Author: Steven Edwards
Date: 06:43:24 02/20/04
Concerning the topic of machine generated game annotations: One of the long term goals of the Symbolic project is to give the program the ability to produce an intelligent annotation of a PGN game score. The question here is: What does the adjective "intelligent" mean in this context? My answer is from Alan Turing's definition: machine behavior indistinguishable from human behavior (from his Imitation Game, a.k.a., the Turing Test). If Symbolic can produce game annotations that could fool a reasonably skilled chess player that such a report came from a human annotator, then I'll claim that the program's behavior, or at least its annotative behavior, is intelligent. It's not such a big deal. Back in the 1980s, there was an expert system named Puffer that produced analysis reports in the pulmonary pathology domain using raw lab result data as inputs. Puffer's reports were remarkably similar to those produced by pulmonary physicians. While some of the similarity was no doubt due to the somewhat rigid traditional style of the physicians, it was still a real achievement for the program. ---- Perhaps I'll take a brief detour with Symbolic's planner and code up a ChessLisp version of Turing's hand simulated chess program. After all, his was the first, and putting his code into Symbolic has an odd sense of appropriateness. And if you've read his paper on chess, you'll see why I'll then have to add a square root intrinsic to the ChessLisp interpreter.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.