Some recent work in artificial intelligence

Abstract

This paper will review certain approaches to artifical intelligence research--mainly work done since 1960. An important area of research involves designing a machine that can adequately improve its own performance as well as solve other problems normally requiring human intelligence. Work in heuristic programming that seems most relevant to this goal will be discussed at length. Important subproblems are devising techniques for self-improvement, the general problem of deciding what task to best work on next in a network of tasks, and the general problem of how to mechanize learning or inductive inference. Some work in linguistics and pattern recognition is directly concerned with the induction problem. Another area of research that will be treated is simulation of organic evolution.