Linus Torvalds: The AI Slop Issue Is *Not* Going to Be Solved with Documentation(phoronix.com)
phoronix.com
Linus Torvalds: The AI Slop Issue Is *Not* Going to Be Solved with Documentation
https://www.phoronix.com/news/Torvalds-Linux-Kernel-AI-Slop
2 comments
> So was the proposed idea be that people would be required to disclose if AI was used to generate code?
Yes. The proposal claimed that there will be too much AI slop, and it will inundate the review process with too much work and bring the entire system to a grinding halt, or worse, end up allowing code that lacks proper review into the kernel.
Linus basically said, "It might become a problem, but the bad guys are never going to admit they used AI, so it's not helpful to demand disclosure".
Yes. The proposal claimed that there will be too much AI slop, and it will inundate the review process with too much work and bring the entire system to a grinding halt, or worse, end up allowing code that lacks proper review into the kernel.
Linus basically said, "It might become a problem, but the bad guys are never going to admit they used AI, so it's not helpful to demand disclosure".
I do agree that there’s no magic in requiring disclosure here as plenty of people simply won’t and we’re in the same situation regardless.