Recap: The GenAI Job Impact Just Got Real
Stanford’s latest AI jobs study confirms what we’ve been seeing in the trenches.
A new working paper from Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen was released this week, bringing the clearest and most granular evidence yet that AI is quietly reshaping the labor market. Using millions of payroll records from ADP, the researchers tracked hiring trends across tens of thousands of U.S. firms. The punchline: AI isn’t killing jobs across the board. It’s targeting the first rungs.
This is not mass extinction or hype. It’s harder to see and therefore easier to ignore: a reclassification of value. The very thing we explored earlier this summer—that AI erodes junior roles while amplifying versatility—is now backed by real-time data.
Six Facts That Stand Out
Early-career roles are eroding: Workers aged 22–25 in AI-exposed jobs (think software engineers, customer service reps) have seen a 13% relative decline in employment since late 2022. Older workers in the same jobs are holding steady.
The problem isn’t the overall economy: Total employment is up. Unemployment is low. But for entry-level white-collar workers, especially in information technology, consulting, and related services and support, the ladder is cracking.
Automation, not augmentation, is the culprit: Roles where AI replaces tasks (e.g., directive or feedback-loop queries) saw the steepest declines. Where AI augments work - collaboration, learning, refinement - jobs held firm or grew.
The firm doesn’t matter, the exposure does: After controlling for firm-wide hiring trends, the AI effect persists. This is not about tech layoffs or macro conditions. It’s occupational and structural across all sectors.
Paychecks haven’t moved (yet): No major changes in compensation. Employers aren’t slashing wages. They’re just not hiring as many entry-level humans.
It’s not just a tech-sector story: Even in non-tech, non-remote jobs, the pattern holds. This is systemic across all types of jobs and professions.
If going to college is about the first job, what does this mean for higher education?
What This Confirms
In Jobs at Risk, Jobs That Rise, I argued:
The bottom rungs of the ladder are eroding.
Routine, junior roles are most at risk.
Interdisciplinary knowledge and adaptability are protective.
The real shift isn’t job loss, it’s role redistribution.
Institutions are still training students for disappearing jobs.
Every one of those claims now has empirical backing. The Stanford team stopped short of offering prescriptions. But the implications are unmistakable. This moment demands present-proofing: a deliberate pivot in how we train, hire, and lead. We must stop preparing people for jobs that no longer exist and instead build capacity for the kind of human value AI cannot easily replicate: range, synthesis, and leadership.
The final word
The future of work isn’t being determined by headlines or forecasts. It’s unfolding quietly, month by month, in the granular shifts of payroll data. The real story isn’t widespread unemployment; it’s the slow erosion of predictable entry points. Routine, junior, narrowly scoped roles are disappearing not with a bang, but with a data trail.
But there is a world where professionals who infuse AI into their work have a bright and shining future. Organizations need orchestrators, synthesizers, and integrators: people who can navigate ambiguity, work across systems, and bring judgment to complexity. AI isn’t coming for everyone; it comes for those lacking versatility.
Note: the authors of the Stanford paper will present their findings at an online meeting scheduled for September 29, 2025. Please visit this link if you would like to participate.