What is tacit knowledge?

Tacit knowledge definition

Tacit knowledge is the part of knowing that cannot be fully reduced to a manual, checklist, or explicit formula without losing something important. It often appears as judgment, feel, timing, taste, or the ability to notice the right signal before others do.

This is why two people can read the same SOP and still produce radically different outcomes. One of them is using a hidden layer of experience-bound cues.

Michael Polanyi and tacit knowing

Michael Polanyi's most cited line is simple: we know more than we can tell. That insight matters because it explains why high performance is often easy to recognize but hard to transfer. Tacit knowledge is not irrational. It is structured, but much of its structure lives below explicit narration.

Why tacit knowledge matters for AI agents

Most AI systems can already follow explicit rules. The bottleneck is not syntax. The bottleneck is judgment. If you want better copilots, better expert systems, or better onboarding, you need to surface the tacit layer: what the expert notices, ignores, escalates, and refuses to do.

Chinese summary

默会知识常被误解成“说不清的玄学”。更贴近事实的说法是，它是专家长期在真实场景中形成的判断结构。它常常体现在火候、取舍、注意力切换、升级边界和失败前兆上。你要把它变成 AI agent，第一步先把这些隐藏线索挖出来。