On Evaluating Agents for Serious Games

Emma Norling

Abstract: With the recent upsurge in interest in agents for ‘serious games,’ there has been a focus on the ‘believability’ of agents in these settings. In this paper, I argue that when evaluating agents in this context, believability is often in fact of relatively minor importance, and indeed that focusing on this criteria can detract from the ultimate goals of the games. I present this argument in the context of a project for which the aim was to extend the BDI agent framework to better support human modelling in serious games.