Since the beginning of the use of technology to support training and learning there has always been the belief that such new technologies would be able to add value either by reducing costs or increasing effectiveness. The 1980s and early 1990s were a period of enormous optimism as to the promise that such technology could bring. The governments of Europe and the US were generous in their funding of research in this area. In Europe research and development programmes such as ESPRIT, DELTA, RACE, ERASMUS and COMETT, to name only a few, funded a wealth of initiatives aimed at advancing the use of technology. At the margins of the early initiatives was the belief that AI must have a part to play in these developments. This paper reviews the early initiatives and suggests reasons why the potential for the use of AI in education and training has never been truly fulfilled.