Possibilities for Learning in Game Artificial Intell igence
Arild Jacobsen, Sindre Berg Stene, Sule Yildirim Yayilgan · 2009
1 Abstract In our earlier research, we looked into the need for and use of AI in video games. Our survey on the existing literature on game artificial intelligence and our hands-on experience with some of the which were developed through 1990s up to today have shown that the Artificial Intelligence in commercially available video has made significant progress over the decades, but one area which commercial have largely ignored is the use of learning AI. Meanwhile, game artificial intelligence research continues to look into and create examples of using such artificial intelligence techniques, e.g. reinforcement learning, evolutionary algorithms, in academic games. At the moment these techniques are largely employed only by game artificial intelligence research; however, considering that game environments in commercial are becoming more dynamic and unpredictable, one would think that these techniques will be more capable of handling such environments and as such would be more widely used by commercial developers. Even so, it is still rare that commercial game developers employ these techniques in their games. In this paper, we will investigate the reasons behind that by looking at the possible benefits and problems, as well as the current state of learning in game artificial intelligence. in the section The history of learning in video firstly we examine the status quo of several academic and commercial in terms of how much learning is present in them, what learning is useful for in these games, and the nature of the techniques behind learning in them. Secondly, in the section can game agents learn?, we list actions and situations that the game agents (can) learn in games. Thirdly, having examined the status quo and the possible actions for learning, under the title The benefits of using learning in we look into more general reasons for using learning in and the benefits that might come from doing so. Are there disadvantages that overshadow the benefits, causing game developers to avoid using other AI techniques than a few common ones? What are these disadvantages? Under The issues and techniques for implementing learning in games title, we focus on the techniques that could be used to implement learning in commercial games, and also examine the problems with making use of these techniques. Next, in the last section before the conclusion, titled Evaluation of game agents' learning capabilities: a commercial perspective, we look at real-world examples, first mentioned in the history section, of commercial using learning techniques in light of the benefits and problems described in the previous sections. Finally, we bring our observations together to present our conclusions. 1 Definition of Machine Learning: A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E (Mitchell, 1999)