Author: Gregory Charles Tiong
Pages: 25–30
Debugging is an essential skill for every programmer, but is overlooked, time-consuming, and prone to error tasks. While various debugging tools like IDE debuggers and Automatic Program Repair (APR) aim to streamline the process, these are frequently challenging for experienced developers and novices alike to navigate. The rise of LLMs and introduction of ChatGPT in 2022 marked a significant shift in this field, with researchers finding that it performs comparably or even surpasses APR tools. In addition, its conversational interface is user-friendly and intuitive, making it a preferred alternative to more traditional interfaces. However, LLMs tend to struggle as the project grows, making them an unsustainable option in the long term. This research will focus on leveraging LLMs to support novices by teaching them debugging skills and guiding them in using debuggers effectively, acknowledging that humans still hold a clear advantage over automatic debugging tools in complex projects.
Proceedings of CHIRP 2025, May 8 2025, Baguio, Benguet