AI in Linux

throwuxiytayq

tl;dr Kernel devs should prompt less because it uses up water or something. Repeat after me: you can't put the cat back in the bag. Pandora's box is open. Sucks, but you have to learn to live with the reality that most developers will use AI to write code from now on.

JodieBenitez

I propose we get rid of tiktok videos and let the kernel devs use AI. That should balance.

datakan

I read through this and must have missed it, but I don't see a single example of AI use in the Linux kernel development posing a problem.

jmclnx

I do not see how AI could be legally used on a GPL project. But if you want to use a non-AI OS, NetBSD banned the use of AI. Maybe some of the talent will move over there. They can sure use the help. As for AI in Linux, I am on the fence about that, but since Linux is pretty much funded/owned by large corporations, AI being forced on Linux is no surprise at all.

colesantiago

Drew, there will be new jobs. Also, the open models are a great alternative to the datacenter ones. I imagine AI will go towards the local models more when the bubble pops.

logancbrown

>The same process is pricing regular people out of the energy they need to power their air conditioner during those heat waves and the technology it enables is pushing them out of the labor market and into poverty. I don't see evidence this is true, and the critical mass of data centers is making electricity cheaper. Economies of scale -> more energy = bigger electric plants = cheaper energy. https://netchoice.org/new-study-confirms-what-weve-been-sayi...

criddell

From the title, I was hoping this was going to be a story about the major AI distros agreeing on an API and basic AI capabilities to be included in Linux just like Microsoft and Apple have done with their operating systems. Instead, it's: > how many tons of CO₂ added to the atmosphere or liters of fresh water supplies disrupted is a tolerable price for a better code review? The answer isn't zero, is it?

Magicrafter13

Torvalds' statement was incredibly disingenuous and rather sad. Every single commit is making a legal and political claim. Each commit is saying "this code is my original copyrighted work, and I am licensing it to the world per the requirements of the GNU GPLv2" (or "this code is someone else's work which was licensed under a compatible license, and I am integrating it into this GNU GPLv2 project"). That's not social warrior, that's simple facts. We don't need to try and guess motives or reasoning for people's contributions - that doesn't matter. What we have is an explicit declaration by virtue of the license, that the code being provided can legally be included in the GNU GPLv2 licensed "Linux" kernel. But with LLM generated code, this legal claim is dubious at best. For trivial (!= simple) code, one can probably argue that the LLM code will not be violating anyone's copyright, however for anything remotely novel, unless you are 100% certain that all of the training data was GPLv2 or compatible, it is not legally safe to use - and aside from that, even if all the training data was compatible with the license, the LLM will virtually always fail to properly attribute the source of its output, which is a requirement of - as far as I can tell - every license compatible with the GNU GPLv2. I cannot share GPL'd code and misrepresent it as my original work, I still have to make clear that it was someone else's creation. LLM generated code is a ticking time bomb, and is going to poison projects. Yes, it is possible to do this already without LLMs, but not by accident. A developer would have to make a conscious decision to violate someone's license in the past. Now it can happen without anyone being the wiser. And as someone who very much uses GNU GPL licensed software because of its license, this worries me for the future of software. On technical merits, it doesn't matter, but if you care at all about the artful expression of software, as well as honesty about who created what, this should concern you.

globalnode

plz dont wreck a good thing, keep ai well away from linux.

vanagandr

This is a “everything is political” rant but now with 50% more AI.

drnick1

The author seems to be making a mountain out of a molehill, to be honest. Linux has never been a "political project" like the author asserts. GNU is closer to that, but even GNU has a purely technical component (the software itself).

lptrap

Linus' statement that Linux is not political is strange. It started as a resistance project and used GPL for ideological reasons. Linus himself was made to sign the corporation-enforced CoC and "retreat" for "counseling". He came back a changed man. Since then I've only see him a) use the CoC as a weapon for the old boys and b) parrot any current mainstream or corporate line. That includes COVID vaccines, overt anti-Russian statements (when not warranted) and now AI. Would he still give Ndivia the finger? I doubt it. These Gen-X millionaires cling to their positions and I'm sure the Linux Foundation has a well oiled HR department. Ted Tso also already interfered pro AI in the Debian discussion, where the pro-AI resolution is also championed by Canonical shills. You can't fork, but you can use BSD.

sav_m

If everything is political, as DDV likes to rant every now and then, then Linus' politics are those related to how software is to be developed and distributed. I don't see why a software project _should_ care about global warming, minorities, state policies and other issues outside of its scope. It could, and it can be a good thing to do, but it doesn't need to. It feels wrong to me, though, to see AI being used in kernel development, as I see the trouble it causes on regular projects.

keeda

> Looking at it from another angle: how many tons of CO₂ added to the atmosphere or liters of fresh water supplies disrupted is a tolerable price for a better code review? Fewer than if a human had done it without AI: https://news.ycombinator.com/item?id=48941143 I guess that means he should advocate for more use of AI for Linux dev! ;-)

loudmax

This is an important conversation to have and we should pretty clearly about what the issues are. I didn't see any objection to LLM assisted development in the abstract. The objection is to relying on proprietary AI providers in general, and Gemini in particular, via Shashiko. The problem is that kernel contributions could become gated by hyperscalers who can draw on compute power normal people can't access. Devault mentions data centers' carbon emissions and water usage. I find both of these objections to be distractions. It's not that carbon emissions don't matter. Keeping greenhouse gasses in check is incredibly important. But there are far more productive ways to cut carbon emissions than to block data centers. Heavy industry consumes far more electricity than data centers, for less economic value. This is even truer for water use. If carbon emissions is a major concern (as it should be), there are better targets to go after than data centers. A more compelling argument is the danger that the Linux kernel becomes something that's effectively controlled by proprietary AI providers: Google, Anthropic, OpenAI, Microsoft, etc. The photo of tech company leaders at Trump's inauguration underscores their lack of any moral compass. But if AI-driven kernel development is actually useful and productive, that's going to be hard to turn down. If you believe that LLM use is technically counter-productive, then this whole AI thing is a bubble and you can just wait for it to pop. If you believe that LLMs can be technically productive if used correctly, and the drawbacks are entirely social, then it means giving up something technically useful. That is a real cost. I think Linus is right here: it would be hard for regular kernel developers to coordinate a fork of the project. It might not be that hard for the hyperscalers to simply fork Linux. They would still release the source code under the GPL, but nobody would be able to keep up with it. Ultimately, this could doom the human-written version of the kernel to irrelevance as most people switch to the faster/stronger/harder LLM-derived fork. Maybe a reasonable medium would be to insist on using an open model for code review instead of depending on Gemini or any proprietary model.

jamii

> The AI built-out is by far the fastest growing energy consumer in the world, and they’re being built with fossil fuels, or drawing green energy demand away from replacing the fossil fuels depended on by other industries. This paragraph links to the IEA report which does indeed call datacenters the fastest growing energy consumer, but also says: > Despite rapid growth, data centres remain a relatively small part of the overall power system, rising from about 1% of global electricity generation today to 3% in 2030, accounting for less than 1% of total global CO2 emissions. For scale, beef production is about 9% of total greenhouse gas emission.

panny

I'm ambivalent about the impact of AI on society with respect to Sashiko, but that is the weak argument to make against AI code review. The strong argument to make is that AI code review sucks. It does a bad job. It denies there are bugs that you have personally verified exists, questions the reason for the patch existing based on this denial, and then demands you create a giant scaffolding of tests to prove to the AI that the bug exists and that your patch will fix it. It is rude to do that to someone contributing their free time to help your project. And yes, I have forked one project as a result of this already. AI code review will atomize your community with great speed. Beware.