New Stanford Study Finds AI Is Disproportionately Reducing Jobs for Young Workers in Automative Roles, Highlighting a Growing Labor Market Divide

The debate over whether artificial intelligence is harming job prospects for young people has taken a new turn, with a fresh study from Stanford economists presenting some of the strongest evidence yet that AI may already be reshaping the labor market for early-career workers. The research, based on payroll data from ADP covering millions of workers through mid-2025, reveals a striking 13% decline in employment for young adults aged 22 to 25 in jobs highly exposed to AI—such as software developers and customer service agents—since the rise of large language models like ChatGPT. In contrast, older workers and those in less AI-exposed roles, like home health aides, have seen stable or even growing employment. This pattern is not just a broad economic trend. The study finds that the decline is most pronounced in roles where AI can automate repetitive, rule-based tasks—like drafting reports, formatting documents, or translating text—while jobs requiring strategic thinking, long-term planning, or tacit knowledge remain resilient. The data suggests AI is replacing certain types of work more than others, particularly tasks that young workers often perform. One key insight from the paper is the distinction between automation and augmentation. When AI is used to fully replace human tasks—what the researchers call "automative" use—employment for young workers drops. But in roles where AI supports human work through collaboration, feedback, and learning—what they term "augmentative" use—there is no significant decline, and in some cases, growth. The study also shows that within the same company, AI’s impact varies by department. For example, a legal team relying on AI to summarize documents and generate reports may reduce entry-level hiring, while a product team using AI to brainstorm ideas may see increased demand for junior roles. This suggests that blanket statements like “AI is killing jobs” are too simplistic. The real story is about how AI reshapes specific job functions, not entire industries. The researchers emphasize that their work is observational, not causal. They cannot prove AI caused the employment drop, but they’ve controlled for other factors like interest rates, remote work trends, and post-pandemic hiring cycles. Even after accounting for these, the decline in young workers’ employment in high-exposure jobs remains clear. The findings align with broader observations about the nature of AI and human skill. Young workers often rely on codified, written knowledge—exactly what LLMs are trained on. Senior workers, by contrast, depend more on experience, intuition, and unwritten workplace wisdom—areas where AI still lags. Tasks with clear, measurable outcomes are more easily automated, while complex, long-term projects remain harder for AI to handle. For educators and students, the message is clear: the future of work isn’t just about avoiding AI—it’s about learning how to use it. Universities may need to update curricula to teach not just technical skills, but also how to collaborate with AI tools effectively. Skills in strategic thinking, creativity, and physical or interpersonal work—areas less susceptible to automation—may become increasingly valuable. In short, the evidence is shifting. After months of conflicting signals, this new data suggests that AI is not just a future threat—it’s already altering the job landscape for young people. The debate is far from over, but the most compelling data now points to a real, measurable impact. As one researcher put it, the best way to understand the future isn’t to predict it—but to pay close attention to what’s happening right now.