AI in Games

In this episode, Kyle Polich interviews the authors of the first textbook dedicated to artificial intelligence in games — Georgios Yannakakis at the University of Malta and Julian Togelius at New York University. There is an increasing use of video games to develop and test AI, and there’s also a long-standing use of AI methods in designing games. Both fields use very different methods, and Georgios and Julius’s text book, Artificial Intelligence and Games try to bridge this divide, as the two fields don’t always understand each other.

Julius and Georgios argue that while games pose interesting and complex problems for AI to solve, AI also helps improve games. Their book looks at three broad areas: playing games as a player or non-player, generating content, and modeling players. In the area of playing games, there are some that involve just players, such as chess and Go, and then there are games in which non-players are involved, such as in Zelda, Call of Duty.

Games also involve generating content involves creating game worlds such as levels, maps, textures or characters. Julius and Georgios argue that if you think of content generation as an AI problem, game designers can do so much more. Finally, when you look at games involving non-players, an area of research interest is modeling players, which involves learning how to predict how players will feel or what players will do.

Learn more about the book and the 2018 summar course being taught by the authors and industry veterans at http://school.gameaibook.org/