Research Summary: Large Language Models Explain Experts Better Than Experts Themselves

A recent study investigates the capability of Large Language Models (LLMs) to externalize tacit knowledge from expert behaviors. The research indicates that LLM-generated knowledge can enhance decision-making quality and enable novices to achieve near-expert performance. This suggests a significant potential for LLMs in knowledge transfer and retention, addressing the long-standing challenge of articulating and preserving expert 'know-how'.

Why it matters

The ability of LLMs to externalize and transfer tacit knowledge has profound implications for organizational learning, workforce development, and operational efficiency. It offers a potential solution to the critical problem of knowledge loss when experienced personnel depart, enabling more robust knowledge management strategies.

Key insights

• Tacit knowledge, the 'know-how' derived from experience, is inherently difficult for experts to articulate and often poorly documented.
• The study explored whether LLMs can externalize tacit knowledge from observed expert behaviors.
• LLM-externalized tacit knowledge improved the quality of downstream decision-making.
• Novices utilizing LLM-externalized knowledge approached expert-level performance.
• LLM-externalized knowledge often surpassed the effectiveness of knowledge articulated by human experts.
• These findings offer empirical support for Polanyi's Paradox, which posits that 'we can know more than we can tell'.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.07488