Directions for AI in the eighties

Abstract

The 1980's will be an exciting decade for artificial intelligence, one in which we can expect to see considerable progress on scientific questions, widespread application of Al techniques to problems of practical significance, and major changes in the social, political and economic structure of the field. In this article, I will attempt to identify some of the commercial, scientific and sociological considerations that I think will be important in coming years. In particular, I will introduce the metaphor of deep versus surface systems, and pose some challenges arising from the recognition of this distinction.