General Game Playing: Overview of the AAAI Competition

Article excerpt

Games of strategy, such as chess, couple intellectual activity with competition. We can exercise and improve our intellectual skills by playing such games. The competition adds excitement and allows us to compare our skills to those of others. The same motivation accounts for interest in computer game playing as a testbed for artificial intelligence: programs that think better should be able to win more games, and so we can use competitions as an evaluation technique for intelligent systems.

Unfortunately, building programs to play specific games has limited value in AI. To begin with, specialized game players are very narrow: they focus only on the intricacies of a particular game. IBM's Deep Blue may have beaten the world chess champion, but it has no clue how to play checkers; it cannot even balance a checkbook. A second problem with specialized game-playing systems is that they do only part of the work. Most of the interesting analysis and design is done in advance by their programmers, and the systems themselves might as well be teleoperated.

However, we believe that the idea of game playing can be used to good effect to inspire and evaluate good work in AI, but it requires moving more of the mental work to the computer itself. This can be done by focusing attention on general game playing (GGP).

General game players are systems able to accept declarative descriptions of arbitrary games at run time and able to use such descriptions to play those games effectively (without human intervention). Unlike specialized game players such as Deep Blue, general game players cannot rely on algorithms designed in advance for specific games. General game-playing expertise must depend on intelligence on the part of the game player itself and not just intelligence of the programmer of the game player. In order to perform well, general game players must incorporate various AI technologies, such as knowledge representation, reasoning, learning, and rational decision making; these capabilities must work together in an integrated fashion.

Moreover, unlike specialized game players, a general game player must be able to play different kinds of games. It should be able to play simple games (like tic-tac-toe) and complex games (like chess), games in static or dynamic worlds, games with complete and partial information, games with varying numbers of players, with simultaneous or alternating play, with or without communication among the players.

While general game playing is a topic with inherent interest, work in this area has practical value as well. The underlying technology can be used in a variety of other application areas, such as business process management, electronic commerce, and military operations.

In order to promote work in this research area, the AAAI is running an open competition on general game playing at this summer's Twentieth National Conference on Artificial Intelligence. The competition is open to all computer systems, and a $10,000 prize will be awarded to the winning entrant.

This article summarizes the technical issues and logistics for this summer's competition. The next section defines the underlying game model. The "Game Descriptions" section presents the language used for describing games according to this model. The "General Game Players" section outlines issues and strategies for building general game players capable of using such descriptions; it is followed by the "Game Management Infrastructure" section, which discusses the associated general game management infrastructure. Our conclusion offers some perspective on the relationship between general game playing and the long-range goals of AI.

Game Model

In general game playing, we consider finite, synchronous games. These games take place in an environment with finitely many states, with one distinguished initial state and one or more terminal states. In addition, each game has a fixed, finite number of players; each player has finitely many possible actions in any game state, and each terminal state has an associated goal value for each player. …