Digitizing the "Shokunin": How we encoded a Master's hammer strike into AI (yusukekaizen.substack.com)

A new approach to capturing expert intuition in AI systems has been introduced, addressing the longstanding challenge of translating the nuanced decision-making processes of master craftsmen, or "Shokunin," into actionable data for machine learning. The author emphasizes that traditional methods focus only on observable outcomes (Action Volume), neglecting the crucial underlying thought processes (Thinking Volume) that inform those actions. By using a novel mathematical framework known as the Verbalization Accuracy Model (TAREFC), which strategically structures interviews to extract the "why" behind decisions, engineers can better harness this hidden intuition for AI training. This methodology signifies a paradigm shift in developing AI systems, particularly in high-risk manufacturing environments where intuition is critical. An illustrative case study involved refining an AI's ability to assess quality based on the sound made by a hammer strike, transitioning from standard frequency analysis to analyzing the decay rate of sound—a vital nuance identified through tapping into the master’s thinking processes. This innovative approach not only enhances AI's accuracy but also redefines the role of Shokunin, transforming them from mere craftspeople into educators who share their intuitive wisdom, ultimately fostering a collaborative relationship between human expertise and AI capabilities.