Inverse Bayesian Inference for Player Agents in Computer Daihinmin Game

In recent years, advancements in game AI research have led to AI systems increasingly surpassing human abilities. However, traditional AI models lack the flexibility and creativity inherent to human decision-making. In this study, we focused on the card game “Daihinmin” (Grand Pauper’s Game) and applied Augmented Bayesian Reasoning to develop an AI agent model that emulates human strategic behavior. The results demonstrated that the model successfully reproduced biological-like behavior, as evidenced by characteristics resembling a powerlaw distribution in strategy selection and consecutive wins. This research marks a significant step towards the development of game AI capable of exhibiting flexible and natural behaviors.