The Linux kernel has always been focused on outcomes and outputs, which is the right focus. Here's Linus: https://x.com/LundukeJournal/status/2110524360377458786
Outputs: Code quality and long term maintenance matters. Outputs: Serving users with correctness, security and performance matters.
Focus on "purity of input" is a faith-based political filter, not engineering.
Someone should ask him if he's also planning on no longer reviewing patches to the kernel and just accepting them unconditionally, since AI is no different from a compiler after all, and nobody reviews compiler outputs.
After the SCO mess, I'm very shocked he didn't reject AI out of hand simply for the provenance and liability issues.
Yes, and that take is bad, because AI is not a tool in the traditional sense. In the software world, tools are generally expected to be deterministic. Would Linus use a compiler that is known to have a random chance to output the wrong instructions? Or an editor that has a random chance to swap something you typed for something else?
> compiler outputs are tested by the compiler projects + the automated testing infra
And AI outputs are tested by the AI companies and the various testing suites that are run on the kernel. If they're really the same like Linus seems to believe, there's no reason to ever read AI generated patches before merging them.
Says who?