GitHub addresses surge in low-quality AI contributions

The Register reports that GitHub is facing significant challenges due to an increasing volume of low-quality contributions, many of which are AI-generated, impacting the open-source community and creating operational burdens for project maintainers.

Product manager Camilla Moraes initiated a community discussion highlighting that developers are spending substantial time reviewing submissions that fail to meet project standards, are often abandoned, or are AI-generated. Maintainers report that only a small fraction of AI-created pull requests (PRs) are legitimate. Some projects, like curl, have even shut down bug bounty programs to disincentivize low-quality reports. Concerns include a broken review trust model, where reviewers cannot assume code authorship, and the difficulty of scaling line-by-line reviews for large AI-assisted PRs. This increases cognitive load and review burden, as maintainers must now assess both the code and the author's understanding.

GitHub is exploring solutions such as disabling PRs, restricting them to collaborators, or implementing AI-based triage tools and transparency mechanisms for AI usage. The influx of AI-generated content raises questions about preserving community incentives for participation and knowledge sharing. Without clear disclosure of AI use, there's a risk of eroding social trust within the open-source community, potentially undermining the collaborative spirit that defines platforms like GitHub, developers said.

Source: The Register