Man I hope this is true, the Linux kernel is such an important piece of our technological infrastructure currently, I don't want it to become slopware.

I think you are right here and this is why I was wondering about the practicality of it. I know there exists other variants of the Linux kernel like Zen but I imagine these are small patches, removing all AI code from the kernel and then having to maintain it would be such an enormous effort. And then you also have the problem of detecting AI code among all the contributions, that can't be easy either.

There are reasons why professionnal rally driver re-train to drive on ice/... before _each_ season.

Do mainly pr review instead of design/code/debug, and your proficiency will slowly erode.

And the best: you could not even detect it without any, objective, external metrics.

@argv_minus_one @nixCraft

> Or just walk away

Sounds like a reference to a Cory Doctorow's "Walkaway" book

BTW strongly recommend to read it!

> In a world of non-work, ruined by human-created climate change and pollution, and where people are under surveillance and ruled over by a mega-rich elite, Hubert, Etc., his friend Seth, and Natalie, decide that they have nothing to lose by turning their backs and walking away from the everyday world of "default reality" (https://en.wikipedia.org/wiki/Walkaway_(Doctorow_novel))

Are they, though, in any meaningful sense? I've heard many complaints that people who use LLMs lose their programming skills, which presumably makes them unqualified to review PRs.

@nixCraft

I following and watching this repo for a long time and the more time passes — the more it reminds me a transcript of the congress of the Central Committee of the CPSU half-mixed with radicalism. Not just a well-structured list like https://noai.starlightnet.work/list.html

So, it wasn't surprise when some OSes ( at least) were included in this list just because they required core approval for LLM commits (reading between the lines: core will not give this approval and as we see from the recently updated guidelines — finally, such commits were clearly prohibited).

@radicalabacus @nixCraft @loadhigh @faraiwe @Stefan_S_from_H @geospacedman @marcusxms

My worry is that their use of AI is going to *create* security issues for bad actors to exploit.

@nixCraft

If Kernel devs find it useful to discover security issues and fix them before a bad actor exploits them, then you can't discount the benefit it provides. Personally I'd put in the same class of tools a fuzzing etc, but far more expensive and advanced.