Tacit Knowledge

Knowledge that resides in individuals’ skills, intuitions, and experiences but cannot be fully articulated or codified. The things we know how to do but cannot explain completely.

The Polanyi Foundation

Michael Polanyi (1966): “We can know more than we can tell.” A skilled cyclist cannot fully explain how they balance. An expert clinician may recognize a diagnosis before consciously articulating the symptoms that led there.

Characteristics

Personal: Rooted in individual experience and context Context-dependent: What works in one situation may not transfer Hard to communicate: Requires demonstration, practice, apprenticeship Often unconscious: Experts may not know what they know

Why It Matters for AI

Large language models work with explicit, textual knowledge. Tacit knowledge, the “dark matter” of expertise, is largely inaccessible to them. This creates systematic gaps:

• Expert intuition about what questions to ask
• Judgment about when rules don’t apply
• Skills that develop through embodied practice

The Knowledge Engineering Challenge

Eliciting tacit knowledge from experts is expensive and incomplete. Knowledge graphs capture explicit relationships; the tacit understanding of when and how to use that knowledge remains with humans.

Related: 06-atom—tacit-knowledge-definition, 00-source—nonaka-1995-knowledge, 06-molecule—tacit-vs-explicit-knowledge, 06-molecule—seci-framework