I’m grateful for my blessings and believe in giving back. I’m organizing a $35,000 giveaway to support people facing financial difficulties. For official details, eligibility criteria, and participation guidelines, contact @Georgestones12 on Telegram and message “GRACE.” Peace 🕊️.
I guess Linus doesn’t care about the environment or any of the other issues surrounding A.I. I’m not anti-AI just for the sake of being anti-AI. I have a huge problem with AI bcz of it’s environmental impacts, intellectual property violations, job impacts, attempts to replace all employees with AI, the general dumbing down of society and worse off the internet is bcz of AI, the surveillance state being built around it. I could go on, but you get the point. I am a staunch, no AI person bcz of all those reasons and a hundred more.
This isn’t an age thing, or a, I am getting older and Im stubborn thing. AI is literally ruining the planet and everything on it, and this guy thinks we should just accept that.
I say fuck off with that noise, I don’t have e to accept anything.
Conversations on this subject will continue to stagnate and miss the important points as long as people continue to act like the discussion of ethics is the same as the discussion of efficacy. People who disagree with AI on a principle of its impacts being negative will often call it “a lying machine which is regularly wrong and offers no utility”.
May I suggest that people with criticisms of the ethics and impact of AI take a charitable position towards the utility offered by AI. Accept something along the lines of a positive account given by an experienced software engineer - it’s useful, it presents risks but also it presents new ways to get certain work done well in much less time etc.
Even if you do not believe this, the point is it is a much more ambiguous argument than the ethical argument or the concerns over the impact of AI on the species, environment or society. You weaken those arguments by putting them along side poorly justified, somewhat falsifiable claims that AI just doesn’t work.
LLM is like a compiler? Then please replace the C sources with English and let it compile!
I’m finding the allure of AI is particularly strong for people who have risen to positions of power where they had to stop doing the work themselves. Instead, they had to work through other people to get bigger projects done.
The allure is that you can just say what you want done, and it does it. If it’s done it “wrong” you can just tell it it’s wrong and not have to worry about arguing the technical differences. As such you feel like you get closer to what you wanted, rather than somebody else’s take on what you wanted.
The things is, when you are removed from the code, you can only work at a high level. That person you asked to do something is considering 10 other low level concerns that you didn’t consider. I think Linus’s recent patch series is a good example. He saw a potential optimisation, did a proof of concept but missed a race condition. He’s now handed it over to the person that normally works in that area and knows the details. Can the optimisation actually work? Who knows?
Does Linus support steaming artwork and a fascist surveillance state? Because using and supporting and entrenching the technology use is exactly that.
I’m no expert but a compiler does not equal and LLM built on stolen data
I think he’s kinda right in that competent engineers can write good code using LLMs. I don’t know whether they’re actually more productive that way though. The problem lies in the fact that incompetent engineers can write a lot of bad, but good-looking code using LLMs, which might be a pretty big burden on the maintainers.
Also, comparing LLMs to compilers doesn’t really work in my opinion. Compilers follow strict rules and translate one kind of logic into a different kind of logic, while LLMs turn natural language and vague notions into some kind of logic. Writing code is telling the computer precisely what it should do and prompting is vaguely telling it what end result you want and those are two fundamentally different things.
He doesn’t get it… Have the big corpo-shilling libertarians rotted his brain?
Like, imagine if using “AI” consisted of whatever big corpo SaaS you subscribe to sending your prompt off to someone to work on then send the result back to you. Isn’t that an idiotic dependency to have? And why the fuck even go through the middleman? Oh they have the “best workers?” Fuck that. Nvidia makes the best cards, but you don’t see me buying from them.
Do people like Linus not realise how silly it is to offload your work to some middleman service that just steals work from people who do not consent to sharing their labour? Who are desperately throwing insane, GDP-level boatloads of cash at it this unknown technology in the hopes that they’ll somehow get people hooked on their particular chatbot?
The intentions of these companies are not good. Their entire valuation is not based on being a “force multiplier” or whatever bullshit, but on being able to threaten or fire workers if they don’t slot themselves into the newest enunciation of hype-based capital.
Anything you do to willingly boost or massage these corporate services is enabling actually vile companies to continue puffing air into their stupid pointless bubble until it bursts and the next generation has to eat tuna helper sans tuna just to be able to afford to live.
Weird, i disagree with Linus.
AI is not a tool. Its an addiction. Maybe the first dozen times (more likely: never) you use it “as a tool”. Very quickly, you depend on it. You need it to refactor. Then you need it to debug. Then you need it to write. Next thing you know, you spend 0 time writing code (the fun part) and 100% of your time as a reviewer (the worst part). Suddenly it isnt fun. A few fools even get TWO AI MODELS, one to write, the other to debug.
Its a virus. I cannot tell you how many 300+ line commits ive rejected because they straight up dont do what the person said they do or also do some extremely asinine 2nd thing that breaks other pieces of code (which they would know if theyd actually used it).
A tool requires input. A hammer can never do anything i dont explicitly tell it to do. A good word editor does exactly what i want it to do. A good compiler does exactly what it says it will do. AI does not. Ai is a glorified autocomplete, and i dont want my code to be automatically generated. Its asking for thousands of issues down the line. If AI was any good, Windows would be a usable platform by now.
Dont get me wrong; i know that it is sometimes used as a tool by people, and meticulously checked over, but more often than not, it isnt.
Except that a compiler is not a stochastic machine.
hes comparing knowing how open concepts work vs using proprietary tech whose purpose is to hook you and then enshittify after you become reliant on it
The amount of bootlicking Linus has been doing recently is insane. I expected this bs from anyone but him. Such a sad reality we are living in.
I like Linus, but comparing an LLM to a compiler is fucking stupid.
Why?
You can predict with certainty what a compilers output would be. Linus is conflating the fact that most people don’t understand how a compiler converts higher level programming language into machine code with the reality that no one can actually predict and audit what an LLM puts out.
You cannot, that’s why reproducible builds are hard. Modern compiler are made with compilers (egg/chicken problem) not assembly nor machine code and you always loose some perfect prediction.
Yes LLMs are less predictible than compiler but their output is readable to human, while machine code is very hard (yet not impossible).
Is that a bad things that 99% of devs have forgot how to write code cutting holes in paper? Tools are changing and while a direct comparison between the move from paper to COBOL or assembly to C is not the same as python to LLM generated it’s hard to deny that LLM will change how human write code.
Thanks for that explanation. Are you saying that if I rebuild the exact same code on the same machine I could get different results for a given compiler? I think this would render solutions like docker useless. Could you elaborate?
Yes you can get a different result. Compilers are designed to be deterministic, but a build is more than the compiler. Usually the bytes differ but the behavior is identical.
Docker is far from being reproducible!
It works on my machine ≠ reproducible. But reproducible will make it works on your machine.
Using a compiler does not guarantee you to get byte-perfect output each time, especially not on different machines.
Talk about misleading. The differences are timestamps and metadata, and some compilers have flags to let you zero out timestamps and force it to produce a bit-for-bit identical file every time.
Yes, that is a great explanation. Thank you.
That is, of course, the whole point of docker (making sure the results of compiling are identical). I totally get the different machines giving slightly different results. This is not really in the same realm of the dramatically different output a LLM will give though.
No the goal of Docker is not to make sure the results of compiling are identitcal. It is to make it work on your machine, there is some nuance to it.
This is more the goal of something like nixpkgs (NixOS).
This is not really in the same realm of the dramatically different output a LLM will give though.
Yes this is not same realm but my point was that eve if LLM would give you different results each time the result is transparent and editable. Also developers tools evolve and Linus have more technical knowledge to use this tool in a way that will gave him satisfying result than me or a junior dev with 1.5 years of experience. He was against LLM generated code a few years ago, not because he hated on AI but because this tool was giving him unsatisfiying results. If compilers were not working 50% of the time he would probably write everything in assembly. LLM models and their coding harness cannot be compared to what they were 2 years ago. And most of the devs are using them. Will that transform bad devs into good ones? Probably not. Does that mean that LLM generated or assisted code is the way to go? Not necessarly. But it means that it’s a tool that find its use in the majority of development teams and thus that they find it quite useful.
There is some issues with it and I’m not here to deny them. Whether they are social, environmental, capitalist, monopolist, learning. Like every new disruptive tool we use them in ways we won’t in the future and I think it’s similar with AI.
He is saying that you have to understand your contribution, so whatever the LLM spits out, you need to thoroughly review and double-check that it makes sense. So, in particular, he’s not trying to rate these tools, he’s merely saying make sure the contribution is objectively good, then he doesn’t care how you got there.
I got that part and I agree with it.
My issue is him saying don’t use a compiler if you don’t really understand the machine language it creates. This conflates LLM with a compiler as a tool when they are inherently different. This is disingenuous because it is possible to understand and audit what a compiler creates. You cannot audit what an LLM creates in even close to the same way.
But why can you not audit what the LLM creates? You can’t do it for the general case, i.e. prove that for a given prompt it always spits out a valid solution, but you can look at what it actually spit out for your concrete prompt and then you can freely decide, if those code changes are good.
I am of the opinion that a thorough review often takes similarly long to actually writing it yourself, and there’s a high chance that you miss architectural aspects, if you don’t investigate yourself.
But yeah, again, I think you’re taking the comparison too literal.
He is just saying that LLMs are a tool. You need to know how to use your tool and produce objectively good results. Then he does not care what tool you use to get there.Auditing typically means tracing the logical path. How do you suggest to do this with an LLM?
I am definitely taking his words literally which is why I disagree with just this one point.
You don’t have to do everything with the LLM. Tracing the logical path very much still happens by looking at the code in your editor.
This is also why I’m saying it isn’t much faster. Understanding the existing code and what changes are necessary is the bulk of the work when handwriting code as well.
You *can* tell the LLM to go ham and implement the change on its own and then only look at the diff afterwards, but anyone who thinks they understand the code changes then, is lying to themselves IMHO.
Can you elaborate on the last sentence a bit from a technical perspective? I’m not using code myself to the extent that I could really judge between LLM-generated code being good or just-works. I can just imagine compiler output being strictly deterministic, while LLM-code, due to the way a LLM works in principle, should be more …unstructured?
I honestly didn’t think you could even audit an LLM. It is possible, but not in the same way you would audit code. It is my understanding, because of the probabilistic nature of how the LLM operates, you can get different results even within the same model. This makes tracing something like logic next to impossible. Why the LLM gave the result is somewhat of a mystery. It is in effect a black box. This may change in the future, but for now LLM suffers from this issue.
Sleeping agent in LLM is a scary topic and the tech itself has mathematically no solution for this: https://arxiv.org/html/2401.05566v2
all the undefined behaviors would like to disagree with you there champ
Can you elaborate?
Compilers are incredibly complex things and the often have undefined behaviors that slip in because the compiler has to interact with the operating system at large, and that means the code actually doesn’t run the same on every platform, or even every version of the same operating system. So, while they are technically deterministic, the reality is that you get surprising and unexpected behaviors in practice.
Also worth nothing that the probability distributions for an LLM are fixed at training time, and that is why the weights are static. When doing inference, models are selecting from a top-k of probable tokens at each autoregressive loop for your typical LLM. When you adjust top-k to 1 or simply set the “temperature” to 0, it will always give you the same result from the same input.
Thanks
Cory Doctorow’s “Reverse Centaurs Guide to Life After AI” has a short section that pertains to this. He talks about how someone like Linus essentially doesn’t have a boss and has untold resources for whatever “AI” “tools” he finds useful.
He isn’t ever forced to use the wrong tool for the job or use a tool that simply doesn’t do what his boss thinks it does. He doesn’t have to worry about getting fired or his coworkers getting fired because his boss was sold a story about reducing labor costs with this cool new “tool”. He simply uses what’s useful to him with the probably millions of dollars of compute made available to him by the tech giants.
95% of people don’t have that luxury. They have a boss itching to get rid of them or 10x their workload with the help of the “tool” that Linus is so happily and defiantly praising on behalf of the monopolists.
I think there’s a broader question here from an evolutionary perspective. Humans are fundamentally tool users, and technology has been central to our development as a species. It’s possible that this evolutionary niche we’ve occupied will eventually reach its limits and contribute to our own downfall, but I’m not convinced AI is uniquely responsible. It’s part of a much longer trajectory of technological progress.
We’ve evolved to solve problems and use tools to make our lives easier. That doesn’t mean technological progress is inevitable, but these tendencies, combined with competition, make it difficult to imagine us ever really stopping. Even if we stopped today, who’s to say another generation of humans 10,000 years from now wouldn’t continue down the same path?
I don’t think this is exclusively about capitalism. Capitalism certainly incentivizes growth and reducing labor costs, but competition, greed, and the pursuit of technological advantages have existed under other economic systems too. That doesn’t diminish the legitimate concerns about how AI is being used against workers.
I also wonder whether our short lifespans make us particularly bad at considering the long-term consequences of our innovations. Who is going to sacrifice the opportunities of their own generation to prevent a possible catastrophe centuries from now, especially when there’s no guarantee it would stop technological progress?
I have a degree in psychology and particularly enjoyed evolutionary psychology, so I think it’s worth asking whether the deeper problem is our own nature, and whether our capacity for collective restraint can keep pace with our ability to innovate.
I know this raises more questions than it answers, but we’ve faced civilization-changing technologies before, from the printing press and electricity to nuclear weapons. AI won’t be the last. Perhaps the more important question is how we learn to manage these transformations when our tendency to create them seems so deeply rooted in who we are.
This argument is conflating the bad boss with the misattribution of the tool. You could do that for any tool throughout history. But I still fail to see how you go from “bad boss is bad because they make us misattribute tool” to “good boss is bad because he’s not wallowing with me about how bad boss is bad”.
As anyone who has known about him for more than a couple of years knows, Linus has a pretty solid track record of calling out stupidity. It’s just virtually always stupidity that affects the kernel. If he had to stop and tell bosses at other companies how to operate, Linux never would have made it out of the 90s.
I don’t see how this means he’s wrong though, if people are hating AI as a proxy for hating their abusive relationship with their employer, that doesn’t seem like a strong argument that everyone who isn’t in that situation must also hate AI.
It doesn’t mean he’s wrong! it means he’s out of touch and doing the propaganda work of capital for free much like many other self-employed/business owning/petit-borgeois people who think they’re on the side of capital.
It also means that his praise for and indeed positive experience with this “tool” is largely irrelevant to the broader social/economic discourse surrounding so called AI.
Blind hatred for AI as a technology is inherantly reactionary, but that doesn’t mean that the implementation of the technology today isn’t problematic in many ways.
Yeah I notice he doesn’t address the costs, economic and environmental, of LLM proliferation. Way to ignore a large part of the argument while calling anyone opposed to your position stupid.
I hope he gets to eat those words one day, but I doubt he would ever admit he was wrong.
@TwiddleTwaddle@lemmy.blahaj.zone
I hope you were shamed by this reply, on the use of commas in a subordinate clause.
Hahaha my writing style is runon and compound complex sentences without commas.
Oh my god a reasonable take for once. Thank you for putting into words why his heel-turn has bothered me so much
It’s funny, the c/fuckai mod that banned me a while back said the same thing. Iirc it was “How very reasonable of you. Banned”
One should look at China when they’re concerned about the USA approach.
Then the picture would be significantly changed if the cost of running useful LLMs goes down to what it costs to acquire a reasonable computer to work on the Linux kernel, no? That is if we could run useful LLMs for kernel programming on our kernel programming computers, and we own those LLMs. In such a world, capital would have no stronger leverage over us than it already does by selling us the computers we need to program. Does this argument make sense?
It also means that his praise for and indeed positive experience with this “tool” is largely irrelevant to the broader social/economic discourse surrounding so called AI.
Yeah, but tbf all of the statements he’s making about it are in the context of organizing linux development and the most effective way to carry that out, and just happen to be public and so end up being widely republished and incorporated into discussions with a broader topic than that.
Yeah, I don’t think Linus should be policing his speech at this level or for this cause. Educate the workers please, Linus is doing plenty.
Shush! They are not supposed to know that the AI–Anti-Ai fight is fabricated bullshit with both sides being equally stupid (like people arguing based on ideological zeal always are…). And that it’s just a show so that the majority is occupied with watching the mud fight so they don’t think about the actual problem.
What’s the point of this argument though? Sure it’s real for every profession and every tool but what’s the agentic choice we ought to pursue here? Whine against AI or actually lead society with policy? As “ban llm” is not a valid policy that would ever work so the only viable path is regulation and growing of culture where AI is used for good, no?
Embrace AI, work towards open weight/free software and protect worker rights. Society wins at the bleeding edge not at conservstism.
The answer is certainly not to embrace “AI” in it’s current forms or allow for big tech monopolists to continue their wasteful and destructive infrastructure buildout. The oligarchy isn’t going to regulate itself to worker self-determination either. Nothing short of the death of capitalism will do.
Nothing short of the death of capitalism will do
Ok, let me know how that goes. While you whine online and have no impact on reality we will embrace “AI” and drain billionaire funds for distillation, open weights and building free software.
You genuinely don’t see the effective differences of thse two strategies? Actually doing something vs having your finger crossed some magical boogeyman called “capitalism” will collapse?
So he lacks empathy is what you are saying.
I guess I have that luxury, but I sacrificed hard to get where I’m at.
And, truthfully, way more people could do it if they didn’t feel entitled to the easy way (college > world destroying corp).
