And no, that's not the position of the Linux kernel.
I realize that some people really dislike AI, but this is an area
where I'm willing to absolutely put my foot down as the top-level
maintainer.
Linux is not one of those anti-AI projects, and if somebody has issues
with that, they can do the open-source thing and fork it.
Or just walk away.
AI is a tool, just like other tools we use. And it's clearly a useful one.
It may not have been that "clearly" even just a year ago, but it's no
longer in question today.
There are other questions around AI (like what the economy of it will
actually look like in the end), but "is it useful" is no longer one of
those questions. Anybody who doubts that clearly hasn't actually used
it.
Yes, it can also be a somewhat painful tool, both for maintainer
workloads and just from a "it keeps finding embarrassing bugs"
standpoint.
But the solution is not to put your head in the sand and sing "La La
La, I can't hear you" at the top of your voice like some people seem
to do.
The solution is to make sure those LLM tools help maintainers
instead of just causing them pain. There's no question on that side.
We're not forcing anybody to use it, but I will very loudly ignore
people who try to argue against other people from using it.
And no, AI isn't perfect. But Christ, anybody who points to the
problems at AI had better be looking in the mirror and pointing at
themselves at the same time.
Because it's not like natural intelligence is always all that great either.
The kernel project has been and will continue to be about the technology.
Sure, the social angle of working on open source is important and
often a very motivating part of the project, but in the end that's a
side benefit, not the point of the project.
This is NOT some kind of "social warrior" project, never has been,
and never will be.
In the kernel community we do open source because it results in better
technology, not because of religious reasons.
And so we make decisions primarily based on technical merit. Not fear
of new tools.
Linus is 100% right , there will always be a subset of the population that dislikes the negatives of technology advances but you can’t be a main player in software today, while still doing 100%...
Linus is 100% right , there will always be a subset of the population that dislikes the negatives of technology advances but you can’t be a main player in software today, while still doing 100% human written code. You can try but to bet the main Linux kernel on artisanal code is crazy.
It's a bit like food production. Sure, you can try to ethically source all of the food at a restaurant... but basically the only way to do that is to run a vegan restaurant. And even then, are you...
It's a bit like food production. Sure, you can try to ethically source all of the food at a restaurant... but basically the only way to do that is to run a vegan restaurant. And even then, are you absolutely sure that nobody was exploited to produce the fertilizer to grow the veggies? And that nobody committed genocide to get that farmland? (spoiler alert: someone definitely did).
I think it makes sense to not let perfect be the enemy of good here, even if I dislike LLMs personally. Let's shame people for slop, but admit that LLMs are a tool that many find very helpful for software development.
100% this. It’d be foolish for the kernel project to outright exclude all LLM contribution based on origin alone, as that necessarily would include bug findings, not just novel code itself. Guess...
I think it makes sense to not let perfect be the enemy of good here, […] admit that LLMs are a tool that many find very helpful for software development.
100% this. It’d be foolish for the kernel project to outright exclude all LLM contribution based on origin alone, as that necessarily would include bug findings, not just novel code itself.
Guess who doesn’t not use LLMs out of morality/principle? Adversarial actors trying to exploit the kernel, finding issues at alarmingly increasing rates already today.
I really wish we (everyone) would discern what type of AI programming we're talking about. There's two types: vibe coding, and AI assisted. There's nothing wrong with AI assisted code, and in...
I really wish we (everyone) would discern what type of AI programming we're talking about. There's two types: vibe coding, and AI assisted.
There's nothing wrong with AI assisted code, and in fact, will make you a better programmer if used properly. It should be treated like stack overflow, and every line of code should be modified/reviewed/approved before implementing.
Vibe coding is the real, objective, problem and I wish it was talked about more. I've seen it first hand. People who let the LLM take the wheel and go all in on agentic programming. Who think they can build a web service for easy money, but have 0 idea what they're doing. Blindly trusting the output with no way to verify its accuracy. LLMs will always have a hallucination problem, and I won't be convinced otherwise. They literally always hallucinate, they just happen to be correct a lot of the time.
Everyone gets wrapped up in talking about "AI bad", but IMO, it's the wrong conversation. As the end user, you really have to pay attention to what kind of services you input your personal information into. It was important before, but now more than ever.
Edit: I'm also not sure wtf an anti-ai stance means in practice. You could mean being against 100% generated content, which is fine. But that term alone implies you can't use AI to help you. It makes me feel like I'm doing something wrong by using AI for assistance, but do they really want to see the code I wrote by myself? I can assure you that is significantly worse.
I didn't read the whole mail chain, but in this case I think the context is an AI tool for finding bugs in the code, and whether that should be applied automatically to commits, or something...
I didn't read the whole mail chain, but in this case I think the context is an AI tool for finding bugs in the code, and whether that should be applied automatically to commits, or something similar. From what I understand it's not even about writing code. Linus is being kinda vague though, so I assume it can also be interpreted as "use any (ai) tools, as long as they improve the outcome".
Yeah, that's essentially my point. Like, I understand what he's saying and I agree with him. But I'm concerned the vibe coders will take that and run with it because the Linux Jesus said so.
Yeah, that's essentially my point. Like, I understand what he's saying and I agree with him. But I'm concerned the vibe coders will take that and run with it because the Linux Jesus said so.
In your mind, what is the scope of "AI-assisted" programming? For example, if I write an AGENTS.md file and spin up some agents and point it at an issue, and the agents generate an MR to close...
In your mind, what is the scope of "AI-assisted" programming?
For example, if I write an AGENTS.md file and spin up some agents and point it at an issue, and the agents generate an MR to close that issue with a bunch of modified files, and I look over the MR and think it looks fine and merge it, would you consider that to be AI assisted coding, or closer to vibe coding?
Actual mailing list email is here: https://lore.kernel.org/linux-media/CAHk-=wi4zC+Ze8e+p3tMv8TtG_80KzsZ1syL9anBtmEh5Z40vg@mail.gmail.com/
Linus is 100% right , there will always be a subset of the population that dislikes the negatives of technology advances but you can’t be a main player in software today, while still doing 100% human written code. You can try but to bet the main Linux kernel on artisanal code is crazy.
"artisinal code"
...i guess that's a thing, now...
It's a bit like food production. Sure, you can try to ethically source all of the food at a restaurant... but basically the only way to do that is to run a vegan restaurant. And even then, are you absolutely sure that nobody was exploited to produce the fertilizer to grow the veggies? And that nobody committed genocide to get that farmland? (spoiler alert: someone definitely did).
I think it makes sense to not let perfect be the enemy of good here, even if I dislike LLMs personally. Let's shame people for slop, but admit that LLMs are a tool that many find very helpful for software development.
100% this. It’d be foolish for the kernel project to outright exclude all LLM contribution based on origin alone, as that necessarily would include bug findings, not just novel code itself.
Guess who doesn’t not use LLMs out of morality/principle? Adversarial actors trying to exploit the kernel, finding issues at alarmingly increasing rates already today.
I really wish we (everyone) would discern what type of AI programming we're talking about. There's two types: vibe coding, and AI assisted.
There's nothing wrong with AI assisted code, and in fact, will make you a better programmer if used properly. It should be treated like stack overflow, and every line of code should be modified/reviewed/approved before implementing.
Vibe coding is the real, objective, problem and I wish it was talked about more. I've seen it first hand. People who let the LLM take the wheel and go all in on agentic programming. Who think they can build a web service for easy money, but have 0 idea what they're doing. Blindly trusting the output with no way to verify its accuracy. LLMs will always have a hallucination problem, and I won't be convinced otherwise. They literally always hallucinate, they just happen to be correct a lot of the time.
Everyone gets wrapped up in talking about "AI bad", but IMO, it's the wrong conversation. As the end user, you really have to pay attention to what kind of services you input your personal information into. It was important before, but now more than ever.
Edit: I'm also not sure wtf an anti-ai stance means in practice. You could mean being against 100% generated content, which is fine. But that term alone implies you can't use AI to help you. It makes me feel like I'm doing something wrong by using AI for assistance, but do they really want to see the code I wrote by myself? I can assure you that is significantly worse.
I didn't read the whole mail chain, but in this case I think the context is an AI tool for finding bugs in the code, and whether that should be applied automatically to commits, or something similar. From what I understand it's not even about writing code. Linus is being kinda vague though, so I assume it can also be interpreted as "use any (ai) tools, as long as they improve the outcome".
This is the message that I gathered the "AI as reviewing tool" from https://lore.kernel.org/linux-media/20260715125058.3692aa03@gandalf.local.home/. The tool itself is called Sashiko.
Yeah, that's essentially my point. Like, I understand what he's saying and I agree with him. But I'm concerned the vibe coders will take that and run with it because the Linux Jesus said so.
I can see your point, I'm not sure it'll change much in practice though. From what I see they already feel pretty justified.
In your mind, what is the scope of "AI-assisted" programming?
For example, if I write an AGENTS.md file and spin up some agents and point it at an issue, and the agents generate an MR to close that issue with a bunch of modified files, and I look over the MR and think it looks fine and merge it, would you consider that to be AI assisted coding, or closer to vibe coding?