The Hybrid Knowledge Elicitation Pattern

When extracting expert knowledge and encoding it into formal structures, the best results come from combining AI-led data collection with human-led semantic structuring.

AI interviews are faster (3.5x), more structured, and stay on topic. But AI-generated ontologies are consistently smaller, miss hierarchical nuance, and confuse types with instances. Human knowledge engineers produce richer, more accurate structures, but the interview phase is expensive and time-consuming.

The pattern that emerges: let AI handle what it does well (rapid, consistent elicitation), then let humans handle what they excel at (interpretation, abstraction, nuanced structuring).

Related:, 06-molecule—seci-framework