Two-Year Repeated Cross-Sectional Study of AI Integration and Learning in Global Game Jams

Abstract

Game jams are intense, collaborative learning environments where rapid skill acquisition occurs. The introduction of generative Artificial Intelligence (AI) has challenged traditional notions of creative ownership and learning struggle. This research offers a repeated cross-sectional, survey-based study comparing data from two years of Global Game Jam (GGJ) participants at a Dalhousie University site (N = 112, 2025; N = 67, 2026), examining the development of AI integration through the framework of Self-Determination Theory (SDT). Reflexive thematic analysis, descriptive statistics, and non-parametric group comparisons are used to compare participants’ perceptions of competence, autonomy, and relatedness. In 2025, AI use was distributed broadly across coding and ideation; by 2026 it had concentrated in troubleshooting and debugging. Our data suggest an ownership paradox, where AI appears to improve technical self-efficacy while potentially reducing the productive difficulty associated with creative ownership and the desirable difficulties required for deep learning. Recommendations are offered on how game jam organizers can design events around AI presence to preserve the educational value of the creative struggle.