Learning, Generality and Problem-Solving
Abstract
This Memorandum is a discussion of the concept of learning in the field of artificial intelligence and its intimate relationship to other concepts such as generality and problem-solving. It is part of a continuing RAND research effort in these areas. The long range goals of artificial intelligence imply the ability for programs to be truly general purpose, in the sense of being able to acquire from their environment the information necessary to develop successfully in ways not envisioned in detail by their designers. The analysis given here shows that learning is generally viewed as the means to accomplish this. These issues are fundamental not only to the field of pure artificial intelligence, but to the whole attempt to develop more sophisticated information processing, such as in Command and Control systems. To illustrate some of the questions brought up in the paper, a discussion is given of some very recent work done at RAND on a learning scheme for the General Problem-Solver (GPS). More details and specifications of this scheme will be given in a subsequent RAND Memorandum. This revision of RM-3285-PR incorporates several changes and additions designed to present a more straight-forward approach to the subject matter.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1963
- Accession Number
- AD0298949
Entities
People
- Allen Newell
Organizations
- RAND Corporation