AI for Dynamic Difficulty Adjustment in Games

Contents

Video Games are boring when they are too easy and frustrating when they are too hard. While most singleplayer games allow players to adjust basic difficulty (easy, medium, hard, insane), their overall level of challenge is often static in the face of individual player input. This lack of flexibility can lead to mismatches between player ability and overall game difficulty. In this paper, we explore the computational and design requirements for a dynamic difficulty adjustment system. We present a probabilistic method (drawn predominantly from Inventory Theory) for representing and reasoning about uncertainty in games. We describe the implementation of these techniques, and discuss how the resulting system can be applied to create flexible interactive experiences that adjust on the fly.