Using a Computer Game to Develop Advanced AI

  • Authors:

  • John E. Laird

  • Affiliations:

  • -

  • Venue:
  • Computer
  • Year:
  • 2001

Quantified Score

Hi-index 4.10

Visualization

Abstract

In computer games, designers can use artificial intelligence to control individual characters, provide strategic direction to character groups, dynamically change parameters to make the game appropriately challenging, or produce play-by-play commentary. Computer games offer an inexpensive, reliable, and surprisingly accessible environment for conducting research in human-level AI design, often--as in the case of Quake II--with built-in AI interfaces. The author's work with the game's Quakebot demonstrated that researchers can successfully pursue serious study of autonomous AI agents within the context of computer games. This research directly applies to computer-generated forces, which require modeling realistic, entity-level behavior. Studying the impact of changes in reaction time, tactics level, and perceptual and motor skills on over-all Quake II game performance helped to model these behaviors. From its scoring method, which rewards the highest number of kills, it's obvious that Quake II epitomizes violent computer games. The author does not, however, believe that the future of AI in games lies in creating ever more realistic arenas for violence. Thus, he is pursuing further research within the context of creating computer games that emphasize the drama that arises from social interactions between humans and computer characters.