Rémi Verschelde, one of the key maintainers of the open-source game engine Godot and co-founder of W4 Games, stated that the increasing influx of low-quality pull requests generated by large language models is draining and demoralizing the Godot developers reviewing the changes. According to Rémi, the Godot project prides itself on its openness to new contributors and allows any user to impact the development of the engine. The maintainers of Godot spend a lot of time helping to get the pull requests into proper shape, but with the current abundance of questionable changes, Rémi is unsure how much longer the team can endure.
Currently, there are over 4,600 open pull requests accumulated in the Godot repository on GitHub. Adriaan de Jongh, director of the game development studio Hidden Folks, described the situation with AI-generated junk pull requests for Godot as a complete mess and a massive waste of time for reviewers who are forced to spend time sorting through changes that often make little sense, are filled with overly verbose descriptions, and are submitted by developers who do not understand their own patches.
With the emergence of AI tools, maintainers are additionally required to determine whether the code was written by a human or generated automatically, as well as to assess whether the author understands the code they submitted. They also need to figure out whether the authors tested their code or if the presented test results are merely fabricated by AI. Furthermore, Rémi does not understand how to handle situations where, when asked about the use of AI, patch authors respond, 'I only used it for the description because I write poorly in English,' which complicates the classification of contributions.
It has been noted that maintainers have to double-check pull requests from new contributors multiple times a day, questioning their quality. Rémi believes that the only solution to maintain the previous approach to development is to increase funding for the project to pay additional maintainers who will review the AI-generated submissions.
The Blender project is facing similar difficulties, having begun in early February to develop a policy regarding patches created using AI. Previously, rules for such cases were adopted by Linux Foundation, Fedora, GNOME, Firefox, Ghostty, Servo, and LLVM. In January, GitHub initiated discussions on implementing measures to limit low-quality pull requests generated by AI assistants that are submitted without manual review, do not meet quality standards, and create serious challenges for maintainers.
Source: opennet.ru
