11 Intelligent Computer - Aided Instruction for Medical Diagnosis

As AIM researchers began to develop techniques for allowing systems to explain their reasoning, some researchers became intrigued by the potential educational role of the developing methods. It became clear that advanced computer-aided instruction (CAl) programming techniques could be applied and extended in the medical setting. Intelligent computer-aided instruction (ICAI) differs from traditional CAl in its use of AI techniques for representing both subject material and teaching strategies. Among ICAI programs, Clancey's GUIDON system described in this chapter is one of the largest and most complex. It contains all of the knowledge of MYCIN (Chapter 5) and uses a variety of techniques for mixedinitiative dialogue, student modeling, and response to partial student solutions. As a Stanford graduate student, Clancey had been involved in much of the early work on MYCIN and also became interested in ICAI and the possibility of adapting MYCIN for educational purposes. Thus GUIDON reflects the tremendous effort that went into building MYCIN's knowledge base of infectious disease rules, as well as nearly a decade of research in building ICAI systems. MYCIN's good performance in reaching decisions and giving explanations made a tutoring application of the knowledge base attractive. GUIDON also demonstrates the value of representing knowledge so that it can be applied in multiple settings, here for both consultation and teaching. This is the main advantage of separating