My Teammate Is an AI: Evaluating Generative AI in Game Asset Creation through a Post-GameJam Study

A post-game-jam study of 37 mostly beginner computer science students examines what happens when generative AI tools such as ChatGPT, DALL-E, and Bing AI take the place of graphic design teammates in creating game assets. Framed within broader debates over labour displacement, artistic agency, and the ethics of training data, the research uses a mixed-methods design combining Likert-scale and open-ended survey responses to assess how generative AI shapes creative potential, team dynamics, and learning in a time-constrained development setting. The study's central thesis is that generative AI is currently better suited to supplementing human artistry than replacing it.

The findings point to a clear division of strengths and weaknesses. AI accelerated routine asset production, such as backgrounds and menus, and enabled hybrid roles and rapid prototyping. However, participants rated AI-generated assets slightly below human-made equivalents overall (48.6%), and most judged the tools unreliable (67.5% rated them somewhat or very unreliable). Creative impact was mixed, with 40.5% reporting enhancement and 21.6% reporting hindrance, while limitations in customisation and creative consistency persisted. Team communication and role clarity remained strong, and nearly half of participants (45.9%) reported substantial learning gains, though many struggled with the steep learning curve of prompt crafting. Most participants were willing to adopt the tools in future projects.

The authors draw implications for both industry and education, calling for responses to labour displacement, AI literacy, cultural bias, and intellectual property. They acknowledge important limitations: the small sample, the educational rather than commercial context, and the absence of professional artists on the teams. These constraints limit how far the results generalize to professional studio practice, though the study offers a grounded early picture of how generative AI functions within collaborative, time-pressured game development.