Zig president says AI coding contributions are 'invariably garbage,' so he banned them

The programming language Zig has taken a definitive stand against AI-generated code, enforcing a strict ban on contributions that utilize artificial intelligence. Managed by a 501(c)(3) organization and a collaborative network of developers, Zig allows anyone to submit code to its repository, provided they adhere to a specific code of conduct. A key rule within this framework is the prohibition of any AI-assisted code submissions. Zig's President, Andrew Kelley, voiced strong opinions on this matter during a recent episode of the JetBrains podcast, labeling AI-assisted contributions as "invariably garbage." He expressed concern that many submissions lack genuine value, stating, "They have negative value, because they take review time away from the team." With a limited number of core team members available to review contributions, Kelley highlighted a significant bottleneck: there are currently over 200 open pull requests pending review. Kelley criticized the influx of what he referred to as "slop contributions" generated by AI, which he believes further delays the review process and wastes valuable time for everyone involved. While Zig may be a smaller player in the programming world, its influence is noteworthy; the language has been instrumental in the development of tools such as Bun, which was later acquired by Anthropic. The ban on AI contributions has ignited discussions within the tech community, particularly as AI-assisted coding tools like Claude Code and OpenAI's Codex gain traction in Silicon Valley. Many developers are now using AI to draft, edit, or modify their code, with major tech companies setting ambitious goals for the proportion of code that should be AI-generated. However, Zig does not operate under the same efficiency-driven mandates as larger corporations. Kelley emphasized that the focus on mentorship is integral to Zig's mission, making AI contributions counterproductive to the goal of improving programming skills. "We’re all trying to get better at programming," he noted, adding that those submitting AI-generated pull requests are unlikely to contribute meaningfully to the core team. To maintain simplicity in policy enforcement, Kelley explained that a blanket ban on all AI contributions is easier to manage than attempting to evaluate the quality of individual submissions. This decisive approach may set Zig apart in the ongoing debate over the role of AI in software development.