An approach to artificial intelligence

In this paper, attention is called to certain problems which are extremely primitive conceptually, and yet exhibit certain features quite relevant for study in the field of artificial intelligence. These issues are presented with the belief that realistic study of them can provide approaches to artificial intelligence. Turing machines are used to present the problems. By this means, the concept of optimality is introduced. The paper then addresses the problem of discovering optimal machines using programmable methods. The requirement for proofs of methods is emphasized. Apparent ways of limiting the difficulties encountered are shown to be deceptive. Finally, the effect achieved by changing the required format is illustrated.