The abstract promises an overview of the contributions of AI to education. This is a formidable task because, as the author consistently observes, measurably effective AI systems for education simply do not yet exist.
The author has done his homework; he covers all the bases: intelligent tutoring systems, AI programming environments, microworlds, Prolog, and databases and expert systems. To illustrate an intelligent tutoring system, he presents a brief but meaningful description of DEBUGGY, a system that functions well within a very narrow domain, subtraction. As an example of an expert system, he describes MYCIN, which serves as a skilled physician working with medical students; again, this is a system that operates within a narrow domain of knowledge.
Just when the reader begins to feel that a paper about AI in education is somewhat premature, the author shifts his emphasis from historical reporting to an analysis of major unsolved problems that must be overcome before AI in education becomes a functional reality. He recognizes one of the truths that educators have been maintaining since Computer-Assisted-Learning (CAL) was put forth as a substitute for a classroom teacher: AI systems will not succeed in education as long as they are viewed as complete environments. Computer-based courseware must be recognized as an element of a larger, human environment in which the emotional context is provided by the human teacher.
The paper is appropriately short (it matches the magnitude of the contribution AI has made to education). It is well written, and the language is neither technical nor condescending. It is recommended to naive readers who would appreciate a gentle introduction to the subject.