Cognitive Debt: The Cumulative Cognitive Cost of AI-Augmented Knowledge Work

AI systems are becoming infrastructural in knowledge work, yet their longitudinal effects on cognition remain poorly understood. This article introduces cognitive debt: the accumulated, delayed costs to attention, learning, and mental health arising from chronic reliance on AI systems that reduce active cognitive engagement. Distinct from momentary cognitive load, cognitive offloading, and automation bias, cognitive debt is inherently temporal — accruing invisibly until individual capacity thresholds are exceeded. Within predictive processing frameworks, it reflects systematic downward adjustment of precision on internal generative models in favor of external AI scaffolding. Three interacting mechanisms — attentional erosion, effort displacement, and affective depletion — are proposed, and neurodivergent knowledge workers are identified as a theoretically critical early-warning population. Three falsifiable predictions anchor the empirical program.