Aysa Fan's essay warns AI breaks education's historic pattern

In a new essay published on arXiv (2605.16283), researcher Aysa X. Fan examines how generative AI fundamentally disrupts a centuries-old pattern. Historically, every new technology automates some layer of cognitive work, and education responds by retreating upward to teach the skills machines cannot yet reach. But generative AI now operates at the very top of that cognitive ladder — the same space where education has always escaped to. The risk, Fan argues, is not that AI replaces teachers but that it replaces the productive struggle through which genuine understanding forms. Drawing on historical analysis, labor economics, and new large-scale data on how students and workers actually use AI, the essay surfaces a paradox: the same tools that augment today's skilled workforce may be quietly eroding the developmental process that produces tomorrow's experts.

Fan notes that current assessment tools cannot distinguish students who are building capacity from those who are losing it. This is primarily a measurement problem, and secondarily a design problem. The essay proposes a research agenda focused on learning outcomes rather than usage patterns, and asks what education should become once AI can perform the cognitive work education was built to develop. Capacities like judgment, character, and epistemic identity have not been central to mainstream educational taxonomies because earlier technologies did not require education to reach so high. Fan offers no final destination, only directions — but the implication is clear: without deliberate measurement reform, AI could quietly hollow out the very process that creates skilled professionals.