This paper's first part, reviewed attempts to model real-world problem solving as machine-executable logic. Part 2 considers an alternative model in which the solution of problems is primarily the work of visualization supported by automatized skills. Consciousness operates at the level of goal-setting and monitoring, and of the construction and communication of after-the-event commentaries, not as a problem solver. Engineering designs based on this model have proved convenient and effective. `Structured induction' is now routinely used to recover and articulate expertise (...) that in the human solver remains tacit. A seminal case of computer-mediated superarticulacy is described in which a human problem solver was enabled to transform an elaborate, but largely blind and unconscious, mental skill into a fully aware, communicable and machine-executable theory. (shrink)

Uses of stored skill-models to accelerate simulator-based real-time training in a control skill are discussed. A real-time coach must deliver advice at three levels: (1) what to do next, (2) what to watch for, and (3) what went wrong. Human learning and machine learning results are presented using different screen representations of a pole-and-cart balancing task.

Consciousness has been widely regarded as the central arena for the mental solution of problems. A variant view locates the business end of problem solving elsewhere, with conscious intervention only for intermittent monitoring and goal-setting. In this scenario conscious awareness, with `intelligent' processes generally, is largely specialized to the construction and communication of appropriate after-the-event histories and explanations.The first part of the paper traces a long march undertaken by main-stream artificial intelligence basing itself on the first assumption. Disappointment with the (...) result has prompted interest in the second view, which forms the main topic of Part 2 , 1995, pp. 52-66). (shrink)

The definition of an expert system as a knowledge‐based source of advice and explanation pinpoints the critical problem which confronts the would‐be builders of such systems. How is the required body of knowledge to be elicited from its human possessors in a form sufficiently complete for effective organization in computer memory? This article reviews recent advances in the art of automated knowledge‐extraction from expert‐supplied example decisions. Computer induction, as the new approach is called, promises both important parallels to the human (...) capacity for concept formation and also commercial exploitability. (shrink)

Recently there has been an unexpected rebirth of Turing's imitation game in the context of commercial demand. To meet the new requirements the following is a minimal list of what must be simulated. Real chat utterances are concerned with associative exchange of mental images. They are constrained by contextual relevance rather than by logical or linguistic laws. Time-bounds do not allow real-time construction of reasoned arguments, but only the retrieval of stock lines and rebuttals, assembled Lego-like on the fly. A (...) human agent has a place of birth, age, sex, nationality, job, family, friends, partners, hobbies etc., in short a profile. Included in the profile is a consistent personality, which emerges from expression of likes, dislikes, pet theories, humour, stock arguments, superstions, hopes, fears, aspirations etc. On meeting again with the same conversational partner, a human agent is expected to recall not only the profile, but also the gist of previous chats, as well as what has passed in the present conversation so far. A human agent typically has at each stage a main goal, of fact-provision, fact elicitation, wooing, selling, "conning" etc. A human agent also remains ever-ready to maintain or re-establish rapport by switching from goal mode to chat mode. Implementation of this last feature in a scriptable conversational agent will be illustrated. (shrink)