Remember everyone, everyone is human. There is no person with all the right opinions and there never will be. Never idolize, you will be burned
Funny how the loudest AI debates often happen without asking the people shipping the biggest software projects on Earth. Torvalds is basically saying: judge the tool by whether it reduces friction. That feels a lot more practical than treating every LLM like either magic or the end of civilization.
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.
I think broadly speaking what he says is reasonable. If you don’t want to use AI, then don’t, but I think it can be used somewhat responsibly.
I have all sorts of issues with AI tooling:
- the centralization of power
- the load it adds on maintainers because of the amount of slop bad developers (and non-devs) produce with it
- the environmental impacts of using larger models
- the data centers causing health risks
- the impact on the PC industry
- the degeneration of skills and knowledge.
That said, it’s a tool, and can be used to amplify good work too.
and can be used to amplify good work too.
But at this point I would argue the downsides aren’t worth it. Especially when the bubble pops, subsidies run out, and people will be forced to pay the actual price
Yeah, it is effectively subsidized by VC right now.
I expect what will happen after the bubble bursts is that the only affordable models will be quite small ones, not the energy chugging behemoths we have now.
They’ve already started cranking up prices in enterprise.
Sir this is Lemmy, I’m surprised you’re not downvoted to hell
It’s really weird for Linus to reference “the open source thing” while encouraging people to use legally dubious closed-source software with legally dubious output
This seems to be his view:
In the kernel community we do open source because it results in better technology, not because of religious reasons.
He doesn’t seem to subscribe to the idea that everything you use MUST be open source, as some more radical open source advocates do, but instead that more software SHOULD be open source.
I would definitely prefer if AI was open source and self-hosted (or at least E2EE if cloud hosted). Sometimes though the best tool for the job as things stand is closed source.
Aa for the legally dubious output, while AI can exactly replicate training data, it rarely does nowadays. It’s usually an amalgam of stuff.
I highly doubt E2EE cloud LLMs are even theoretically possible. Let alone the computational overheads for said theoretical zero knowledge computing.
As a non-dev person who recently started using AI to get into doing some computer programming, I think the marketing by the CEOs has done a lot of harm. People who truly know what they’re doing and who are very careful thinkers could see all along that this technology is not at a point to where the following hype was true:
-it will not entirely replace human workers,
-it’s almost certainly not going to be better at what it’s doing than a true expert,
-and it remains to be seen whether or not it will ever be able to generate truly unique, bespoke, and fundamentally new solutions… you know, like what humans had to do for every single issue that came before…
And what it’s really good at from my perspective is allowing a non-expert person to generate productive materials in an area where they are not a master-level technician/developer/knowledge worker. It has really dissolved the barrier of entry and the value associated with routine technical skills, for better or worse.
The practical issues though involved with massive data centers, the price rises associated with computer parts, and all the other environmental garbage that’s happening is pretty discouraging though. This technology has legit potential for benefit, but the companies and CEOs have ultimately made it into a moral conundrum whether or not it should be put to use.
Typical capitalism, imo …
Yes, anyone who wants to protest any use of AI whatsoever in software development, just on pure principle, is on the fringe at this point. It is a tool and needs to be used responsibly but it’s no longer credible to say that it has no use or can’t be used responsibly.
And it’s the job of lawmakers to rein in the harmfulness of AI, not of open-source software developers.
And even if it isn’t open source, expecting companies to be the moral police without actual jurisdiction involved has already led to censorship, like Steam purging a bunch of games because of pressure from Visa and MasterCard. This isn’t something to be encouraged, even when it’s for things we don’t like, until it’s actually legally clear how AI is harmful and not “dubious”.
Not that this mail is about users anyway, but about contributors. I didn’t get from this if Linus wants to push AI use for Linux contributors or not, but I doubt he does. But blocking people from using it by their own decision would be a bad move, as long as the output is reviewed and works.
A lot of my smaller issues with ai could be solved if it was used better like how I’d much rather see an ai that could be used to make music instead help you find similar songs or if ai giving you all the answers messes your ability to think why not put in the effort for a great search engine that doesn’t require you to already know what the perfect search terms are like a libertarian in a way more so than an oracle, reverse an ai art bot to find similar human made art from an ai image if you can’t describe it
I believe most people’s problem with AI comes from the potentially massive financial bubble and economical damage to come from over-adoption of the technology where it should not be used.
Many people, like Torvalds, know that AI is useful for certain things when used responsibly. Low level programming is one of the best places for it. You’re not solving business problems. These are usually optimization and security problems with verifiable outcomes(ex AI finds bug, you can actually confirm bug exists).
Unfortunately the ease of which you can produce crap with AI causes big problems for everyone trying to actually improve things.
Economical, environmental, ethical. How can it be used responsibly when the Earth is burning and making it required pilfering the sum total of human knowledge and creativity for private profit?
Running an open model on your local PC to assist with programming seems pretty environmental, ethical, and economical.
Other uses, not so much.
Economic damage, environmental damage, concentration of power with multi-billionaire fascists, cultural damage due to huge quantities of slop crowding out everything else, mass unemployment and the exploitation of remaining workers, disempowerment of everyone who doesn’t have the resources to control this technology, destruction of ordinary people’s access to suitably powerful computing machinery so that we’re all forced to either rent access to AI from its corporate gatekeepers or not use it, total surveillance by the state and its favoured corporations, profiling and predictive policing, marketing, etc. That’s quite a lot to object to, and I’m sure there’s more.
AI can also be run and trained locally, on hardware and data you own. The fascist-owned AI is not the only AI in existance
You first have to obtain the equipment.
As a programmer I’ve been doing this, but only because a friend lent me a very powerful computer that I could never afford myself. That’s the problem right now: local LLM use requires a huge upfront investment in equipment that most workers can’t afford. But it does feel a lot better (and cheaper, once you have the equipment) to run the LLM on a server at home than on a server owned by some nasty tech company. If everyone could run open models locally some of the main objections would fall away.
So I disagree with the poster who said anything you can run at home isn’t worth it, but the catch is you need an absurdly expensive computer. The one I’m using would cost about 6 times what I used to pay for a powerful development PC. The model and agent I’m using can do some complex things and get decent results. But right now no affordable computer can run them. On regular affordable computers the models you can run are indeed limited.
Anything that you can run locally isn’t worth it, unless you can’t do anything by yourself at all, and I mean anything.
AI can be both a bubble and a useful tool at the same time, a crash won’t make us forget about all about it instead I have a feeling a crash would help mature AI the way dotcom boom did web 2.0
Lemmy users are sure to have a reasonable reaction to this take.
Well most of them won’t read beyond the headline; they’ll likely just make massive assumptions based on title and their own predispositions. Sounds to me like Linus is trying to leave room for some of the practical nature of LLMs and implement them as useful tools, but it seems like he may be a little ignorant of the various agendas of LLM developers themselves. Just my stupid take.
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.IMO it’s a pretty clear eyed take. LLMs are tools and, like all tools, are more fit for some purposes than others. And, like many tools, they’re hard to understand, easy to misuse, and dangerous when they are misused (intentionally or otherwise). And, like any tool, they can be used for evil if the user has a mind to do so. Just because they are currently being used to do great harm doesn’t mean they can’t also be put to good use.
I think the FOSS community, in particular the Fediverse community, does itself a disservice by refusing to engage with them at all. They aren’t evil in and of themselves, they’re just new and selfish people will try to find an angle with them, same as it’s always been. I think it’s up to us to understand them and try to find a way to use them ethically. Show the world that it’s not that we invented scary Artificial Intelligence, it’s just that we created a cool, if wonky, search interface for text. Wonky in what way? Well, I think it’d be nice if there were robust FOSS tools and documentation I could use and/or contribute to, and learn about them with a community of my peers working for the common good, instead of having to either take the megacorps at their word or just renounce anything “AI branded” altogether, which doesn’t feel like a good long term strategy.
LLMs are tools and, like all tools, are more fit for some purposes than others
This is the whole problem, this take is disingenuous. Nobody who actually argues in good faith and knows what they’re talking about, argues that LLM isn’t a tool, or it’s something bigger than that. The argument that is being ignored is that LLM is a tool that inherently has downsides that are bringing more “bad” than it brings “good” when it’s useful. But proponents successfully argue that they can use LLM with some benefits, and act like they answered something, and any retaliation to that is unreasonable.
My favourite analogy is asbestos, and I say it as a huge disservice to asbestos. Asbestos is an amazing material that legit has many very important uses (way more than any LLM will ever have), but we’re not using it because we understand that no matter how great of a insulator it is, and how nice it looks as snow replacement on a movie set, and how wonderful it is in a cigarette filter, the supercancer it causes kinda nullifies all that.I think asbestos is a good analogy, but I think your claims that the inherent downsides of LLMs outweigh any possible upsides are as-yet unfounded. I also think it’s kind of strange that you assume anyone who thinks otherwise is being disingenuous.
Maybe they are too dangerous for broad use, and we need to regulate them like asbestos, or uranium. Maybe they shouldn’t be used outside of a laboratory setting or by anyone who doesn’t have extensive training with how to interact with them safely. It seems pretty clear that Torvalds has decided they’re worth the downside, and while I don’t know if that’s a good call, I don’t think he’s operating in bad faith to the detriment of the kernel project. That doesn’t sound like something he would do.
I feel confident that I don’t have the expertise to say for certain one way or the other, though the experience I do have with software tools makes me think there’s probably an application for them where the downsides can be mitigated to the point where they become worthwhile. I think there’s probably some single-purpose or tailored application of LLMs related to textual analysis that don’t require the theft of the whole internet, and don’t require insane amounts of energy to run. I don’t think we have discovered them yet (at least I haven’t), but saying “this kind of software is only bad and can have no ethical uses, ever” seems premature.
There are, broadly speaking, two categories of problems with LLM generated code (well, everything, but code here specifically). One is purely technical, does it bring anything to the process, does it broadly speaking saves you time or effort. And the second is everything else: environmental impact, the bubble of it all, the consolidation of power to all the worst people, erosion of skills, death of education, inflation of prices of consumer electronics, psychological impact, and so so so much more, all lumped together as non-technical downsides.
A bunch of llm-proponents, and here we sadly have to include even Linus, are only engaged with the first category of problems, throwing the whole second category aside as “irrelevant”. And a lot of those people are smart enough so I don’t believe they don’t see this.
If there was only the problem of the tool being kinda shitty but useful for some people, I wouldn’t be so against it. I still don’t believe they actually benefit from it, but if they think they do, I’m not going to argue with them about it.
But it’s fundamentally not true. You can’t ignore the second category in your evaluation. To quote Linus himself,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
about the fact that even one of the downsides from the second category makes the whole “but I, a person with 40 years of experience, can use llm in such a way that it sometimes produces some benefit” not worth it. No benefit of it is worth the fact that the new generation of junior developers don’t know how to write code on a fundamental level, while elon mask owns all of humanity’s personal data and can do whatever he wants with it. Even if those benefits were kinda big, it wouldn’t worth it, but they aren’t.
Well all I will say is that I do understand the usefulness of LLMs as tools, though I only really use them passively through search results, and always verify the results separately. And I have a fundamental understanding of how they work. While I understand what Linus is getting at in a practical sense, I worry he is putting his head in the sand regarding the heavy handed agendas at work in the implementation of AI as we know it now. And I think that is a definite bad thing to be ignorant of. I personally won’t embrace “AI” until it is liberated from the agenda of power and inequality. Though, I cannot but accept its level of influence in daily life, and will attempt to understand and recognize if it fundamentally evolves or changes.
I think if we, as a community, really put our heads together, we could figure out how to define and make useful user-respecting tools that incorporate LLMs. There are a lot of hard problems, like the massive power consumption, and the ethical use of data, that I don’t really know how to solve.
I think a start would be to focus on making smaller, lower-complexity models that are built for purpose, rather than trying to make a jillion-parameter jack-of-all-trades trades model. I think it would also make sense to focus initially on areas where there are already large corpuses of freely available text, like all the writings in the public domain. But I don’t really have a good idea of what these tools would be used for, exactly, which is where I’m stuck.
I understand and appreciate where you desire to go with it, I just don’t see a current system to support it. Afaik people do make their own AIs/LLMs (recalling an artist’s court case regarding the legitimacy of their work), but large scale LLMs do their thing by running roughshod and stealing tons of data and driving up hardware and electricity prices. It is difficult to see an environment where small models prosper compared their lawless corpo counterparts. It would probably be more beneficial to humankind, though.
Yeah, you put your finger on maybe the biggest problem that I don’t have a good answer to yet. The ability to interact with LLMs in natural language comes from an analysis of huge tracts of contemporary human writing, and there’s not an ethical corpus that can compare in scope to the corpus of just snatching the whole internet. Maybe they don’t need to be strictly “natural” language interactions? Maybe a sufficient ethical corpus could be compiled and maintained somehow? I don’t know, this is kinda where I get out of my depth.
The energy consumption can be measured, very approximately, by the cost of the tokens. If you design agentic tools to make use of fewer, cheaper tokens then you’re likely also minimizing the energy usage.
How vague of a guess is that when people rely on those closed systems? We all know Corporate AI is hiding the actual costs with creative accounting while they enjoy abusing the environment…
Comparisons can be made using open AI models. I run some locally on my own hardware, I know exactly how much energy they use.
I had enough of this “Big AI agenda”.
Yes the oligarchs at the top are the most evil and corrupted people ever. Sure. Ok. Understood.
But what does it have to do with Linus and the kernel project ? He is not like vouching for them or even for their specific models. And it’s not like there isn’t open source AI models too. There is, and pretty good one at that.
But no, AI is the obligatory Claude code or chatgpt obviously…
Damn if only Lemmynautes could understand that we can’t be dumb enough to consider AI as one big proprietary monolotic block. It is not and Linus is well aware of that.
And by the way, Linus I’m sure understand very well the agenda of the big evil corps. It’s just not his problem. He is leading the kernel open source project. He is not involved in a crusade against AI and I’m so glad he will focus on that instead of “picking a side”. If there is anybody I trust to use the tech effectively and pragmatically it’s the kernel open source project.
It’s not complicated truly, all AI models are not the same. They dont come from the same business, the same country, they don’t have the same licence.
If we used half the energy we use to vociferate about big corps AI to actually create open source (and copyright respecting) models that would really improve the situation in the future.
It’s more than time for the open source community to wake up and actually create their own models, built ethically and in accord with the OSS ethos.
Obviously, I’m well aware of the destruction of our planet by AI datacenters but again that is a problem for politicians, not freaking Linus Torvald.
This is such a bullshit take. Yeah, “open source models” exist, and someone probably uses them for something. Most of the slop being commited is generated by the big ones though. Because, as any sloperator will tell you, only the latest and greatest claude is actually good, and all the problems with your slop is because you didn’t use the latest the bestest. So while most of the slop is generated by big corpos, we need to talk about evils of big slopgen corpos, and how they impact the world.
And when you say “yeah, who cares, if we can use it we need to use it”, you’re vouching for this shit, you’re putting your voice, your money, your support, to worst people that own the worst companies.
Linus is a very practical man. If your code is good, ai or not, it’s good, but if you use ai to break the userspace…
So they are going to autocomplete their takes based on the headline, I hope the irony isn’t lost on them.
He doesn’t really specify use cases beyond Sashiko (I believe that’s the project they’re discussing), which, if I understand correctly, uses an LLM to review kernel patches using a pre-set list of prompts. The output of those prompts presumably gets added to the PR for anyone who cares to read it.
These outputs sometimes have mistakes in them, so the maintainers must read them with a critical eye, but I think Linus is arguing here that it’s sufficiently… “good enough” that including them is helpful to the maintainers, that it catches enough real stuff without wasting too much time on fake stuff. There are always tradeoffs when adding a tool to the stack, and sufficiently rare misfires are probably worth the times it catches real problems that humans miss.
Stupid leftists think orange man and AI bad? Cringe! Updoots to the left!
Lemmites deciding whether to delete their anti-AI comments or their pro-linux comments
I felt the message was OK-ish but it bothered me that he used the term “social warrior”, which is an older term used by right-wing chuds before they got married to “woke”
To be honest I would not be surprised of any political views Torvalds would have. 😅
He did not use the term “social warrior”.
The quotation marks indicates a distance from the expression. They’re clearly used to show that Linus considers the term someone else’s label and something he doesn’t endorses.
While I also disagree with how he have behaved and his opinions, I think it’s important to not make up alternative versions of what he has said or not said.
I read this Linus Torvalds quote out loud and my dog freaked out.
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.
It’s important to note that this man’s contributions to Open Source can be taken freely, and he and his opinions can be thrown in the wastebasket at any time.
Welp, I guess it’s off to BSD I go.
J/k. This seems like a pragmatic approach from Linus.
deleted by creator
I’ve felt I should walk away since I heard his views on GPLv2 vs GPLv3… and his infamous lack of compassion when communicating with others.
I’m interested using GNU Hurd as an alternative to Linux, eventually.
Given that GNU Hurd started over 35 years ago and the only distro, such as it is, that’s using it has latest news from 2018, you could be waiting for a while.
Gentoo GNU/Hurd experimental was released April this year 2026.
I may wait but perhaps I shall get involved 🤔
Well, that’s something!
What about his lack of compassion?
"Torvalds is known for vocally disagreeing with other developers on the Linux kernel mailing list. Calling himself a “really unpleasant person”, he explained, “I’d like to be a nice person and curse less and encourage people to grow rather than telling them they are idiots. I’m sorry—I tried, it’s just not in me.” - Wikipedia
Ah something we all can relate to honestly
