CONCEPT
The Disembodied Generative Model
Clark's diagnosis of what distinguishes large language models from biological cognition — a generative model without embodied grounding, statistically fluent but unable to check its outputs against reality.
Both the brain and the large language model are generative models — systems that predict outputs based on learned statistical regularities. But the brain's generative model is tethered to reality by embodiment: the organism acts on the world, receives feedback about consequences, and updates its predictions when the world pushes back. The language model is not. Its predictions are constrained only by linguistic patterns, which are correlated with reality but not identical to it. This architectural difference, Clark argues, is the structural source of AI hallucination and the reason the biological component of extended cognitive systems is architecturally necessary.
In The You On AI Field Guide
The distinction matters because it identifies what the AI cannot supply. Language follows patterns. Reality is one thing that generates those patterns, but not the only thing. Literary convention, argumentative structure, rhetorical expectation, and sheer frequency of co-occurrence all generate patterns too. The model cannot distinguish between patterns that reflect reality and patterns that reflect the