Thin Knowledge, Thick Knowledge

In the AI Story

The distinction is not absolute. It is a spectrum, and different kinds of knowledge fall at different points on it. Factual information — the capital of France, the boiling point of water — transfers relatively well across contexts and can be decontextualized without significant loss. Procedural knowledge — the steps for tying a knot, the syntax of a programming language — transfers somewhat less well but can still be usefully communicated in abstraction. The kind of knowledge that matters most in professional practice — the judgment about which facts are relevant, which procedures to apply, when to deviate from the standard approach, how to respond when the situation is not quite what anyone expected — resists decontextualization almost completely.

Thick knowledge is produced by the process Lave's framework describes with precision: sustained participation in a community of practice, through which the practitioner accumulates not just propositional knowledge but contextual understanding — the kind of understanding that manifests as judgment, intuition, the ability to act wisely when the formal knowledge runs out. The process is slow. It is often frustrating. It is inefficient by any output metric. And it is irreplaceable.

The distinction is often invisible in normal operations. When the system is running smoothly, when cases are routine, when situations fall within parameters that formal knowledge anticipates, thin knowledge is sufficient. The junior developer with Claude's explanations can maintain the system, ship features, fix bugs. The junior lawyer with AI-generated briefs can serve clients competently. This is why the thinning of knowledge is so difficult to detect — it does not show during normal operations. The distinction reveals itself at the margins, when the situation is abnormal and formal knowledge proves insufficient, and the practitioner must rely on judgment.

Judgment is the application of thick knowledge to situations that thin knowledge cannot resolve. It is the capacity to act wisely in the absence of clear rules. AI-augmented workflows tend to favor breadth over depth — producing practitioners who are wider and thinner, more capable on the surface and less reliable at the margins. The consequences are systemically invisible in the metrics that organizations typically use and structurally consequential at the moments when thick knowledge is the difference between a managed problem and a catastrophe.

Origin

The terminology of thin and thick knowledge in On AI extends Lave's own vocabulary, which spoke of "situated" versus "decontextualized" knowledge. The thin/thick framing draws on Clifford Geertz's distinction between thick description and thin description in ethnographic method, adapting the anthropological insight to the epistemological question of what AI produces and what it cannot.

Key Ideas

The distinction is not quantitative. Thick knowledge is not more thin knowledge. It is a different kind of knowledge, produced by a different process, reliable in different ways.

Normal conditions hide the distinction. Both kinds of knowledge produce adequate outputs most of the time. The distinction surfaces at the margins.

Judgment requires thick knowledge. The capacity to act wisely when rules run out is produced by situated participation and cannot be short-circuited by information transfer.

AI favors breadth over depth. The economics of tool-mediated work reward scope expansion over sustained engagement in any single domain — producing practitioners who are wider and thinner than their predecessors.