Knowledge Management and AI

Knowledge is not neatly packaged. It is difficult to extract, it is ephemeral and may only be approximately correct.

One of the chief barriers to the construction of expert systems is the difficulty of knowledge acquisition—and this barrier must also be overcome to achieve success in knowledge management. However, there is some good news. The problem may be more tractable in the new context.

Systems that support knowledge management typically do not try to solve a problem alone. Rather, they try to find the knowledge (best practices, lessons learned, tips, solutions to related problems, and so on) that assist people in developing their own solutions. Stated another way, the goal today is to help people solve problems.

Contrast this with the original goal of expert systems—to solve problems themselves at an expert level. This is not to say that the AI community should give up on the expert systems goal, but rather, that achieving the goal of giving powerful assistance to people as they solve problems is of great interest to the knowledge management community.

Extraído de The Road Ahead for Knowledge Management
An AI Perspective
Reid G. Smith and Adam Farquhar
AI MAGAZINE, WINTER 2000