Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft I can’t find the post but someone pointed out recently that LLMs are not like compilers. With a compiler you generate a disposable artifact that you can reliably recreate from the input source. With LLMs you discard the input prompt and keep the output, and running the same prompt twice will generate different results. This is an infuriating stance given how much of the last decade I’ve dedicated to making reliable deterministic build pipelines.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft there's hope that Claude suggests rewriting the kernel in C++.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft "nobody in their right mind"... what happened to the ring of maintainers?
well, then itll take the time it takes to put in the code... seriously, just because youre flooded doesnt always mean youll have to open the flood gates...
i remember linus saying that someday linux will become obsolete, so we are steering right into that i guess
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft Two things. First, most of the time, discovering problems are easier than building complex systems, and the same goes for ML/AI models, of course. Second, ML/AI is "bound" to make mistakes, and ML/AI scientists have been saying for years that you need mistakes for ML/AI to be useful (because otherwise, it means overfitting). I understand that ML/AI tools can be helpful for productivity, but it's guaranteed that it'll make mistakes (if not intentional).
-
@nixCraft Two things. First, most of the time, discovering problems are easier than building complex systems, and the same goes for ML/AI models, of course. Second, ML/AI is "bound" to make mistakes, and ML/AI scientists have been saying for years that you need mistakes for ML/AI to be useful (because otherwise, it means overfitting). I understand that ML/AI tools can be helpful for productivity, but it's guaranteed that it'll make mistakes (if not intentional).
@nixCraft What's the measurement against them? Is there any attempt to (officially) verify the kernel code with ML/AI, Syzbot style, not just for development? I assume ML/AI tools will soon plant some complex mistakes that are difficult for overloaded human reviewers to find out.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft Comparing deterministic tools to #AI is wild.
Yes, I would create rules about not using compilers & editors if they gave different output each time I used them, occasionally gave wrong output, sometimes gave SLIGHTLY wrong output, and were just running predictions based on my past inputs rather than following sets of rules.
-
Just so you know. Linus who is adopting AI tools. Here is how it is going right now. Linus Torvalds let Claude Code fix the Linux page cache. Then reviewers found a race https://thenewstack.io/torvalds-claude-code-kernel/
@nixCraft
This is why I'm looking at NetBSD and Plan9 (9front) because the use of AI slop in Linux woke me up to the fact that the whole ecosystem is built around profits for corporations rather than good code -
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft Uf what a bad take. Not all tools are created equal.
“No sharp knives in here because they can be misused for violence.”
“Oh, so I guess we can’t use straws or spoons either?! For that matter, this food shouldn’t be on a plate!”
-
@nixCraft Idk how Linus uses AI. I hate AI, but sometimes is kinda useful ONLY IF it's used locally hosted, like with Ollama.
I'm more concerned about environmental impact than stealing knowledge, although both are major concerns
AFAIK he at least slop generates python code with LLMs and confuses gen AI to be some kind of tool. How things (apparent psychosis) are going i suspect he frequently converses with the machine, too.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft I may be wrong about that but it seems to me that the debate shouldn't be about whether an AI wrote the code or not but about code quality and whether trusted humans validate it. If your contribution goes in a queue that is reviewed at human pace and you have a storage quota for your contributions there's no way slop of any origin get into mainstream. AI or not AI becomes irrelevant. But maybe I'm overlooking something.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft I love it when techbros begin their argument with "I don't understand" as a rhetorical stance because it's always true in a way: they don't understand.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft It should be easy to maintain stats on the slop producers regardless of origins and slow the worst offenders even further somehow. There's always a human motivation behind any AI production. That shouldn't be too hard to control I think.
-
Just so you know. Linus who is adopting AI tools. Here is how it is going right now. Linus Torvalds let Claude Code fix the Linux page cache. Then reviewers found a race https://thenewstack.io/torvalds-claude-code-kernel/
@nixCraft But also… “Amanda Caswell is an AI journalist, certified prompt engineer, and technology commentator whose work and expertise have been featured on Fox News and CBS News […] TNS owner Insight Partners is an investor in: Anthropic”
(and the article is sprinkled with weird ambiguities. Wouldn’t surprise me if it was LLM-summarized or LLM-rewritten)
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft Man, that last paragraph, what a nasty guy.
My respect goes to the maintainers who manage to handle working with an asshole like this. I know I wouldn't be able to do it.
-
@nixCraft Man, that last paragraph, what a nasty guy.
My respect goes to the maintainers who manage to handle working with an asshole like this. I know I wouldn't be able to do it.
-
@nixCraft But also… “Amanda Caswell is an AI journalist, certified prompt engineer, and technology commentator whose work and expertise have been featured on Fox News and CBS News […] TNS owner Insight Partners is an investor in: Anthropic”
(and the article is sprinkled with weird ambiguities. Wouldn’t surprise me if it was LLM-summarized or LLM-rewritten)
@nixCraft Anyway, apparently LWN also (very briefly) covered this: https://lwn.net/Articles/1100449/
-
@nixCraft Comparing deterministic tools to #AI is wild.
Yes, I would create rules about not using compilers & editors if they gave different output each time I used them, occasionally gave wrong output, sometimes gave SLIGHTLY wrong output, and were just running predictions based on my past inputs rather than following sets of rules.
@syntaxseed @nixCraft
I agree...and I most definitely would have rules for non-deterministic tools. Those were two inappropriate analogies. -
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom now@nixCraft This post is actually a pretty good reply to that, even if it is on LinkedIn... https://www.linkedin.com/pulse/ai-tool-rich-hickey-kjl8e
Use the tools you want - but don't use automation where your brain should do the job.
-
Linus Torvalds shares his thoughts on Anti-AI coding rules on Linux mailing list.
https://lore.kernel.org/lkml/CAHk-=whTs09_ozN+e09GU68N1wPP56MUzDjrg9rnKxQUa0A_4Q@mail.gmail.com/
He is doubling down on AI. This will come back to bite him. He thinks people using LLMs for coding are actually checking their work
Code reviewers are just rubber stamping PRs. Nobody in their right mind can look at 100s of PRs per day &d manually go through them especially since they don't even know if they'll have a IT job left at the end of the day. It's a race to the bottom nowSomeone needs to explain logical fallacies to Linus because he doesn't seem to understand False Equivalence and Straw Man, among others 🫤
️ -
@epistomai there are better tools than ollama for local AI. https://sleepingrobots.com/dreams/stop-using-ollama/ explains why.
@__jm It is shady, yes, but there are also locally models, not only cloud. I've learned a lot how to 'bubble' mine. Still, I know there are other tools, but are more convoluted to me.
I only use it because I am forced to right now, once I get rid of that nonsense I will
drop AI in any form
