Games: The Next Challenge
Résumé
After the successes of many chess-playing programs, one may wonder whether it is necessary to keep on developing chess programs. Perhaps it would be better to realize programs for games such as Go, where the best programs are still weak. In this contribution, we suggest that it is time to shift focus on general game-playing systems and to realize them. Such systems will receive the specification of a new game. Subsequently they will analyze the specifications so that they will be able to play the game reasonably well. We will see how one can define such a specification and how a system may study it in order to extract useful knowledge. The problem is ultra difficult, but also highly challenging for AI. Obviously, as long as we call a system intelligent when it solves only one problem we must face the critique that most of the intelligence is in the brain of the author and little in the program itself.