Linux

I had to admit it, but he’s right. I’m no fan of using AI either, but there are times and places where it can actually be useful when used correctly. What’s really important is the quality of the contributions made to the project, and the usefulness should be judged on its merits alone of those individual contributions, not the “icky“ feeling people get from the idea of using AI at all.

Torvalds is clearly not here defending “vibe coding”, and he’s perfectly happy for people to not use AI at all. But when AI, as a tool, is used to responsibly, it can clearly be quite useful. And for the coders who can use it in a responsible and useful way to make meaningful contributions, I also think they should be permitted to.

AI is not the problem. It's the people and companies behind the popular public AI services that are the problem.

I used to see anti-AI people as myopic and foolish, but seeing how it's being developed and implemented is really eye-opening. They're pushing an unfinished, unrefined technology that often adds to the ecological crisis and might be poisoning people's drinking water, so they can roll the dice in terms of what it'll do to the economy while using it to spy on everyone. The public face of all this is these companies telling us that we should be terrified of what it'll do to the job market, but we need to invest in it big time, right now. I apologize to anyone I spoke down to. I get it. It's not a way forward, but I get it.

It's not fear over new tools, it's frustration with how it's wielded and how inappropriately, irresponsibly, unethically, and greedily it's pushed.

AI, itself is not the problem. The people who use AI, and the people who steal and trample and bludgeon to get it and while using it are the problem.

To disregard this entire aspect of the conversation is... Alarming and disappointing.

I believe we should all be pushing for ethical and responsible AI use and development, and what constitutes that and why. If not, the only resulting conclusion should and will be, simply: "Fuck AI", which is where we are, now.

Yeah he's had a very neutral take on it. Like it's a thing and it does good work some of the time but ultimately code not written by a person is harder to maintain. At least when a person writes code they typically have a strong understanding of what they wrote and can accurately communicate that to others.

I'm pretty sure he already handed out a ban for someone who submitted a slop PR.

He's been pretty clear that he's only interested in clean code additions. It doesn't really matter how you achieved your PR so much so thst it's quality code with a useful purpose for the kernel.

the mailing list entry that phoronix links to: https://lore.kernel.org/linux-media/CAHk-=wi4zC+Ze8e+p3tMv8TtG_80KzsZ1syL9anBtmEh5Z40vg@mail.gmail.com/

"

On Tue, 14 Jul 2026 at 19:01, Roman Gushchin roman.gushchin@linux.dev wrote:

I think it makes the point of sashiko - helping maintainers - unachievable. If the point to not use
LLMs in general, let’s discuss this, not how to make each use case more complex.

It seems like [1] expresses a very anti-LLM position in general

Yes.

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  

"

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.

No one is saying humans are perfect either. But Christ, the scale is clearly not the same. Before, an average bad dev had a limit blast radius. Now they can push a million lines of code and write a convincing description of why this is actually correct with little effort.

later

(Oops. Wait. You're right to push back. This is totally wrong. This should have never happend.)

"AI is a tool" is a thought-terminating cliche. The phrase is intended to foreclose other ways of framing the discussion.

While he is correct from a purely technical perspective, a lot of the backlash against other people using AI (LLMs generally) on collaborative work is due to how it is currently being implemented in a way that harms the environment, drives up hardware prices, sucks up mass amounts of data it shouldn't, breaks laws, negatively impacts those lower on the social latter , and a bunch of stuff outside of how it impacts code.

Like my limited experience with LLMs has been bad because it is frequently wrong, but I hate it because it is shoved into everything and has all the negative impacts I listed above. I wouldn't hate it and discourage others from using it if it was just new tech that didn't work.

Past the whole thing about theft of code, theft of wages and theft of rights, my next problem with AI is that it's viral, in even worse terms than say the GPL is viral.

Once you accept slop submissions into a project, you have the problem that someone would have to check it. But why spend your own mental and emotional energy, better dedicated to dealing with and interacting with fellow humans, to trudge through a mass of hallucinated, gaslighting crap? Better set up some AI that checks the submitted code.

And now you have two problems.

Fortunately, lists like Open Slopware exist.

Linus is severely mistaken that Linux is not a "social warrior" project, but I guess that's mostly because he's the face the community has to try and "sanitize" for corporate the posture that people like RMS have. Gone are the times where he would openly flip Nvidia the bird. Which is not bad, but it's unfortunately ill-timed: if history has taught us anything useful this decade is that, right now, bootlicking to corporate and its trends such as "uniformity" and "neutrality" is not the win you think it is.

what is a social warrior? is it some sort of shortening of social justice warrior?

it’s not just about the removal of the human, or run away corporations, or the destruction of the planet, or the squandering of art, or social warfare

it’s that ai is FUCKING DUMB.

you spend more effort trying to catch their dumb before it gets out into the wild, and to top it off it has the whole ‘semi self driving car’ problem where the point is laziness, so obviously the focus to whip the ai in shape is constantly waneing until it fucks up bad. and then you get to blame the ai, which is the PERFECT SCAPE GOAT. on top of that it’s sycophantic, where truth in its outputs is not even in its top 10 motivating pressures. it’s got a memory that will lose older information at inconsistent rates depending on fluctuating token availability, and when it suddenly sees a gap in knowledge, it will out right LIE rather then admit a failing. it does not have a true comprehensive under standing of any topic, because the very nature of an LLM is probalistic connection of abstracted nodes representing data/text. the longer it must work on a single conversation the more polluted its weights gets! it can’t prioritize certain expectations because again, not a true comprehension of the discussion.

and you want this to develop code! something that is understood to take deep planning and concentration, and interlocking systems?

I suppose that due to his position Linus is forced to have some kind of opinions about it, but I am so sick of all the "AI" talk that I can't even be bothered to click the link and find out what it is.