

Every earlier technology wave heard the same warning, and the jobs changed shape instead of vanishing

ATMs were going to end bank tellers, and teller numbers held for decades while the job changed underneath the title — branches got cheaper to run, banks opened more of them, and tellers sold and advised instead of counting bills. Spreadsheets were going to end accountants, and accounting grew, because cheaper number-crunching made analysis affordable for firms that had never bought any. The executives selling AI make a version of the same argument today: a worker using AI beats a worker replaced by AI, and the smart shop buys leverage for its people rather than a pink-slip machine.

The Stanford researchers behind the scariest number in this debate print their own caveats in bold type. Their finding counts as an early signal inside specific exposed jobs, and overall youth employment outside those jobs kept growing, with no widespread displacement across the economy. And the cyclical case deserves full weight, because the cycle did real damage first: the post-pandemic hiring binge and the interest-rate climb afterward cut tech's junior hiring hard before any chatbot entered the building.

The companies selling AI pitch the tools on doing the exact work junior staff used to learn from

Grant the premise first: the tools themselves are real, and they're good at the work. The companies selling AI to employers pitch the tools on doing the work junior staff used to do — first drafts, boilerplate code, basic research, ticket triage. The sales pitch counts as the agreement. Employers nod along because those tasks are what a first job was for. A first draft teaches judgment about drafts. Boilerplate teaches a codebase. Triage teaches what breaks and why. The chores were reps in disguise. Every senior worker in every field was once a junior collecting those reps, and nobody selling or buying the tools has described where tomorrow's seniors collect theirs. A skill nobody gets to practice grows scarce rather than cheap, and scarce senior judgment commands a fortune precisely because the pipeline ran dry.

You can automate the apprenticeship or keep the craftsman pipeline. The arithmetic won't let a company, or a country, keep both.

Payroll records show workers aged 22 to 25 in the most AI-exposed jobs standing 19 percent behind their less-exposed peers

Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen at the Stanford Digital Economy Lab tracked ADP payroll records covering millions of American workers, and their paper, "Canaries in the Coal Mine?", puts employment for workers aged 22 to 25 in highly AI-exposed occupations 19 percent below where employment would stand had that group kept pace with same-age peers in less-exposed jobs, on data running through June 2026. Experienced workers in the same exposed jobs show no comparable gap, and the authors describe the pattern as an early, descriptive signal rather than proof of cause, running mainly through slower hiring rather than waves of firings.

Scale checks agree on direction. SignalFire, a venture firm, reports in its State of Tech Talent 2025 that new graduates account for just 7 percent of hires at Big Tech — the top 15 tech firms by market value — with new-grad hiring down 25 percent from 2023 and down over 50 percent from 2019. Challenger, Gray & Christmas counted AI cited as a reason in 48,414 announced job cuts through October 2025, with 31,039 in October alone.

Rising interest rates and the post-pandemic hiring binge hurt junior hiring before AI took any blame

The hangover. Interest rates climbed, the post-pandemic hiring binge ended, and companies that had staffed up for a boom started cutting back — and junior hiring took the early damage, before AI took any job at scale. Tech firms hired classes of graduates they lacked work for, then froze the door when money stopped coming free. Plenty of cuts now wearing an AI label trace back to that rate cycle and that binge. AI collects blame for damage the hangover ordered first, and my side should say so plainly.

My grading instrument stays the country test, and the test asks two questions of every hiring pitch and every tool. Does the arrangement work for the country, beyond one quarter and one company's ledger? And can a 26-year-old get on the ladder under it, collect the reps, and become the senior worker the whole system will need in ten years? A pitch that fails either question fails at my table, whoever sells it.