New grads vs experienced professionals: AI is splitting the job market in two
Employment for 22 to 25 year olds in the most AI-exposed jobs fell a relative 16% while older workers in the same fields grew 6 to 9%. The ladder's bottom rungs are disappearing. Here's what the data says, and what both groups should do about it.
In short
AI is hitting new graduates and experienced professionals in opposite ways, splitting one job market into two.
- Stanford research finds employment for workers ages 22 to 25 in the most AI-exposed occupations fell a relative 16% since ChatGPT launched, with entry-level software roles down nearly 20%, while older workers in the same fields grew 6 to 9%.
- SignalFire reports big tech cut entry-level hiring 25% in a single year while hiring more senior people.
- The reason: AI automates the tasks juniors used to learn on, and amplifies the judgment seniors already have.
- The answer for both groups is the same skill: directing AI instead of competing with it.
One job market became two
For decades the deal was stable: graduates traded low wages for experience, doing the grunt work that taught them the craft. AI just bought the grunt work. Stanford researchers tracking payroll data found that since late 2022, employment for 22 to 25 year olds in the most AI-exposed occupations dropped sharply while workers over 30 in the same occupations grew. Same fields, opposite directions.
The numbers are blunt. Entry-level employment in AI-exposed jobs fell 6% from late 2022 through September 2025 while older workers gained 6 to 9%. For young software developers the decline approaches 20%. Recent-graduate unemployment hit 5.7% in late 2025, worse than during the 2008 financial crisis. And SignalFire found big tech cut new-grad hiring 25% in one year while increasing experienced hires. This is not a single report you can wave off: Stanford and ADP now refresh these figures monthly in a live canaries dashboard, and the early-career gap has held as new data lands.
I've hired thousands of people across my companies, and I've watched this shift from the hiring side. We didn't stop valuing juniors. The tasks we used to justify a junior seat, first-draft work, research summaries, basic analysis, simply stopped costing anything.
relative employment change, ages 22 to 25 in the most AI-exposed jobs since ChatGPT (Stanford)
big tech entry-level hiring, 2023 to 2024 (SignalFire)
recent-graduate unemployment, late 2025
employment growth for older workers in the same AI-exposed fields
Sources: Stanford Digital Economy Lab (Brynjolfsson et al.); SignalFire State of Talent; federal labor data.
AI eats tasks, not jobs, and juniors are made of tasks
The mechanism matters more than the headline. AI doesn't replace a person; it replaces tasks. Entry-level roles are built almost entirely from the tasks AI does best: drafting, summarizing, formatting, first-pass analysis. Senior roles are built from the things AI still can't do alone: judgment, context, accountability, knowing which output is wrong.
That's why the same technology that shrinks junior hiring makes experienced professionals more valuable. A veteran with AI fluency now produces what a small team used to. The World Economic Forum still projects 170 million new jobs created against 92 million displaced by 2030, but the new jobs assume you can direct AI, and 39% of core skills are changing underneath everyone.
Neither group gets to opt out. New grads can't out-grind a model at entry-level tasks, and experienced professionals can't coast on judgment while a younger, AI-fluent rival pairs judgment they're building fast with tools the veteran never learned.
New grads vs experienced professionals in the AI job market
Same technology, opposite effects, because the two cohorts are made of different tasks.
| New grads (22 to 25) | Experienced professionals (30+) | |
|---|---|---|
| Employment trend in AI-exposed jobs | Down a relative 16% since late 2022; entry-level software near -20% | Up 6 to 9% in the same fields |
| What AI does to their work | Automates the drafting, summarizing, and first-pass analysis they used to learn on | Amplifies the judgment, context, and accountability they already have |
| Hiring signal | Big tech cut new-grad hiring 25% in one year (SignalFire) | Same firms increased experienced hires |
| Main risk | Competing with a model at entry-level tasks it does for free | Coasting on judgment while an AI-fluent rival pairs both |
| The move | Walk in directing AI; ship portfolio work, not first drafts | Pair domain judgment with real AI fluency |
Sources: Stanford Digital Economy Lab (Brynjolfsson et al.); SignalFire State of Talent.
What each group should actually do
New grads: skip the ladder
The old bottom rung is gone, so don't compete for it. Walk in directing AI like a manager directs a team: ship portfolio work that proves you deliver senior-shaped output with AI doing the heavy lifting.
Experienced pros: arm your judgment
Your edge is knowing what good looks like. Pair it with real AI fluency and you become the person AI can't replace, and the one who decides how it gets used.
Managers: rebuild the rungs
If AI does the grunt work, apprenticeship needs a new shape. Give juniors AI-supervised real work with review, or your senior pipeline dries up in five years.
Everyone: prove it
Degrees signal less when AI compresses entry-level work. Verifiable, demonstrated AI skills, shipped work, not watched videos, are the new credential.
Common questions
Is AI really taking entry-level jobs?
In the most AI-exposed occupations, yes, measurably. Stanford research shows employment for workers 22 to 25 fell a relative 16% in those fields since late 2022, with entry-level software down nearly 20%, while older workers in the same occupations grew. Fields less exposed to AI show no such split.
Are experienced professionals safe from AI?
Safer, but far from safe. Judgment and accountability still command a premium, which is why older workers in AI-exposed fields grew 6 to 9%. But that premium goes to experienced people who can direct AI. Veterans who skip the skills hand their advantage to those who don't. Start with an honest read of where you stand.
What should new graduates do differently in the AI job market?
Stop competing for disappearing grunt work and walk in AI-fluent instead: learn to direct AI on real deliverables, build a portfolio of shipped work, and target roles where you operate like a junior manager of AI rather than a human first-drafter.
Whichever side of the split you're on, fluency is the move
Find your AI level in two minutes, then build the skills that put you on the right side of the data.
Sources
- Stanford Digital Economy Lab: Canaries in the Coal Mine? Brynjolfsson, Chandar & Chen, working paper (Nov 2025)
- Stanford Digital Economy Lab + ADP: Canaries Dashboard (monthly early-career employment tracker)
- SignalFire: State of Talent report (entry-level hiring down 25%)
- World Economic Forum: Future of Jobs Report 2025
Power users save 10+ hours a week. Learn how.
The practical AI habits behind it, one a week.
Written by
Adrián Ridner
Co-founder of Candova, founder of Study.com, and O'Reilly AI author
Adrián has spent two decades as a serial entrepreneur opening the doors to the life-changing impact of education. Before Candova, he founded and scaled Study.com into the largest platform for online college-credit courses, certification prep, and career-aligned degree pathways, helping millions of learners earn credentials for the modern workforce.