There has been a significant increase over the past several years of interest in research focused on the applications of artificial intelligence to computer games. The goal of AI agents in these cases is to maximize their payoff. In recent years similar approaches have been applied to newer games of real-time strategy, first person shooters, and more. Despite the relative complexities of these environments compared to chess, the fundamental goals of the AI agents remain the same: to win the game.
We argue that game AI techniques should operate to maximize the interestingness of sequences of states of states should tell a good story of a game experiences. In AI and ML techniques should focus on telling the player enjoying a game. A hard fought battle that results in a loss can be more enjoyable than an easy win. How they reach the ending can often be as important as what the ending is.
Making the connection between a trajectory of states in a game and a narrative provides us with a set of tools for thinking about and evaluating sequences. We argue that formalizing computer games in a manner similar to a story provides the groundwork for developing “better” game AI. We conclude with an enumeration of open research questions entailed by our approach that, if met, may dramatically improve the quality of game AI in the future.
Story_auto3