AI, Human Cognition and Knowledge Collapse

Summary

A dynamic theory paper modelling how generative — and specifically agentic — AI shapes human learning incentives and the long-run evolution of society's information ecosystem. Successful decisions require combining a community-level stock of general knowledge with individual-level context-specific knowledge; human effort jointly produces a private signal and a "thin" public signal that accumulates into the shared stock, creating a learning externality. Agentic AI delivers context-specific recommendations that substitute for that effort, while a richer stock of general knowledge complements it.

Contribution

The authors identify a regime in which the economy "can tip into a knowledge-collapse steady state in which general knowledge vanishes ultimately, despite high-quality personalized advice." Welfare is shown to be non-monotone in agentic accuracy, implying an interior welfare-maximizing precision and motivating information-design regulation; greater aggregation capacity for general knowledge unambiguously raises welfare and resilience.

Method

Conceptual / theoretical: a dynamic economic model of learning and decision-making under social-learning externalities, solved for steady-state regimes and comparative statics in agentic accuracy and aggregation capacity.

Relevance to RISE

Acemoglu and co-authors model knowledge collapse — the field-level epistemic risk when humans defer to AI-generated knowledge faster than they verify it. Macroeconomic foundation for the field-level RISE evaluation question.

Critique / open questions

Stylized model with no empirical calibration in the visible excerpt; the agentic-accuracy threshold is a free parameter, leaving the practical question of when real-world systems cross it unresolved. "General knowledge" is collapsed to a single scalar, abstracting from disciplinary heterogeneity.

Key quotes

"Agentic AI delivers context-specific recommendations that substitute for human effort. By contrast, a richer stock of general knowledge complements human effort by raising its marginal return."

"The economy can tip into a knowledge-collapse steady state in which general knowledge vanishes ultimately, despite high-quality personalized advice."