Long-term Consequences of Generative AI on Education: Cost Disease and Widened Inequality in the Gen-AI Era

Generative AI promises efficiency, yet in education it risks intensifying Baumol’s cost disease and widening inequality. We propose a five-step mechanism: (1) AI lifts productivity and wages elsewhere; (2) quality-preserving teaching tasks remain weakly automatable; (3) schools face rising wage pressure with limited gains in quality-adjusted output; (4) shortages deepen in high-demand subjects and hard-to-staff locales; and (5) coping capacities diverge, producing uneven class sizes, staffing, and AI access. Drawing on a narrative synthesis of recent research, audits, policy documents, and procurement records across K–12 and higher education, we integrate early signals on adoption, workforce strain, and assurance costs. We explain why time savings are typically reinvested in personalization, feedback, and safeguarding rather than used to raise student–teacher ratios, leaving the production function largely intact. Anticipated outcomes include a larger teacher pay penalty, persistent vacancies, and uneven deployments that risk widening achievement gaps. We conclude with testable predictions and an equity-oriented policy toolkit (market-indexed pay, targeted retention incentives, pooled infrastructure, interoperable procurement, strengthened data governance, and weighted funding) to realize modest, real benefits while avoiding a two-tier system in diverse national contexts.