"TODO: Fix the Mess Gemini Created": Towards Understanding GenAI-Induced Self-Admitted Technical Debt
Abdullah Al Mujahid, Mia Mohammad Imran
2026-01-13
Summary
This paper investigates how developers are acknowledging the use of AI tools like ChatGPT in their code, and specifically looks at instances where they admit the AI-generated code isn't perfect and creates future work for them.
What's the problem?
As developers start using AI to help write code, they often leave comments explaining what they did. The researchers noticed that sometimes these comments not only mention using AI, but also admit that the code might have problems or need to be fixed later – essentially creating 'technical debt'. The core issue is understanding *why* and *when* this happens when AI is involved, and what it means for the quality of software being developed.
What's the solution?
The researchers analyzed over 6,500 code comments from public projects on GitHub that mentioned AI. They specifically searched for comments that *also* admitted to having technical debt. They found 81 such comments and categorized the reasons developers gave for this debt, like needing to do more testing, not fully understanding the AI's code, or knowing they'd need to adjust it later. They then proposed a new term, 'GenAI-Induced Self-admitted Technical debt' (GIST), to describe this specific situation.
Why it matters?
This research is important because it highlights a potential downside to using AI in coding. While AI can speed up development, it can also lead to shortcuts and code that isn't fully tested or understood, creating problems down the road. Understanding this 'GIST' phenomenon can help developers and teams make informed decisions about how to best integrate AI into their workflows and manage the resulting technical debt.
Abstract
As large language models (LLMs) such as ChatGPT, Copilot, Claude, and Gemini become integrated into software development workflows, developers increasingly leave traces of AI involvement in their code comments. Among these, some comments explicitly acknowledge both the use of generative AI and the presence of technical shortcomings. Analyzing 6,540 LLM-referencing code comments from public Python and JavaScript-based GitHub repositories (November 2022-July 2025), we identified 81 that also self-admit technical debt(SATD). Developers most often describe postponed testing, incomplete adaptation, and limited understanding of AI-generated code, suggesting that AI assistance affects both when and why technical debt emerges. We term GenAI-Induced Self-admitted Technical debt (GIST) as a proposed conceptual lens to describe recurring cases where developers incorporate AI-generated code while explicitly expressing uncertainty about its behavior or correctness.