An Exploratory Case Study of LLM-Assisted Refactoring and Gameplay Feature Generation in an Endless Runner Game

This paper presents an exploratory case study utilizing GPT-4o for software development tasks in a Python/Pygame endless runner game, focusing on localized refactoring and gameplay feature generation. The study found that while all localized refactoring tasks were successfully implemented, only one of three gameplay feature generation tasks was correctly integrated, indicating that GPT-4o performs better with localized transformations than with complex interactions across existing systems. These findings highlight the potential and limitations of using LLMs in game development, providing insights for practitioners on the effective application of AI in software engineering contexts.