26 May 2003, Spinoza Lectures (held in English), Prof. Hubert L. Dreyfus (University of California, Berkeley)

From Socrates to AI researchers, rationalist philosophers have assumed that intelligence is based on features and rules. Thus, expert system builders assume that expertise consists in problem-solving and that problem-solving consists in stepping back, analyzing a situation in terms of its objective features, and then finding a situation-action rule that determines, on the basis of those features, what should be done. But such systems have failed to exhibit expertise. I will argue that this failure results from a misunderstanding of the nature of expert understanding. Phenomenological description reveals that expertise is normally acquired in a five-step process: The beginner does, indeed, pick out objective features and follow strict rules like a computer. The advanced beginner, however, responds to meaningful aspects of the situation that are recognized as similar to prototypical cases, whithout similarity being analyzed into objective features. At the next stage, the competent performer learns to figure out a strategy and to pay attention only to features and aspects that are relevant to his plan. The fourth stage, proficiency, is achieved when the performer no longer has to figure out his strategy but immediately grasps an appropriate plan. Finally, the expert, after many years of experience, is able to stay involved and spontaneously do what works, without having to reflect and reason at all. The above progression suggests that the expert's brain becomes able to discriminate a vast number of types of whole situations and associate an appropriate action to each without using conscious or unconscious rules. that is why expert systems based on rules obtained from experts can at best exhibit competence.

For more information, please contact illc at science.uva.nl