The role of AI tools (LLMs, mainly) in Linux is under discussion, or it was, until Linus Torvalds “put his foot down” in support of the use of AI in Linux kernel development.
You must log in or register to comment.
I think Linus has realized that LLM’s are not going anywhere and can actually be a useful tool if used correctly.
And he is right in that respect. I know Lemmy is very anti-AI, and for good reason. Because we do not NEED it. And heat producing, power and water guzzling datacenters in this time of climate change is absolutely NOT what we need.
Yet I think LLM’s are here to stay. When used as an “employee” you instruct what to do while keeping total control of the code it produces it is actually pretty useful.
However, I really do hope that open weight models keep improving (which they are) and running them locally on midrange gpu’s on your own computer will be the future.
Because the way the big AI companies are doing things now is absolutely not sustainable and I hope they all fail soon.
He wasn’t in support of it per say, he was against a strong stance against it and makes it clear it’s a tool. In/out grouping is a shitty rhetoric that to me feels unenlightened. Especially when combined with twisting someones words.
The linux kernel has very high standards and a culture of people owning their code, which makes them pretty immune to slop. When code is under this kind of scrutiny, what difference does it make if it was produced this way or that way. I think he’s just being pragmatic, there are better hills for him to die on.
Arguing over use of a tool is foolish.
Arguing over correct use of a tool absolutely has merit.
AI as an industry has extremely debatable morality, however you kinda can’t put this genie back in the bottle.
Like him or don’t, his ultimate message of “This shit is here so you may as well learn how to embrace it” isn’t inherently bad either in context of bug-hunting and skilled individuals using the tools as an aid rather than as the alpha and omega solution.
How is this controversial besides coming from the guy who said it?
Some of us perceive benefit in not touching “the shit” and not letting “the shit” get in our environment (computers, phones, lettuce)
There may be no tech way to put the genie into the septic tank, but surely we can’t be forced to deal with it, and until the foot dropped there was a social way to rebottle the genie
Oh I agree. I jumped off Windows the day I got my laptop and saw what a horror show windows 11 is and am outright refusing to update my android until I can unlock my bootloader and entirely degoogle.
I also don’t use any of the big LLMs because I see no point and don’t have a job that requires I do so. I personally believe AI is a net-negative for humanity in general.
One could also argue the very same for alcohol, gunpowder, firearms, etc. The only recourse left is to argue ethical use, not cry about not wanting it in the world anymore.
Besides, what does using AI to debug the kernel and optimize it for people who have reason to run local LLMs have to do with you and me? I’m all for sticking it to The Man when he does some Seriously Dumb Shit, but this whole situation qualifies for neither.
This is the correct take, and I kinda hate many things AI is doing.
“Rape and adultery are a major boost to reproduction rate and infancy survival rate therefore are very useful. Embrace it. Enjoy it.”
Haha. Do you people have no morality?
Read further before responding next time and don’t “you people” me when you’ve got no idea what bucket I fit in.
Lucky we still have OpenBSD.
As a member of the community, I won’t have a problem if the interactions with LLMs are only with open source ones, and the conversation is available as open data to all, without restrictions. The same which CommonVoice does for Voice.
Data matters so much that the companies keep it to themselves. And I absolutely don’t want the data to be used for training models which enslave the developers and the people of this planet. The least is that data should be open to all, so the FOSS community would have a chance to use the data to contribute back to FOSS.
Note that it doesn’t matter if the model is open source. For Machine Learning the data matters as much as the code does, if not more. And in this case, data matters much much more. The code is widely available and can also get researched independently.
in Linux is under discussion
Are you referring to LKML discussion? Or some other metaphysical discussion we should be aware of? Yes. Linus replied in the thread.
It’s not really very surprising point of view.
Torvalds has years of experience as a developer, he wrote:
AI is a tool, just like other tools we use. And it’s clearly a useful one.
Which I firmly agree with. I think if AI is used responsibly it can be beneficial for certain tasks. More education around better AI usage and practices is more important now than ever before.
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? Temperatures during heat waves are exceeding 50°C in India, causing tens thousands of deaths.
Literally a negligible amount of emissions and seems to be the core of their argument. It doesnt make a dent in any of those stats nor does it have any impact on the temp in India. Reminds me of the people who screech over private jets as if banning them will do anything to slow down climate change. Even his own link “The AI built-out is by far the fastest growing energy consumer in the world” undermines his point showing that computing has been the fastest growing industry for years by % increase (its a tiny % of the total) not by actual power increase. The projected 1.5% to 3% of global energy use seems reasonable given how massive computing is in our day to day lives. We are lucky we have something so efficient that is so useful we spend far more energy on far worse things than computers.
