An Empirical Study of GenAI Adoption in Open-Source Game Development: Tools, Tasks, and Developer Challenges

Abstract: The growing capabilities of generative AI (GenAI) have begun to reshape how games are designed and developed, offering new tools for content creation, gameplay simulation, and design ideation. While prior research has explored traditional uses of AI in games, such as controlling agents or generating procedural content. There is limited empirical understanding of how GenAI is adopted by developers in real-world contexts, especially within the open-source community. This study aims to explore how GenAI technologies are discussed, adopted, and integrated into open-source game development by analyzing issue discussions on GitHub. We investigate the tools, tasks, and challenges associated with GenAI by comparing GenAI-related issues to those involving traditional AI (TradAI) and NonAI topics. Our goal is to uncover how GenAI differs from other approaches in terms of usage patterns, developer concerns, and integration practices. To address this objective, we construct a dataset of open-source game repositories that discuss AI-related topics. We apply open card sorting and thematic analysis to a stratified sample of GitHub issues, labelling each by type and content. These annotations enable comparative analysis across GenAI, TradAI, and NonAI groups, and provide insight into how GenAI is shaping the workflows and pain points of open-source game developers.