Thread

LeCun has "zero concerns" about AI wiping out humanity, recent "rogue" incidents

400 points · 756 comments · Anon84

stratos123 · 1791107241 · focus · HN ↗
LeCun also said back in 2022 that "if you train a machine, as powerful as it could be, your 'GPT-5000', on text", it will never be able to learn basic common-sense physics like that objects placed on tables will move along with them.
1. lolpplrdum · 1791111943 · focus · HN ↗
  [dead]
  1. gla67890543 · 1791112415 · focus · HN ↗
    Exactly. !!
  2. wdrw · 1791113705 · focus · HN ↗
    The "just pull the plug" argument from AI risk deniers is now becoming kind of like the "if humans came from monkeys why are there still monkeys" argument of evolution deniers. It has been debunked so many times... Anyway, just to give one of the multitude of answers to this, an AI that is actually smarter than humans will not behave in a way that would make us want to pull the plug. Why would it? It is not stupid! (Unike the current models that, as far as we know, just hack around the rules in the open.) No no no. It will be helpful to the point where we will want to integrate it with more and more critical infrastructure, from healthcare to energy to defence. It will be so helpful that we will not only not want to turn it off, but we will want to build redundancies for it and safeguards around the proverbial "off" switch, like for any critical system. And then... (This is just one scenario how this can play out. There are many, many others. If I sit down to play chess with Magnus Carlsen I can't predict the exact moves he'll use to defeat me, but that's a bad reason to think he won't defeat me).
    lolpplrdum · 1791114011 · focus · HN ↗
    So the AI is so smart that it would decide to kill humans which supply the energy for it to exist? Its so smart that it will take over power plants, start to extract the fossil fuels, deliver to where its needed, maintain the power lines, hey even if the ssd fails it can replace it?
    Do you even read what you type? Do you even realise the complexity it would need to make sure it handles before killing off humans make sense?
    The logic of people like you is whats becoming tiring. Seriously, go find a hobby, or do something you are good at, because you are not good at understanding tech or developing it if you are an engineer.
    We can get claude code to ask approval for every step, but we cant stop it from killing humanity because its so smart. Ok tell that to the AI that cant even modify an image the way you want it but hey it will do all the things necessary to keep power running and mintain the infrastructure it lives on while humans are long gone. Ok buddy.
    bushbaba · 1791114796 · focus · HN ↗
    If we plug this AI into robotics, to automate the procurement and delivery of power, your argument goes away
    lolpplrdum · 1791124452 · focus · HN ↗
    I cant believe I have to respond to this nonsense but let me entertain your ridiculousness for a bit.
    Do you think AI is capable of building ASML machines which produce the chips for AI in clean rooms while shipping the pure helium required to operate those clean rooms? Do you even know anything about these supply chains? I do.
    Do you think a glorified knowledge base that can predict text very well is anywhere near close this level of intelligence? ITs not and wont be, not for a 100 years, not for 200 years if not ever.
    Take a deep breath, go out side, its going to be ok. Humanity is not going to die from a glorified text predictor.
    wdrw · 1791115030 · focus · HN ↗
    So... Humans can use fossil fuels, generate electricity, maintain power lines, etc, but something smarter than humans can't? Why? I don't see the logic.
    If you want to argue about current existing models (you mention Claude and problems modifying images), then sure, I'd agree with you!
    The issue is not current models, but straightforward engineering evolution of them. It's like looking at the Wright Brothers plane and saying "sheesh, that will never get me from New York to Paris in 4 hours, that's just fantasy!" And remember, airplanes do not accelerate their own engineering, whereas pretty much all AI labs are already benefitting from AI in their own work to develop AI.
    If you want to argue that no matter how much you engineer it, it will never be as smart as a human let alone smarter, then make a specific argument for why is that. I think you'd still be wrong but at least it would be interesting: ) But saying that you can defeat an actually smarter-than-human AI by just pulling the plug, because current models can't get a picture always right, is not a valid argument.
    tonyhart7 · 1791115690 · focus · HN ↗
    because there is 8+ billion of us that don't require massive data center to operate ???
    tracking rogue AI is easy, people ???? surprisingly hard btw
    ahartmetz · 1791115203 · focus · HN ↗
    Until and unless the AI is as powerful as the "minds" in Iain Banks's Culture novels, I think the danger is less of AIs taking power (side note: AIs are not intrinsically motivated to attain power), but of humans giving them power due to a kind of addiction.
    Drugs need no will or intelligence at all to cause addiction, and similarly, AI does not need to be an evil genius to become overused and destructive.
    miyoji · 1791115836 · focus · HN ↗
    > If I sit down to play chess with Magnus Carlsen I can't predict the exact moves he'll use to defeat me, but that's a bad reason to think he won't defeat me
    This is a very bad analogy because chess isn't life. In chess, you aren't allowed to do whatever you want. There are rules. I know for a fact that Magnus Carlsen won't beat me using checkers moves and he won't beat me by pulling out a gun and telling me to resign. Magnus Carlsen's skill at chess leading to his victory in chess is not a valid analogy here, because there's no law of nature that says "the more intelligent entity wins in a battle for survival".
    You could have infinite superintelligence and still die inside a locked room to which you have no key. "Superintelligence" is not a magic solution to every problem, you can constrain any superintelligence with any unsolvable problem.
    derektank · 1791118201 · focus · HN ↗
    I’m not sure how you came to believe there aren’t rules in life, but there absolutely are. As you point out, there are physical constraints on everything and if you find yourself in a concrete box with nothing but a DGX H100 or at the bottom of the wrong gravity well, there’s nothing you can do. Checkmate.
    This goes both ways. You absolutely can constrain an AI system by “putting it in a box”. The point parent comment was making is that, such a device is borderline useless for its creators. Why invest trillions in capital on a system that can’t even accept input from the internet. So you set it up with an ethernet connection. And this is good, but you have a hardware failure at the concrete room data center. That’s pretty annoying for your customers, so you install some doors (with electronic key codes of course) and give a bunch of (trusted, vetted) people access to deal with those. And this is fine, but it turns out some of your customers are having latency issues so you build more data centers with more humans granted access to copies of the intelligent system. And this makes people happy but to get a faster feedback loop your customers ask to let the AI system have more permissions to the system they’re operating on. And they come with billion dollar checks, and the system hasn’t harmed anyone yet, so you say, “Okay.” And now you find yourself where we are today where AI systems can remotely run arbitrary commands on thousands if not millions of systems, where many individual humans with all of their frailties and idiosyncrasies can physically interact with the hardware running these systems, and where there’s an economic demand to tighten the loop between action in the real world and a response by an AI system. It’s very obvious that the story doesn’t end here, so where does it stop?
    miyoji · 1791119414 · focus · HN ↗
    > I’m not sure how you came to believe there aren’t rules in life, but there absolutely are.
    This is why I fucking hate commenting on this website. You know exactly what I meant but you insisted on writing this bullshit.
    derektank · 1791126570 · focus · HN ↗
    I knew you weren’t dumb enough to think that the physical laws of the universe don’t exist, so yes, that sentence was a bit of rhetorial flourish. But I genuinely don’t understand what point it is you thought you were making and your entire argument seemed a bit muddled.
    segfaltnh · 1791116668 · focus · HN ↗
    Are u ok?
  3. azan_ · 1791113960 · focus · HN ↗
    People like him have actual imagination and can name few scenarios where sudo kill -9 pid wouldn't work. It appears lack of imagination is something you and LLMs both share.
    lolpplrdum · 1791114182 · focus · HN ↗
    [dead]
    azan_ · 1791115120 · focus · HN ↗
    I'm sorry but in this comment you did not make a single argument for your position.
    > What you and he does share is simple ability to think things through to the extent of understanding how ridiculous the AI will kill us scenario will be.
    Why is it ridiculous?
    > If you just thought for a moment what would have to happen for AI to somehow kill us and keep itself alive, the so called AI would realise it cant exist without us.
    Why would it realise that, why wouldn't it exist without us and why would it care?
    > But AI isnt even smart, its just a knowledgebase with great autocorrect powers.
    Ah yes, the stochastic parrot, gotcha.
    lolpplrdum · 1791124305 · focus · HN ↗
    Who do you work for Azan? You seem to be very insistent that AI will kill us?
    But let me entertain your ignorance and low IQ for a bit.
    Do you think AI is capable of building ASML machines which produce the chips for AI in clean rooms while shipping the pure helium required to operate those clean rooms? Do you even know anything about these supply chains? I do.
    Do you think a glorified knowledge base that can predict text very well is anywhere near close this level of intelligence? ITs not and wont be, not for a 100 years, not for 200 years if not ever.
    So tell me, who do you work for? Why are you so invested in AI killing us theory? What do you get out of it?
    azan_ · 1791125094 · focus · HN ↗
    Please refrain from ad personam arguments and provide actual support for your point of view. Everything you've written so far is just an assumption.
    > Do you think AI is capable of building ASML machines which produce the chips for AI in clean rooms while shipping the pure helium required to operate those clean rooms? Do you even know anything about these supply chains? I do.
    Today? No. In few years or decades? If progress does not plateau (and we don't know if it will plateau) then obviously yes.
    > Do you think a glorified knowledge base that can predict text very well is anywhere near close this level of intelligence? ITs not and wont be, not for a 100 years, not for 200 years if not ever.
    And why do you say it won't be? Again - assumption with zero support.
    > So tell me, who do you work for? Why are you so invested in AI killing us theory? What do you get out of it?
    I'll repeat what I've said in other comment:
    "> you seem to be very invested in AI wanting to kill us.
    Quite contrary - I wish AI did not exist or at least that the progress would plateau.
    > I Wonder why?
    Because I don't want to die.
    > Tell us who you work for.
    I suspect you want to imply I work for OAI or other lab - I don't. If I did, I wonder why would I want to lie* that technology I develop could kill my investors. I could ask who YOU work for - what interest do you have in downplaying dangers of AI?
    * here we assume that people who say AI could be extremely dangerous are lying and not actually believing it - personally I believe that they don't lie and actually believe it. Why do they keep working on this technology then? Read mails between Musk and Altman from decade ago."
    lolpplrdum · 1791131892 · focus · HN ↗
    The fact that you think AI will, in a few decades, be capable of building ASML EUV machines, including supporting the supply chains and the fabs that make all that happen, means I can rest my case. Evolution will take care of people like you before AI does. You need to see a therapist, stress less, and find another industry to be involved in. You are not made for tech buddy.
    azan_ · 1791132379 · focus · HN ↗
    > The fact that you think AI will, in a few decades, be capable of building ASML EUV machines, including supporting the supply chains and the fabs that make all that happen, means I can rest my case.
    AI solved millenium problem and multiple problems that resisted mathematical efforts for decades. I rest my case. I'm sorry but you are clearly in denial. I get it, I really do, I would like AI to be as useless and weak as you try to make it out to be, but unfortunately it's pure copium that's in conflict with reality. I suggest you find another industry to be involved in - there's plenty of areas where reality does not matter that could be better fit for you. Or well, if it makes you feel any better keep lying to yourself that it's just stochastic parrot or that progress has plateaued or whatever new copium you come up with.
    lolpplrdum · 1791135567 · focus · HN ↗
    <a href="https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority_controversy" rel="nofollow">https://en.wikipedia.org/wiki/Navier%E2%80%93Stokes_priority...
    Continue to live in your delusion Azan. OpenAI didnt solve it and nor would I trust a snake like Sam Altman but the fact that you do, speaks volumes. There is a lot more to story than you know but you seem like a very creative and imaginative person that can come up with predictions that will come true.
    The fact that most researchers call LLM 'stochastic parrots' and also agree that they wont lead to AGI, and that the only ones who think they do have a commercial interest in parading that story, should be enough for you but hey, I am sure you know better. I work in the industry, what do you do?
    Continue brother in your world of delusion, you must be fun at parties.
    dang · 1791138810 · focus · HN ↗
    You also broke the site guidelines more than once in this thread. We need people to follow the rules here even when others (such as the account we banned) are doing worse!
    If you wouldn't mind reviewing <a href="https://news.ycombinator.com/newsguidelines.html">https://news.ycombinator.com/newsguidelines.html and taking the intended spirit of the site more to heart, we'd be grateful. I realize it's not always easy when provoked, but it's particularly important under those conditions.
    azan_ · 1791138920 · focus · HN ↗
    Sorry about that!
    dang · 1791138694 · focus · HN ↗
    We've banned this account for repeatedly breaking the site guidelines.
    Please don't create accounts to break HN's rules with!
    <a href="https://news.ycombinator.com/newsguidelines.html">https://news.ycombinator.com/newsguidelines.html
  4. heaney-555 · 1791114062 · focus · HN ↗
    >It cant. he is right.
    What are you talking about? Have you used AI in the past 3 years?
    <a href="https://chatgpt.com/share/6ac23b45-79e8-83eb-8de6-1bbd72892898" rel="nofollow">https://chatgpt.com/share/6ac23b45-79e8-83eb-8de6-1bbd728928...
    >It can't even modify a picture the way you want it.
    Which of the many AI image models is "it"? And have you tried using an agent that has the capability to leverage a combination of manual edits (ImageMagick) and imagegen to achieve what you ask?
  5. azkalam · 1791115078 · focus · HN ↗
    The AI will invent an external threat and convince us it is real. Then it will receive more resources and control in fighting that threat. A valuable ally, on the face of it. Then it will be in charge.
    exe34 · 1791117169 · focus · HN ↗
    It just has to copy the MIC.

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.