Stanford Study: The AI Job Gap for Young Workers Hits 19 Percent

Stanford economists have put a number on what recent graduates have been feeling, and the number is an AI job gap that keeps widening. In a revised study of American payroll records, researchers at the Stanford Digital Economy Lab report that employment for 22- to 25-year-olds in occupations most exposed to generative AI now sits about 19 percent below where it would have been if those young workers had kept pace with same-aged peers in less exposed fields. The analysis, released as an update to the lab's "Canaries in the Coal Mine?" paper in August 2026, uses high-frequency payroll data from ADP covering millions of United States workers.

Workers past their mid-twenties in the same occupations show no comparable shortfall, according to the lab's own summary of the findings. The result lands in the middle of a broader conversation about whether AI is shrinking the career ladder, and it gives that conversation the sharpest measure yet of the AI job gap.

The paper, by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, sets out six facts about how employment has evolved since ChatGPT's release. The headline is that there is no evidence of widespread, economy-wide job displacement associated with AI, which is what makes the AI job gap for the youngest workers stand out. The shortfall is specific: young workers in AI-exposed occupations are falling behind their less-exposed peers, while experienced workers show nothing comparable. The authors caution that these are descriptive patterns rather than causal estimates, noting that the timing and structure of the changes are suggestive but that the data alone cannot establish how much of the divergence was caused by generative AI as opposed to other forces in the labor market.

The shortfall is growing, and it runs through hiring

The AI job gap has widened steadily since the researchers first documented it. Using the same measure, the shortfall was 15 percent in the July 2025 data and had grown by the June 2026 data, the lab says.

Measured in absolute levels, employment of the youngest workers in the two most exposed quintiles fell roughly 11 percent between the launch of ChatGPT in November 2022 and June 2026, while employment of the same age group in the three least exposed quintiles rose about 10 percent. The researchers say the adjustment works mostly through companies hiring fewer young workers rather than laying off staff already in place, and that the declines concentrate in roles where AI substitutes for tasks a person would otherwise do, such as software development and customer service. Where AI mainly complements workers, employment is flat or rising, especially for experienced staff. The pattern echoes a widening generation-level jobs gap that economists have flagged across the economy.

Why experience seems to be protective

The revision adds a new distinction the authors say may explain why the AI job gap lands where it does. Employment has declined among young workers in occupations that lean on codified knowledge, the kind taught through textbooks, formal training, or written procedures. In contrast, employment has grown among experienced workers in occupations that rely more on tacit knowledge picked up through practice, mentorship, and repeated exposure to real situations. The lab's reasoning is that generative AI is unusually good at reproducing knowledge already encoded in text and digital information, while experience-based judgment remains harder to replicate. The paper also notes that women face greater AI exposure on average, a pattern the lab says it will keep tracking.

To keep the evidence current, the lab and ADP Research have launched a Canaries Dashboard that refreshes the headline figures every month, so the AI job gap will be measured continuously rather than in occasional snapshots.

What it means for someone planning a career now

For anyone mapping a career, the AI job gap is a signal about where hiring still flows. Young workers may be shifting toward roles that are harder to automate, and Fast Company reports the Stanford researchers see evidence of young people gravitating toward so-called AI-proof jobs that involve physical labor, social interaction, and skills that resist automation. New graduates are still finding openings in fields such as real estate, utilities, and construction, even as entry-level hiring in exposed white-collar roles has slowed.

The data favor roles where AI is used to assist the work rather than perform it, and the AI job gap itself argues for building the kind of tacit knowledge that accumulates through internships, projects, and mentorship. That advice lines up with the surge in year-end job searches among Gen Z workers reported this month, driven by pay, benefits, and advancement. Entry-level work is concentrating in different places than before, and the AI job gap is the figure that shows why.