Zig creator bans AI tools over code quality and reliability
The creator of the programming language Zig has rejected contributions from artificial intelligence tools, claiming they lower code quality and waste time. Andrew Kelley also relocated the language’s repository from GitHub to Codeberg, citing reliability failures and a mismatch with the project’s goals.
Kelley’s position rests on two main arguments. First, AI-generated submissions consistently produce poor-quality code. In an interview, he stated, “And at some point we realize this is never going to be good quality because they have no idea what they’re doing.” This approach consumes review efforts without delivering useful progress, he explained.
The second concern involves Zig’s emphasis on mentorship. The project uses a system called contributor poker to direct limited review time toward developers who demonstrate potential for long-term involvement. AI-assisted contributors, Kelley argued, do not benefit from this process. They’re not going to join the core team later.
Kelley contrasted Zig’s philosophy with that of Bun, which recently rewrote its codebase from Zig to Rust using AI tools. While the change was presented as a technical decision, Kelley dismissed it as a difference in values. In a rather critical reaction titled “My Thoughts on the Bun Rust Rewrite,” Kelley affirmed: The main issue here had nothing to do with the language features of Zig vs Rust, and everything to do with the diverging value systems of the two projects.
The shift away from GitHub stemmed from repeated failures in the platform’s continuous integration system. “GitHub simply stopped working for us,” Kelley noted. “We would not have results for our continuous integration runs anymore.” The project now uses Codeberg, a German nonprofit, which Kelley described as more dependable than corporate-run alternatives. “Nonprofits focus on stability,” he said. “That’s what matters to me.”
Zig’s design intentionally avoids automatic memory management, a choice rooted in Kelley’s early frustrations with other languages. While developing a digital audio workstation, he found JavaScript lacked low-level control, Go introduced unacceptable delays, Rust’s borrow checker hindered progress, and C++ generated memory corruption errors. Zig’s explicit memory model was created to address these problems.
Since its 2018 launch, Zig has been used by projects like Ghostty, a low-latency financial database, and TigerBeetle, as well as Uber’s cross-compilation tools. The language’s focus on performance and predictability reflects Kelley’s original objectives: minimizing unexpected behavior in hardware interactions and runtime operations.
