Debian Linux, one of the oldest Linux distributions, which emerged in 1993, has decided to incorporate modern technologies into its project. Recently, community members debated and voted in favor of allowing the use of generative Artificial Intelligence in the operating system's development, provided that it is done responsibly.
After weeks of deliberations, the vote concluded last Friday, August 28th, at 23:59. Eight suggestions were presented, all focused on determining whether Large Language Models (LLMs) could be employed in the development of the Linux distribution.
This discussion arises because, similar to other software projects, Debian already utilized collaborations with tools like Claude or GitHub Copilot, but lacked a clear and specific guideline regarding this working methodology.
The arguments were divided: some advocated for using generative AI to increase productivity or accelerate problem-solving, while others opposed it due to concerns about the quality, security, and reliability of the produced code.
To resolve the deadlock, a vote was held starting on August 15th. Proposal number 5, titled 'Responsible Use of Generative AI,' prevailed. This vote followed the Condorcet method, where the winner is the one that surpasses all other options in direct confrontations, which was the case for proposal 5.
In essence, the approved proposal authorizes the use of generative AI in the development of Debian Linux, but imposes the condition that the developer who uses such a resource must understand, review, test, and, if necessary, modify the generated code.
This establishes that the developer retains full responsibility for the AI-generated code, just as they would if all the work were done by them, moving away from the idea of a simple assisted coding policy.
A section of proposal 5 states that such tools can significantly boost collaborators' productivity when used responsibly, allowing volunteers to dedicate more time to tasks requiring technical knowledge, discernment, review, and collaboration. However, the Debian Project requires that all submitted contributions meet the same standards of quality, correctness, maintainability, and legal compliance, regardless of the tool used.
Although the vote seems fair, not everyone agreed with the outcome. Developer Antoine Le Gonidec expressed his disagreement with the use of LLMs in Debian and, for this reason, chose to leave the project. Despite the disagreements, the 'father of Linux' supports the responsible use of artificial intelligence, evidenced by the fact that Linus Torvalds recently used AI to fix a bug in Linux.