The Register’s Steven J. Vaughan-Nichols has a column out today on Linus Torvalds and AI, and it lines up nicely with the Linux 7.4 and Mesa stories I covered earlier this week.
What the column says
- Some people reacted to Torvalds embracing AI for kernel code with “slop” complaints. The author, who has known Linus for decades, doesn’t see it as abandoning his standards.
- At the Linux Kernel Maintainers Summit in Prague, kernel developer Sasha Levin announced that maintainers had decided to “deprecate the Assisted-by Tag.” You no longer need to say whether you used AI. The code still has to pass review, so slop won’t get in.
- Torvalds, in a note to the author: “I really think we want competent people to use AI as a tool.” The problem is “clueless people using AI,” and he wants to discourage “the drive-by patches” and people “who just want to feel like they sent a patch and never really wanted to become anything more.”
- At Open Source Summit Europe he called vibe coding “a wonderful way to find joy in programming” and said he loves “AI as a gateway drug” to programming.
My take
I think that’s a sensible line. Judge the patch, not the tool. A competent person who reads, tests and owns the result is fine, and someone who fires off code they don’t understand isn’t, with or without AI. That’s how I think about my own agent workflows too: the bot drafts, a human is still responsible for what ships.
Have a blessed day! And Happy Hacking!
References & Further Reading
- Torvalds: AI is fine for competent devs, but spare us the ‘drive-by’ patches (The Register)
- AI Coding Assistants (Linux kernel documentation)
- Linus Kills the Unsafe __get_user() and __put_user() in Linux 7.4 (EtherNinja)
- Mesa Loosens Its AGENTS.md Rules for AI-Drafted Comments and Commits (EtherNinja)
