The enduring appeal of Mahjong: Navigating the challenges of AI integration

Abstract

Game is an important part of entertainment. This study applies Game Refinement Theory (GRT) to the zero-sum game of Mahjong, aiming to uncover its enduring appeal in terms of player engagement and to explore whether Mahjong can continue to attract players in the AI era. By applying Game Refinement Theory (GRT), we analyze real gameplay data from the online platform Tenhou and AI self-play data by akochan, following the Riichi Competition Rules (RCR). We calculate Game Refinement (GR) values to evaluate the balance between skill and chance in Mahjong. Our findings reveal that the GR value for human players averages around 0.088, indicating high engagement, while AI players exhibit GR values within the balanced range of 0.076, all within the range of game attractiveness. Key factors such as the number of riichi, furo, and player skill levels significantly influence game balance and excitement. Importantly, even as high-level players adopt AI strategies, Mahjong remains competitive and entertaining, suggesting that AI does not diminish its inherent appeal. These insights offer practical implications for game designers and AI developers aiming to enhance player experience and maintain game balance in the AI era.