Linux kernel devs drop obligatory confession of AI tool usage
2 days ago by ☆ Yσɠƚԋσʂ ☆ to c/linux
The open source software movement was very much founded on the idea of being concerned about what kinds of tools we use as a society. Software that's "just a tool" can lock you down and out of what you can do with it. Same argument can just as well be made that people were "shying away from using new and interesting tools" by running free software instead of Microsoft's new priority stuff that locks you in
AI similarly gives massive control to a handful of companies. They have already used that leverage to intentionally sandbag tasks without telling you, when they think it's against their economic interests (model distillation and AI research). Is it not possible to imagine that goes further, they would use that leverage to slip in backdoor, not report classes of bugs, etc.?
Torvalds has never been a Free Software guy. Open source is a convenience to him. Why should he allow blobs in the kernel otherwise? They do the job and arguably don't violate the GPL. That's all he's ever been concerned with.
Torvalds has never been a Free Software guy.
That's all I need to know.
Do explain how open models like Qwen that you can run locally give massive control to handful of companies. In fact, what seems obvious is that using and improving these models is precisely what takes control away from these companies. People who started the free software movement didn't cry like a bunch children about how much they hated software. They had the brains to understand the problem wasn't the software itself, but rather who controls it. So, they went out and built open versions of the software that weren't corporate owned. And try guess what tools they used initially to do that since open software didn't exist yet. Your mind is about to be blown.
Only a handful of companies make open weight models, and those models are not made for charity. They want either market share and/or control. Making your own model is not like making your own software because the capital cost are extraordinarily high. The masses will not have the billions in funds they would need need to train these models
Unlike open source code, you cannot check what's actually in the open weight models. Worse still, you can hide things that trigger on very specific words (poison attacks can be intentional by the model maker or put by other actors in the public data). Fine tuning attempts are also no escape from hidden features you don't even know to look for
The reason Chinese companies are making open models is very simple actually, and has fuck all to do with charity. LLMs are a form of a general commodity, and it's not a viable business model to monetize them directly. Trying to compete on model quality is a race to the bottom. Anything one lab can do, others can figure out a few months later. So, you have to keep investing more and more money just to keep ahead. Chinese companies realized this early, and that's why they release the model in the open, aiming for the high end of the value chain building customized services and integrations tha capture the market. My expectation is that LLMs will become like Linux where nobody really makes any money off the tech directly. But you'll see platforms like AWS built on top of it.
Also, not sure what the rationale for companies releasing open models would be to trigger some unspecified poisoning in a coding model. Like what exactly is it gonna do when running locally?
Meanwhile, you absolutely can inspect the weights and there are now plenty of tools for doing just that. It's an active area of research, but these things aren't black boxes people make them out to be.
Finally, there are completely open models as well with training data, and source released in the open. None of them are frontier quality, but they are certainly a starting point.
There also projects like petals for doing distributed training. There's absolutely no reason why you can't have something like SETI@home for training models on a network. This is what people should be putting their energy into instead of whinging endlessly, as if OpenAI or Anthropic give a flying fuck that some brave lib made yet another 'AI slop' comment on a niche social media platform.
Software that’s “just a tool” can lock you down and out of what you can do with it.
AI is not “just a tool”. It is highly unethical.
It harms the environment, it harms actual developers, it introduces all sorts of legal issues where copyright is only one of them, it locks you in to a vendor, it makes you dependent of giant tech companies that steal your data and want to monetize your work.
Linus seems to ignore that or does not care about it.
It harms the environment, it harms actual developers, it introduces all sorts of legal issues where copyright is only one of them, it locks you in to a vendor, it makes you dependent of giant tech companies that steal your data and want to monetize your work.
Could you elaborate on this a little? I came here as I'm not very knowledgable when it comes to AI and am genuinely curious. Sadly you're the only person I've seen so far that has made coherent (even if broad) arguments against AI.
I know the environment argument is probably the enormous power demand of AI, and as far as the copyright thing goes, I've seen first hand how people abuse voice actors' livelihoods with shitty imitations of their voices that are also completely devoid of anything resembling skill. But how does AI harm developers, lock you in to a specific vendor or make you dependant on giant tech companies?
LLMs and machine learning techniques are just a tool. This iteration of frontier AI was harmfully developed by bad people. IMO it’s important to recognize the underlying technology as being useful while focusing on the current crop of products and companies as the problem.
These risks are not different from before and still open source has the advantage of being open to anyone and any agent.
Kernel modules written by US employees with unknown ties to US intelligence was just as much at risk for containing a backdoor as if written by an agent using an LLM with US intelligence ties.
And in both cases intelligence actors or security researchers from around the world can review it all and try to spot backdoors. So can agents with different LLMs behind them.
It would be an issue if there was only one state sanctioned LLM available, but as long as there is world wide competition in training LLMs, opensource will be “reviewed” with competing interests in “mind”.
That's not really beast mode. That's a guy with a god complex not understanding the problem after refusing to acknowledge he's not the only one who cares about Linux for the past twenty years.
the people I respect and call intelligent are not the ones that shy away from using new and interesting tools.
You know who else isn't shying away from the new tools? Trump and his cronies. Like this argument means literally nothing. People like Peter Thiel have said they want the world to end, these are the lines of people not shying away from the tools that Torvalds is aligning himself with.
Dude needs to fuck off and stop being a single point of failure.
You know who else isn’t shying away from the new tools? Trump and his cronies.
Hard to think of a poorer argument against AI use.
In line with the rest of the comments here so far where there's lots of anti AI rage but little to no coherent arguments that specify how Linus using AI to code is a bad thing. No specifics, lots of vague beating around the bush, highly emotive language.
My point wasn't to lay out every argument against using LLMs. That's been done ad nauseum elsewhere (not this comment thread no but other communities).
My point was his supposed beast mode was a poor argument for them by pointing out the exact same argument works both ways. Both smart people and dumb people are using them.
And quite honestly, in my experience, the people I respect and call intelligent are not the ones that shy away from using new and interesting tools.
(But) Both smart people and dumb people are using them
This is not the gotcha you seem to think it is...
If you think I'm arguing here that they're gonna go away then please, please, please, go take a reading comprehension class.
I'm arguing Linus's point that smart people use LLMs is a piss take that shows zero understanding of the world around him. That's it.
I just also happen to wish LLMs would go away for moral and ecological reasons.
I'm arguing Linus's point that smart people use LLMs is a piss take that shows zero understanding of the world around him. That's it.
That itself is a piss take from you. What projects have you used LLM's in for code assist activities (and I'm not talking vibecode slop)?
Ahhhh fuck bro. I'm sorry you're too busy meming to comprehend a point.
Wow, that's an incredibly stupid take. Like, the kind that you generally expect from brainwashed dumbasses that don't even know what is the difference between a screen and a computer.
That's scary that people would listen to a guy that completely lost his understanding of reality.
"Everyone who disagrees with me is stupid" - your brain on slop machines
Like that's Linus's entire point. "I'm smart, you're dumb. Fall in line". Not great
Nah, Linus's entire point is that LLMs are just a tool people use to build things, and drawing some arbitrary line between the tools you're used to and ones you're not is idiotic.
Wow, he's such an asshole.
If the code is obvious crap, it doesn't matter if AI was involved. If the code is perfectly obviously fine, it doesn't (directly) matter if AI was involved.
In the grey area in between, if someone doesn't say AI was involved it might mean AI was used undisclosed to vibe code the result from beginning to end. If someone said AI was involved, it might mean AI was used solely to speed up testing to track down the cause of the bug. In the former case, you'll get nothing useful out of the submitter; in the latter case the submitter can engage usefully in pushback on their proposed flawed fix. In neither case did the AI disclosure (or lack thereof) actually help.
And that's more or less where this news comes from - the tool usage disclosure turned out to not be useful, partly due to the vagueness of the method of disclosure (a tag in the commit message), partly due to people not reliably following the requirement, so they decided to stop requiring it.
I'm not going to start using AI voluntarily any time soon, but I'm also not going to start panicking that the kernel is suddenly completely turned into vibe coded AI slop like so many crazy people seem to have been doing lately.
honestly the code should be stood behind by the developer. then it won’t matter. and by stood behind, i mean responsible for the outcome as if they wrote it, able to explain exactly what the code does, how it does it. and have been reviewed before deployment. often times that takes longer then just coding it yourself
you can’t use ai as a scape goat when the code does not work.
Exactly. While I do have issues with the ethical and environmental aspects of AI, I do believe that many of the criticisms are disingenuous. Some made me wonder if the one criticizing never worked in a real dev project and just parroting
Bad developer have been copy pasting from stack overflow, writing slop, not following documentation/standards for years. This will never go away.
AI can do amazing good stuff, keeping the exit criterias checked, act as a second opinion, verifying upstreams and helping building a good prototype. making harnesses, hell even find a forgotten RAII pointer or concurrency.
But it can also lead to making gibberish that no one understands, mess up with productivity factor (if someone do code fast, why dont you??) and many other issue.
I think its all about approaching how we can gatekeep the bad code, instead of trying to pin how it was made.
Exactly, that's the only rational way to look at it. The person submitting the code is responsible for it. Period. How they went about producing it is their own business, but they are the one who takes ownership of the result.
That's a sensible stance and I don't disagree in principle, the problem is that I feel there are too few people that fall into that camp and a fair amount of them have had their brains absolutely cooked and believe they do, while they absolutely do not.
Of course I don't have any real data to back those assumptions up, it's not the kind of thing you're going to get solid data on, mostly just observations. I'd like to think the Linux kernel devs are a cut above, but given the state of things my faith is in the gutter.
So your stance is "trust without verification".
This will be fun when it inevitably blows up in all of our faces.
AI literally atrophies your brain.
Linus Torvalds ego has betrayed him, no one is perfect.
I don’t mind people using AI tools to code for them. As long as they understand the code when they submit it and can fix it, if there are errors/faults in it
So your stance is “trust without verification”.
I'm curious how you came to this conclusion
So your stance is “trust without verification”.
No. The reviewer is going to ask the submitter questions about the code and the submitter needs to be able to answer them in a way the reviewer is comfortable seeing it merged.
The former seems unlikely to happen (or at least for the long term).
Edit: other relevant post
If people are hurt enough, they will push towards it.
While I totally disagree with both, there are distributions without systemd and there are NEW projects that keep X11 alive and shun Wayland.
Something like this could happen with the Kernel. I’m not the one doing it because while I dislike Linus’ pro-slop stance I don’t really care enough for actively working on something like that..
But if available I’d install it instead of the slop Kernel.
Y'know how you're not supposed to just yell "somebody do something" in an emergency situation? If I ever hear someone yell such a thing, I'm going to be the "somebody".
I am going to fork Linux.
This is the kind of arguments people made against the free-software movement in the first place. That it's somehow more mature to not care about consequences of what we do and how we do it
I think it's exactly the other way around
The difference is that free-software movement is about rolling up your sleeves and actually making alternatives as opposed to throwing a tantrum like spoiled children which is what you all are. You lot are just a modern day iteration of Luddites who were raging against the technology as opposed to social relations that resulted in the technology being used in abusive ways. And lacking basic understanding of the problem you are all doomed to fail the same way as your predecessors.
The comparison to the luddites is interesting as you're tacitly saying "no you guys are right but your fucked anyway".
There is no good end with this technology. None. It will be used to abuse the working class, it will alienate workers from their labor, and it will bring an enormous amount of harm to vast numbers of people. This technology gives nothing in return for this loss except the ability to consume faster.
These discussions never focus on the fact that this tool optimizes the, already, most efficient part of the human supply chain, ideation and design. This tool does not make any of the other bottlenecks that slow human progress faster, those being implementation, adoption, and translation of solutions to different environments. The human race had the tools to solve world hunger shortly after the invention of nitrogenated fertilizer; yet here we are. The human race created a source of nearly infinite clean energy with negative temperature-coefficient fission reactors; yet here we are.
But you're ultimately right. The luddites lost, we'll lose, and we're fucked.
They were not "against technology", that was the propaganda of the ruling class which still gets repeated today. The Luddites were put down by militarily and violently killed. They didn't fail because they weren't publicly supported (they were wildly popular) or weren't against the social relations (they very much were)
Building frontier AI models cost billions of dollars in capital. Any solution involves raising that kind of money, the forces of the money will be put right back against you. It's also just as likely to be banned if it's seen as a big enough threat.
Arguing for the use of LLM on lemmy? Bold move. Don’t forget that 90% of the people are unemployed, jealous commies, attacking everybody who has even the slightest positive thought about LLM’s
This was only keeping honest people honest.
Linuslop Rustokenvalds
@lemmy.ml
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
go to feed...
@lemmy.ml
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
go to feed...
Linus going beastmmode
"
Linus Torvalds responded with an emphatic rejection of tool usage declarations.
He wrote:
What you are about is something entirely different. It has nothing to do with "expectations", and it has everything to do with your hangups.
I don't understand this whole idiotic concept of "don't use AI for X".
Here's a new rule just for you: don't use compilers for any of your kernel submissions.
Because if you can't show that you really understand what the machine is actually doing by writing your code in assembler, then I don't want to see your code contribution. You clearly aren't competent if you can't write it that way.
And btw, you had better not use an editor to write that commit message either.
Because I need you to show me that you can actually write a good commit message without using tools that make it easy to backspace and hide the mistakes you made while typing.
Ok?
Because otherwise those tool-assisted commits of yours clearly aren't showing that you know what you are doing.
Do you not see how UTTERLY STUPID those kinds of rules would be?
So no. We're not going to make stupid rules that say "you can't use tools for X".
The kind of expectations I have is that people do their best and I expect people to understand what they are doing.
But that understanding has very little to do with what tools they use.
And quite honestly, in my experience, the people I respect and call intelligent are not the ones that shy away from using new and interesting tools. ®
"
save