At the AAAI Conference on Artificial Intelligence for Interactive Digital Entertainment Conference (AIIDE) in Edmonton, Canada, Florian Rupp and Manuel Eberhardinger presented their paper on explainable board game agents at the EXAG Workshop, and Florian Rupp presented his doctoral work on procedural content generation and balancing for games.
Cognitive-Plausible Explanations for Board Game Agents
At the Experimental AI in Games (EXAG) workshop during AIIDE 2025, Florian Rupp and Manuel Eberhardinger presented new methods to generate interpretable explanations for board game AI agents. The paper, “Towards Cognitive-Plausible Explanations for Board Game Agents with Genetic Programming”, by Manuel Eberhardinger, Florian Rupp, Florian Richoux, Johannes Maucher, and Setareh Maghsudi, explores how genetic programming and decision trees can produce post-hoc explanations for simple games like Tic-Tac-Toe and Connect 4. This approach highlights key factors behind agent decisions, helping designers and researchers understand AI behavior beyond black-box outputs.
Doctoral Consortium Contribution by Florian Rupp
In addition to the workshop, Florian Rupp participated in the AIIDE Doctoral Consortium, where he presented his ongoing research in the context of his dissertation on on ** game balancing through procedural content generation**. His contribution focuses on combining agent-based simulations with reinforcement learning and search-based optimization in order to create balanced content for all players of a game. The consortium offered a platform for discussion and feedback with other PhD students as well as senior researchers in game AI research.