222

AI godfather Geoffrey Hinton: ‘We’re creating alien beings. I’m willing to be selfish and choose humans’

#technology
the Guardian AI godfather Geoffrey Hinton: ‘We’re creating alien beings. I’m willing to be selfish and choose humans’ The pioneering computer scientist on why Trump is using ‘superintelligence’ wrong, whether AI could wipe out humanity – and how our biggest challenge will be accepting we’re not the only intelligent beings www.theguardian.com

check post at origin

89 comments

log in to comment.

  • @Zwuzelmaus@feddit.org · 14 Oct 2026

    And now he says we’re going to call AI ‘superintelligence’. You can’t make it smarter than people just by calling it smarter than people.”

    I actually laughed out loud. 😄

  • @nonentity@sh.itjust.works · 14 Oct 2026

    It’s so demoralising to perpetually witness apparently intelligent people be convinced that Algorithmic Iterators can in any way be mistaken as a form of intelligence. I’ve struggled for decades to relate to the world; but this recent, collective debasement of rationality has nearly obliterated the few remaining foundations suitable to build hope upon.

    • @red_tomato@lemmy.world · 14 Oct 2026

      AI is about making machines that mimic human level intelligence. Whether they possess actual intelligence doesn’t matter. They just need to act intelligently.

      You can disagree with the terminology, but it has been an active field of study since the 1950s.

      • @TronBronson@lemmy.world · 14 Oct 2026

        Remember the Turing test? Damn goal posts move faster than the ball. I’m not sure how one could fail to be impressed by the technological advancements in the last 3 years in the field

        • @schipelblorp@sh.itjust.works · 14 Oct 2026 · edited

          Sort of.

          I think Turing thought, “well something that could do this would surely HAVE to be intelligent”, but I seriously doubt that he would have waved the white flag for an intelligence that can convincingly mimic human communication but can’t count letters in a word.

          • @SreudianFlip@sh.itjust.works · 14 Oct 2026

            Well the Test now includes casual tasks like counting letters, doesn’t it? You have to move your head in a certain way to spot the leopard in the shaded forest.

    • @Eheran@lemmy.world · 14 Oct 2026

      Define intelligence.

      • @nonentity@sh.itjust.works · 14 Oct 2026

        Comprehension, something LLMs are intrinsically incapable of demonstrating, let alone possessing.

        • @nialv7@lemmy.world · 14 Oct 2026

          you are just replacing one underspecified concept with another. what’s “comprehension”?

          • @nonentity@sh.itjust.works · 14 Oct 2026

            Literal dictionary definitions:

            Comprehension: the ability to understand something.

            Understand: perceive the intended meaning of something.

            Things LLMs are categorically incapable of embodying, and anyone who believes otherwise is demonstrating an innate challenge of grasping themselves.

          • @h0tbeef@lemmy.zip · 14 Oct 2026

            It’s what you don’t posses about the workings of LLMs (or basic English).

            www.merriam-webster.com/dictionary/comprehension

            • @p03locke@lemmy.dbzer0.com · 14 Oct 2026

              Jesus fuck… did you actually read your definition?

              1. the act or action of grasping with the intellect

              Okay, we’ve replaced the concept of “intelligence” and just swapped it out with “grasping with the intellect”.

              1. knowledge gained by comprehending

              Nice, using the verb word of “comprehending” to define “comprehension”.

              1. the capacity for understanding fully

              Okay, maybe that’s a start, but it’s incredibly vague.

              Also, clicking on the links just resolves back to the most generic circular definition possible. Remind me to never use Merriam-Webster.


              At least this gets further: en.wiktionary.org/wiki/comprehension

              If we settle on “A thorough understanding” and define understanding as… fuck, we’re back to this circular bullshit again.

              Okay, fine, let’s just look at intelligence:

              The capacity of mind, especially to understand principles, truths, facts or meanings, acquire knowledge, and apply it to practice; the ability to comprehend and learn; the ability to process sentient experience to generate true beliefs with a justified degree of confidence.

              Breaking this down:

              • Understand principles, truths - Sure, sounds like an LLM to me. If you ask it to define a scientific principle or fact or Newton’s laws, it can tell you what it is.
              • facts or meaning - And, depending on the kind of thing we’re talking about, it can you tell you why it exists.
              • acquire knowledge - While it’s not permanent (because context are just thrown away like short-term memories), LLMs have MCPs and other tools to acquire new knowledge.
              • apply it in practice - It takes the knowledge it has naturally (training data), acquires new knowledge from its tools, and then applies it to the task it’s looking at.
              • the ability to comprehend and learn - Here we go with “comprehension” again. But, let’s assume we all know what “learn” is. A LLM can learn about things, but the main problem is the context size and how we throw that away each session. But, that can be solved by storing facts elsewhere, Markdown rules, other techniques to condense the immediate context into long-term storage. In humans, we call that “long-term memory”. We have our own process to stash short-term memory into long-term memory. We just haven’t experimented enough with the long-term storage part in LLMs.
              • the ability to process sentient experience - sentient… well, that’s a pretty weighted word. I also feel like there’s a clear difference between intelligence and sentience. We’re now entering the realm of sensation, thought, and feelings, which are entire different parts of the brain.

              I also don’t think “sensations” are impossible to emulate, either.

              • @h0tbeef@lemmy.zip · 14 Oct 2026

                Words have meanings, and you don’t get to change them in order to strengthen a weak argument.

                I’m sorry that those meanings aren’t convenient for your untenable position.

            • @nialv7@lemmy.world · 14 Oct 2026

              anyone can quote a dictionary without comprehension.

              • @h0tbeef@lemmy.zip · 14 Oct 2026

                I didn’t quote it, I sent you the entry in the hopes that you might read it yourself and learn something

        • @TronBronson@lemmy.world · 14 Oct 2026

          I’m pretty sure comprehension was like the original success of LLMs, and the responses were a bit wonky. When was the last time an LLM failed to comprehend?

          • @Fishnoodle@lemmy.world · 14 Oct 2026

            They never comprehend. They ingest information and find close approximations for responses, but they don’t comprehend any part of the process. If they did, they wouldn’t hallucinate complete bullshit answers and spit them out like they were fact

            • @SreudianFlip@sh.itjust.works · 14 Oct 2026 · edited

              Oh, so, like my coworkers then.

              /s

            • @encelado748@feddit.org · 14 Oct 2026

              To be honest human spew bullshit all the time when they do not understand something. Or people are confidently incorrect all the time as they are remembering incorrectly. The amount of bullshitting is more a meter for intelligence, not something that discriminate between intelligence or the lack thereof.

              • @DudeImMacGyver@kbin.earth · 15 Oct 2026

                You are anthropomorphizing LLMs: This is a mistake. They are not people and they do not think, reason, or feel. They are completely incapable of these things. Even calling it "AI" is misleading at best.

                • @encelado748@feddit.org · 15 Oct 2026 · edited

                  I am not. I want to understand what is the difference between “feeling” as result of electrochemical reactions in our brain and “feeling” as result of math performed on a numerical matrix. I am perfectly aware those are not the same, but the result may be similar, and what we care of the result determines the conclusions we can make.

                  You need to clearly explain what is thinking and what in the “thinking” process of LLM is not thinking. If you define “thinking” as electromechanical reactions in human brains then LLM is not thinking, but it is not a useful definition.

                  • @DudeImMacGyver@kbin.earth · 15 Oct 2026 · edited

                    I don't need to explain shit, you're the one made the claims and didn't provide proof. The burden of proof is on you. We can (and should) dismiss your baseless unproven claims about "AI" outright.

                    • @encelado748@feddit.org · 15 Oct 2026

                      Do you even know what a claim is?

                      they do not think, reason, or feel. They are completely incapable of these things.

                      This is a claim. You made a claim. My assertion that you need to substantiate your claim definitionally does not constitute a claim.

                      continue reading (4 more)
              • @mysticalone@lemmy.world · 14 Oct 2026

                This gives me “are you smarter than a fifth grader” vibes

            • @TronBronson@lemmy.world · 2d ago

              Silence luddite lmaoo. All of ur jobs r mine

          • @schipelblorp@sh.itjust.works · 14 Oct 2026

            LLM’s comprehend statistical correlations. If you feed it every chess game, it can tell you which spaces the knight is likely to move to, and which it very unlikely to move to, but it does not understand how the knight moves.

            You might say that’s a distinction without a difference, but programs can actually know thing, like a chess program actually does know how a knight moves, and it does so with minimal electricity.

            • @skibidi@lemmy.world · 14 Oct 2026

              There’s some semantics here.

              Large neural networks have been shown to build parallel models of the world that they learn through their training data. A chess model trained only on chess move notation does have an interior board and state that it tracks, despite never being told about a board, the number of pieces, how they move, etc.

              LLMs do something similar with language, mapping tokens to ideas in ways that can produce meaningful and grammatically correct output without needing to be formally taught any of this.

              Our brains also do things like this - we observe X and then try to intuit Y, the truth of the world and reason for X.

              Where this breaks down is humans can then take our guess for Y and propose Z, and try to find it out in the world. If we can’t, then Y is probably wrong, and our new understanding of the world is Y* that is consistent both with X and the absence of Z.

              I cannot stress enough how far we are from machines that can do this. They can learn to approximate Y through training, but they have no means to extend that to propose an unobserved Z, no means to update their internal state given the results of that observation (without fully retraining the model on new data - which makes it a new model), and no way to question X given Z - they can talk at great length about data biases but can literally never excise bad data from their training weights. There have been experiments with training models on time-boxed datasets (e.g. stuff up to AD 1880) and asking it about modern concepts - predictably hilarious results.

              All the current ‘AI’ stuff is smoke and mirrors around this very hard problem. The machines can talk, but they can’t think (Z) and can’t learn (Y*).

              • @p03locke@lemmy.dbzer0.com · 14 Oct 2026

                they can talk at great length about data biases but can literally never excise bad data from their training weights

                And you can’t excise 20 years of Fox News from some boomer conservative. They can literally never back down from their biases.

                Training data is a lot more permanent than context, but at the same time, it’s not absolute, either. I’ve seen plenty of thinking blocks where it knows about X from its training data, but perhaps it’s out-of-date and should check other sources. It has the web tools, finds newer information, processes it as part of its problem set, and solves the problem with newer information.

                The core issue is we throw all of this data away when we’re done with the problem! A LLM only has so much context, so it starts to compact after a while. And even if it doesn’t, we voluntarily chuck that shit out the window, as soon as the session ends.

                It’s like Lenny from Memento. You got a baseline person who constantly believes he’s got to find his wife’s killer. And even when he tries to supplant his lack of long-term memory with photos and tattoos, it’s a shit system with shittier outcomes. Also, he doesn’t get to choose when his memory disappears. It just happens with some kind of stress trigger. Because of how flawed Lemmy is, do we call him somebody who can’t think and can’t learn? Is he intelligent? Is he even sentient?

                Though, in the case of an LLM, it doesn’t have to be set up this way. We can still stash a long-term version of memories as rules and other permanent files. In fact, this is encouraged. Everything starts with a system prompt, a set of rules, and the user prompt. All of that is adjustable.

                And yes, you can train out biases. Rather easily, actually. You don’t even have to adjust weights. You post-train through rules files.

                But, in terms of philosophical questions, can we call LLMs intelligent if we can edit their memories at a whim? Is a brainwashed boomer intelligent because Fox News hacked their memories and biases? Is Lenny intelligent because he can’t form new memories?

                You have to remember that just about every argument for or against LLMs can also apply to humans, and you can’t really arrive at definitions without clearly separating what doesn’t apply to LLMs and does apply to humans, in all cases.

                • @skibidi@lemmy.world · 14 Oct 2026 · edited

                  And yes, you can train out biases. Rather easily, actually. You don’t even have to adjust weights. You post-train through rules files.

                  The assertion is not about whether ‘you’ can do it. It is that the model cannot do this itself.

                  A human recognizing a bias in the model and correcting it is not evidence robot intelligence, it is further evidence for human intelligence.

                  Training is absolute, it determines the set of relationships the model can represent between input and output. Context, rules files, system prompts, etc. operate only on the conditional probabilities within this set and are impermanent. This is ‘priming’ in human psychology, and why a person in a herpetology class will talk about Pythonic features differently than in a coding class.

                  As an example of how this is not real learning and is inadequate to solve real problems, here is a paper where the authors challenged multiple models to solve unfamiliar games where they need to discover the rules through play and feedback. Even the ‘best’ model performed pitifully compared to humans.

            • @TronBronson@lemmy.world · 25 Oct 2026

              I really feel like yall spend too much time focusing on the how, and not the results. Was the analysis of chess moves useful? Great then it’s a good tool. Even better you found the limitation of the tool. That would make you a better user of the tool.

              I approached the topic asking “ can the tool do real world work?” Or “can I use the tool to benefit myself?” Or even “ am I enjoying learning how to use the tool”

              A piece of software that cost $20 a month and can help me create a game I can play and share with my friends in under 10 days. Pretty cool if you ask me. I learning how to make a video game felt good, the result is real, I’m going to make another one.

              And to be fair they know exactly how the game works… they trained my ais and programmed my game. I don’t know how my game works.

              • @schipelblorp@sh.itjust.works · 25 Oct 2026

                You gotta use the right tool for the job. Using AI to play chess is like using a sledge hammer to drive a finishing nail.

                For programming, LLM AI is a great tool, no doubt.

            • @p03locke@lemmy.dbzer0.com · 14 Oct 2026 · edited

              If you feed it every chess game, it can tell you which spaces the knight is likely to move to, and which it very unlikely to move to, but it does not understand how the knight moves.

              Well, it knows how the knight moves, but it might forget that’s it’s an important rule after it starts to lose context. That’s also how LLMs start to lose track of what’s actually on the board.

              There are definitely differences between LLMs and Neural Networks, but at the end of the day, they are both just gigabytes of weights. And really, both types are similar to what the human brain possesses. People tend to forget that the human brain isn’t just some monolithic thing. There’s a lot of difference specialized pieces. If you ask Magnus Carlsen how he calculated the best move on the chess board in a blitz game, he literally cannot tell you! It just appears, because that part of the brain is trained and hardwired to calculate good moves, based on the current board state.

              A LLM uses logic and reasoning to arrive at a solution. That’s not really compatible with intense pre-trained calculation. You can’t talk yourself into a chess move at the speed the game requires. You can quickly think about possibilities, but you have to already have pre-trained calculations to get at least 90-95% of the way there, to pick from a few different lines.

              On the other hand, Neural Networks beat the shit out of the greatest grandmasters in chess. Hands down. On average, they are at least +500 ELO to the best GMs. And everybody in the chess world knows this, which is why everybody uses these tools outside of 1v1 human to human games. (Obviously, cheating is taken very seriously, and has made plenty of headlines over the years.) Every GM analyzes their own games using those engines, to pick apart mistakes.

      • @gandalf_der_13te@feddit.org · 14 Oct 2026 · edited

        pretty sure it comes from latin “inter-lego” (collect or pick out information, like you would pick out the good fruit from a bush with good and bad fruit).

        en.wiktionary.org/wiki/intelligent

        so it’s about filtering what you want to have from what you don’t want to have. i.e. a mathematical function f : X -> R which maps an object x to a real number to figure out how desirable that object is, i.e. assign value to it.

        technically you don’t even need neural networks for it, sorting algorithms employ this kind of function all the time to sort objects according to some criteria.

    • @MyVeryRealName@lemmy.world · 14 Oct 2026

      Why does the result of algorithmic iteration seem unintelligent to you?

    • @CovfefeKills@lemmy.world low rep · 14 Oct 2026

      You are assigning a magical meaning to the word intelligence. Your objection is superficial semantics. You are effectively being a science denier proudly displaying your ignorance as evidence. This anti-vax tier anti-ai circlejerk of ignorance as evidence is just the opposite side of the same ignorant coin. You are going to get pissed on by researchers and the actual smart people when are you are anti-ai like this you are not changing minds you are ignored. Seriously move off semantics it’s just fucking embarrassing.

      • @nonentity@sh.itjust.works · 14 Oct 2026

        Your bulk discount tokens must have just hit your account, I’m honoured you chose to spend them replying to my post.

        It’s always fascinating when an entity exhibits such a strong reaction to a critique of associating LLMs as intelligence, what was the prompt you engineered to conjure it?

        I’ll concede my perspective of LLMs is limited, and skewed by agnostic ignorance as I’ve never actively, willingly, or consciously used one on purpose, but to accuse me of wilful ignorance smells more like Lynx Admission.

        I’m not interested in the opinion or approval of anyone whose fallen for this grift, too much of my time and energy has been exhausted appeasing them. I’ll reengage with them once they wake up, or are excreted from the ‘AI’ logo sphincters, until then I need to relegate them as dark matter.

        • @CovfefeKills@lemmy.world low rep · 14 Oct 2026 · edited

          I’ll concede my perspective of LLMs is limited,

          but to accuse me of wilful ignorance smells more like Lynx Admission.

          I like the way you have no ability to assert your opinions and just send insults it really highlights your character and my heart is moved. You can be a better person who intelligently argues their point or you can be a MAGAt tier anti-vaxx tier science denier and fight with your feelings. Guess which one I want for you? Move off semantics it’s fucking embarrassing.

    • @TronBronson@lemmy.world · 14 Oct 2026

      I enjoy speaking to the AI more than humans on most topics I find worthy of discussion.

      • @nonentity@sh.itjust.works · 14 Oct 2026

        LLMs are impressive inversely proportionally to the level of understanding and proficiency one has in the field they’re deployed. They’re also progressively corrosive to competency and credibility with one’s exposure.

        All I can see is their lobotomisation of the world, and I’m sorry you’ve fallen victim to them.

        • @TronBronson@lemmy.world · 25 Oct 2026

          I always wonder if it’s different learning styles or lack of intellectual curiosity. The one point that I’m certain on is the ego just cannot fathom itself losing its intellectual position in the world. The technologically illiterate and the technological gatekeepers both are interestingly aligned on this topic.

          Like imagine, talking to something that wasn’t rude and dismissive, and also could articulate a point that could be debated? So much wasted Internet space on elaborate ad hominem

      • @mrmaplebar@fedia.io · 14 Oct 2026

        Sounds like something to talk to a (human) therapist about.

        • @TronBronson@lemmy.world · 25 Oct 2026 · edited

          Would you believe it if I told you rural America did not have much to offer in regards to mental health care?

      • @frongt@lemmy.zip · 14 Oct 2026

        Speaking to even a brick wall would be better.

        • @TronBronson@lemmy.world · 25 Oct 2026

          Baby boy I got a home grown qwen that does a perfect Theodore Roosevelt, and I got 15 fucking morons in my Lenny inbox

      • @h0tbeef@lemmy.zip · 25 Oct 2026

        That’s phenomenally depressing

        • @mysticalone@lemmy.world · 25 Oct 2026

          I bet he can set the bar lower

          • @h0tbeef@lemmy.zip · 25 Oct 2026

            Ahh, buckle up for the rest of the comment thread lmao

            • @TronBronson@lemmy.world · 25 Oct 2026

              Exactly show up 12 days later and not a single intellectually curious individual, or comment that was well thought out. Just a bunch of dismissive shit talk from people either about to lose their position in society or people who never had a position in society.

              • @h0tbeef@lemmy.zip · 25 Oct 2026

                Ahh yes, it’s the people who moved on with their lives after this conversation who aren’t curious.

                Certainly the person still ruminating about how chatbots are better to interact with than people 12 days after self-pwning on a Lemmy thread is intellectually superior

                • @TronBronson@lemmy.world · 2d ago

                  Did you create anything yet? I have 3 working games, an engineering project, and electrical engineering project in the last 23 days.

        • @TronBronson@lemmy.world · 25 Oct 2026

          I mean, read through this comment thread. Do you really feel like you guys are all the type of personalities anyone wants to have an intellectual conversation with?

          • @h0tbeef@lemmy.zip · 25 Oct 2026

            I have lots of intellectual conversations with other intelligent people.

            You cannot have an intellectual conversation with an inanimate object, it possesses no intellect.

            If you can’t find a human you want to have a conversation with, that says way more about you than it does about the rest of humanity. You’re experiencing a user error.

            • @TronBronson@lemmy.world · 2d ago

              Again, not a single thread of intellect anywhere on this technology sub. Not a single nuanced take about technology. I’m just enjoying the new tool I’m using to expand my knowledge and capabilities in computer science. you’re not an intellectual you’re an asshole.

        • @TronBronson@lemmy.world · 25 Oct 2026 · edited

          The collective charisma of a fucking rabid racoon pawing at your door

          • @h0tbeef@lemmy.zip · 25 Oct 2026

            Not sure if you meant to reply to me twice, but your sentence fragment doesn’t make any sense

      • @psx_crab@lemmy.zip · 25 Oct 2026

        This read like you just want people to agree with you.

        • @TronBronson@lemmy.world · 25 Oct 2026

          Oh. Well it should read like…. “I enjoy calculus, astrophotography, 30 year old games and sci-fi, politics, and novel technology, and I rarely find a person who enjoys two of those things to discuss in tangent.”

  • @username_1@discuss.tchncs.de · 14 Oct 2026

    our biggest challenge will be accepting we’re not the only intelligent beings

    But it isn’t challenging at all. All animals have some level of intelligence. Some of them even rather comparable to us. And the probability of existence of some aliens doesn’t sound too crazy, they might be quite clever too. See no problem in acceptance here.

    • @Zwuzelmaus@feddit.org · 14 Oct 2026 · edited

      See no problem in acceptance here.

      Mice? Super intelligent ones?

      • @username_1@discuss.tchncs.de · 14 Oct 2026

        I read some popular-science paper where they claimed that mice even have friends. That demands some nontrivial intelligence I think.

        Super intelligent

        But I suppose Trump is somewhat cleverer than a mouse.

        • @Zwuzelmaus@feddit.org · 14 Oct 2026 · edited

          I read some popular-science paper

          The mice that I was thinking about have been researched by the famous Douglas Adams, and they were by far the most intelligent beings on this planet, as long as they stayed here.

    • @queermunist@lemmy.ml · 14 Oct 2026

      That’s extremely challenging for a lot of people. A lot of people aren’t even willing to accept animals experience pain.

    • @h0tbeef@lemmy.zip · 14 Oct 2026

      I don’t accept the computer as intelligent or a being, and I love my computer (it’s objectively an inanimate object tho).

    • @TronBronson@lemmy.world · 14 Oct 2026

      Oh the people with jobs in peril, they are definitely in a state of denial about the state of AI. It’s easier to accept when your survival doesn’t depend on your ability to think and do things.

      On a funnier note the people who believe they are genetically superior to others and naturally smarter, those people probably won’t accept this at all

      • @eestileib@sh.itjust.works · 14 Oct 2026

        The people who think they’re genetically superior to other people seem to be the very biggest AI boosters of all, I’m not sure how to read your post here.

  • @melfie@lemmy.zip · 14 Oct 2026 · edited

    Not surprising that someone who evangelized artificial neural networks and algorithms like backpropagation at a time when people had largely written them off would have an inflated view of the technology’s potential.

    Edit:

    His peer, Yann LeCunn, seems to have a better perspective, basically stating that letting companies like Anthropic have a monopoly is the real threat to humanity: en.koreadaily.com/yann-lecun-ai-threat-humanity/

  • @prole@lemmy.blahaj.zone · 14 Oct 2026

    We’re creating alien beings

    No we’re not, shut the fuck up

  • @AceFuzzLord@lemmy.zip · 14 Oct 2026

    I’m willing to be selfish and choose humans

    Is the selfishness in the room with us?

    In all serious, choosing humans over AI of any type is always the correct answer.

  • @FauxLiving@lemmy.world · 14 Oct 2026

    Roko’s Basilisk was displeased at that.

  • @jh29a@lemmy.blahaj.zone · 14 Oct 2026

    This piece of crap technology is created by leonarci da vinto.

  • @InfiniteHench@lemmy.world · 14 Oct 2026

    Hey. It isn’t fair to slap that label on the entire city of San Francisco

  • @jaykrown@lemmy.world · 14 Oct 2026

    What we need are some regulations that are understandable and sensible.

  • @Darkonion@lemmy.world · 14 Oct 2026

    But if the current batch of ‘AI’ had delicious meat, would we still eat them?