Here is the uncomfortable truth about AI and work: the layoffs everyone keeps predicting aren't the story. The story is the job that never got posted.
Since ChatGPT launched in late 2022, the documented damage from AI has landed almost entirely on people trying to start careers, not people already in them. That's the finding that should reorganize how we talk about this.
AI's Subtle Shift: Beyond Mass Layoffs
I keep hearing the same question: when will AI trigger mass layoffs? When does the axe fall?
Wrong question. As Yale's Jeffrey Sonnenfeld argues, the AI jobs debate isn't exaggerated—it's framed incorrectly. We keep expecting disruption to show up all at once, in visible waves of pink slips. But that is not how this transition unfolds.
The evidence points somewhere quieter. A widely discussed study by Erik Brynjolfsson, Bharat Chandar, and Chenyu Chen, using payroll records, found a notable decline in employment among early-career workers in AI-exposed occupations since ChatGPT's release in 2022. Meanwhile, employment among older workers in those same occupations remained relatively stable or kept growing.
The authors call these young workers "canaries in the coal mine"—the first to feel labor market disruption from AI. Other researchers have since found similar negative effects on hiring of young AI-exposed workers in both the U.S. and the U.K., beginning in 2022.
So the disruption is real. It's just concentrated at the entrance, not the exit.
The Quiet Mechanism: Reduced Entry Points, Not Exits
How does this actually happen inside firms? Not with dramatic restructuring announcements.
It happens through hiring decisions that never make headlines. A team absorbs the work of a departing junior employee with AI tooling instead of backfilling the role. A manager reclassifies routine entry-level workflows as tasks delegated to AI systems. The headcount quietly shrinks at the bottom of the pyramid while the top stays intact.
Sonnenfeld's analysis is blunt on this point: as agentic AI scales across industries, the changes are already happening—"just quietly." The key question, he writes, is no longer which jobs disappear, but which tasks and workflows are being delegated, and where humans still retain a comparative advantage.
That reframing matters. A layoff is a visible event with a name, a date, and a severance package. A hiring reduction is an absence. Nobody protests a job that was never created.
And the mechanism compounds. Sonnenfeld notes that agentic AI will likely reduce labor demand "not through sudden disruption but through a steady narrowing of opportunities at the entry level." The biggest impact on jobs, in his words, "will not be the layoffs we can see."
