Work

Using Generative AI for Teaching Game Development Using Unity

Year: 2024
Type: dissertation
Abstract: Game development education’s interdisciplinary and time-intensive nature presents several challenges for students. These challenges include insufficient and untimely feedback, steep learning curves in mastering development tools and concepts, and insufficient project planning and communication skills. Generative Artificial Intelligence offers potential solutions to these challenges. While research indicates that Generative AI can effectively address individual problems in fundamental computer science education, there is a risk of over-reliance on such technologies. Generative AI may be inadequate for resolving long-term, complex challenges. This thesis introduces a Generative AI-based game development teaching tool through a project-based teaching mode. We divided the game development process into four phases: Design, Asset Creation, Programming, and Testing. To support this, we developed a multiagent teaching system with four AI tutor agents, each specializing in one of the phases and a supervisory agent. The AI tutor agents provide tailored guidance and feedback through timely interventions. They can perform web searches and access local knowledge base to offer upto-date information and explanations. The supervisory agent ensures overall supervision of teaching quality and progress. Additionally, we have implemented an innovative multi-agent memory system to enhance the AI tutor agent’s performance in addressing long-term, complex challenges. An anonymous user study involving 22 students indicated that our teaching approach increased participants’ motivation and the quality of their game projects compared to text-based learning methods. However, no significant differences in motivation and learning effectiveness were observed among participants with different personality traits and game development backgrounds. Most participants were satisfied with their final game projects and willing to continue using this teaching tool. This study highlights the potential of Generative AI in game development education and provides insights for the integration into future curricula.
Author: Li, Shaojun
Language: English
Open Access status: green
OpenAlex ID: W7053252978
Cites:
Cited by: 2
Related to: 10