LLMs as Explicit Knowledge Processors

LLMs excel at combining and restructuring explicit knowledge but struggle with the tacit-to-explicit conversion that defines expertise.

In knowledge elicitation research, AI-led interviews produced more structured, information-dense outputs than human interviews. The AI was better at explicit-to-explicit transformation, organizing stated information into consistent formats. But the resulting ontologies missed the depth and nuance that human knowledge engineers captured.

Human experts bring tacit knowledge to the structuring task: intuitions about what belongs together, implicit understanding of hierarchical relationships, sense of what’s essential versus incidental. These don’t transfer through text.

This maps directly to knowledge management theory: LLMs operate primarily in the Combination quadrant (explicit→explicit), while humans are still required for Externalization (tacit→explicit).

Related: 06-molecule—seci-framework, 06-atom—tacit-knowledge