I think that if we are looking for computational models of cognition, we should look at brains. They work. But artificial intelligence has turned out to be most successful at making programs that make people smarter. We shouldn't be reluctant to admit this, indeed programs using artificial intelligence technology that help us organize and manage an increasingly complex and technical world will probably be the most profound and valuable legacy of artificial intelligence. Cognitive science students need to be trained in programming so as to develop the intuitions about the nature of compu- tation that they will need in order to understand how the brain implements cognitive processes, and to learn how artifacts can extend cognitive abilities. At this point in the development of the field we can't be so sure that any specific aspect of computation will turn out to be irrelevant. Students should thus be given a wide-range of programming experience. Artificial intelligence models, taught as programming exercises and situated within a large set of approaches to cognitive modeling and cognitive engineering, can provide part of this experience. Intellectual Archeology P a t Hayes Beckman Institute University of Illinois 405 North M a t h e
/lp/association-for-computing-machinery/intellectual-archeology-byNOaJPFgB