Lottery and Sprint: Generate a Board Game with Design Sprint Method on AutoGPT

This research introduces the Lottery and Sprint methodology, a framework designed to facilitate collaborative board game creation by integrating human design intuition with the structured Design Sprint process executed by AutoGPT. The primary objective is to determine if an AI-driven system can effectively generate playable, engaging, and original board games, even for users with limited design experience. The methodology employs a cyclical, iterative approach where AutoGPT first generates multiple game concepts—the lottery phase—before refining a selected design through a customized Design Sprint process that incorporates user feedback and playtesting.

The study evaluated the system’s performance through a user experiment involving 12 participants organized into three teams, each consisting of a Game Master and three players. Quantitative data was gathered using the Creativity Support Index (CSI) and post-gameplay surveys, while qualitative insights were derived from participant feedback and researcher observations. Findings indicate that the system successfully produces original and fascinating game concepts, achieving a total CSI score of 77.58. Participants reported positive experiences, particularly regarding the system's ability to foster collaboration and exploration.

However, the analysis also identified significant challenges. While Game Masters noted improvements in rule clarity and fairness after iterative updates, these refinements did not always translate into a better experience for players, suggesting a disconnect between design intent and player perception. Technical limitations, such as the system's restricted working memory, occasional generation of contradictory rules, and difficulties in creating coherent board graphics, were also noted. The authors conclude that while the current framework shows promise for democratizing game design, future advancements in AI memory, constraint adherence, and graphical capabilities are necessary to enhance the overall quality and consistency of AI-generated board games.