The application of artificial intelligence to computer aided learning promises to produce a new generation of computer based tutors that are able to provide individual tuition of a comparable standard to that of a human tutor. However, artificial intelligence based computer aided learning is still in its infancy and existing methodologies contain numerous fundamental and practical pitfalls. Among the problems faced by various artificial intelligence based computer aided learning approaches are: prohibitive developmental overheads; prohibitive operational overheads; inability to handle complex domains; lack of sufficiently rich knowledge representations; lack of sufficient meta knowledge; and need for excessive domain knowledge. This paper examines the scope and limitations of existing artificial intelligence based computer aided learning methodologies. This examination is set in the context of a historical examination of the major systems and approaches to be produced.