Post by @AndrewRadev@hachyderm.io

In case you were wondering how the linux kernel is doing: https://lore.kernel.org/lkml/20260818212923.2725202-1-kuba@kernel.org/

> This is not telling the full story either because 1/3 to 1/2 of the net-next patches also *seem* like AI-driven low priority fixes, cleanups and clarifications.
> We are completely overwhelmed, of course. The glimmer of hope is that we secured sufficient LLM budget and access (thank you Meta!) to run reviews with multiple frontier models on each patch. This eliminates some hallucinations.

That's right, "Thank you Meta!", this kernel maintainer enthusiastically exclaims. Presumably, for subsidizing their slop machines trained on pirated books with money from online fraud. Which eliminates "some" hallucinations. An incredible achievement, I'm sure.

View

1 Reply

1 Boost

Andrew Radev

@AndrewRadev@hachyderm.io 3 hr. ago

Do you think that this kernel developer who describes his team as "completely overwhelmed" is capable of carefully reviewing and fixing hallucinations, which he is aware are present? Do you think he and his team are even trying? Do you think that the plagiarism machine trained on all insecure code on the planet is maybe going to occasionally squeeze in some well-hidden bugs and/or security issues in those "low priority fixes, cleanups and clarifications"?

I think I'm going to avoid updating my kernel for a long while. At this point, I have a hard time guessing if an update is going to fix security bugs or add worse ones.

bonfire.cafe