New Theory Explores AI's Cognitive Debt and Systemic Fragility

Researchers have developed a formal theory termed "cognitive debt," which describes the accumulation of unverified reasoning obligations when artificial intelligence is used to replace fundamental cognitive processes rather than augment them. The model posits that agents possess both cognitive capital and cognitive debt, with cognitive capital acting as collateral that influences the returns from AI adoption. The theory outlines six key propositions. It suggests that individuals rationally incur cognitive debt because its costs are often deferred, partially externalized, and masked by immediate productivity boosts. Periods of stability can lead to lower perceived risks, increasing the intensity of AI substitution and compounding this leverage, potentially resulting in a "cognitive Minsky moment" where subjective risk decreases while actual systemic fragility rises. Furthermore, the model indicates that expected crisis losses are convex with aggregate leverage. Post-crisis, pressure to meet output targets can produce a "false-correction loop," where AI failures are addressed by deploying more AI. The decentralized adoption of substitutive AI is shown to be suboptimal compared to a social optimum, primarily due to systemic risk, the public good nature of cognition, and arms-race dynamics.