Pioneering AI researcher Geoffrey Hinton joins Joel Hellermark for a candid, reflective conversation tracing the arc of neural networks, machine intelligence, and the future of AI research. Hinton explores his early inspirations, the creative collaborations that shaped deep learning, and the profound role of intuition and skepticism in scientific progress. The discussion delves into the mechanics of large language models, the promise of multimodal AI, and the parallels between brains and machines. With insight and humility, Hinton weighs the societal impacts, open questions, and the enduring value of curiosity-driven research on the road to human-level intelligence.
Watch the full episode above or read the transcript of their conversation below.
Have you reflected a lot on how to select talent, or has that mostly been like intuitive to you? It just shows up and you're like, this is a clever guy. Let's let's work together. Or have you thought a lot about that?
Can we really.
Should we should we roll this?
Yeah. Let's roll this. Okay, good. Yeah. Yeah yeah. Jumping into a repetition. Okay, so, Joe.
That song is working.
So I remember when I first got to Carnegie Mellon from England, in England, in a research unit, you would get to be 6:00 and you go, go for a drink in the pub, a Carnegie Mellon. I remember after being there a few weeks, it was Saturday night. I didn't have any friends yet, and I didn't know what to do.
So I decided I'd go into the lab and do some programing because I had a Lisp machine and you couldn't program it from home. So I went into the lab at about 9:00 on a Saturday night and it was swarming. All the students were there, and they were all there because what they were working on was the future.
They all believed that what they did next was going to change the course of computer science, and it was just so different from England. And so that was very refreshing.
Take me back to the very beginning, Jeff at Cambridge, trying to understand the brain. what was that like?
It was very disappointing. So I did physiology. And in the summer term, they're going to teach us how the brain worked. And all they taught us was how neurons conduct action potentials, which is very interesting. But it doesn't tell you how the brain works. So that was extremely disappointing. I switched to philosophy. Then I thought maybe they tell us how the mind worked and that was very disappointing.
I eventually ended up going to Edinburgh to do AI and that was more interesting. At least you could simulate things so you could test out theories.
And did you remember what intrigued you about it? I was at a Pittsburgh. Was that any particular person that expose you to those ideas?
I guess it was a book I read by Donald Hebb that influenced me a lot. he was very interested in how you learn the connection strengths in neural nets. I also read a book by John von Neumann early on, who was very interested in how the brain computes and how it's different from normal computers.
And did you get that conviction that this ideas would work out, at that point or what would was your intuition back at the Edinburgh?
This it seemed to me there has to be a way that the brain learns, and it's clearly not by having all sorts of things programed into it and then using logical rules of inference that just seem to be crazy from the outset. so we had to figure out how the brain learn to modify connections in a neural net so that it could do complicated things.
And von Neumann believed that Turing believed that. So phenomenon Turing were both pretty good at logic, but they didn't believe in this logical approach.
And what was your split between studying the ideas from from neuroscience and just doing what seemed to be good algorithms for for like how much inspiration that you take early on?
So I never did that much study of neuroscience. I was always inspired by what I learned about how the brain works, that there's a bunch of neurons, they perform relatively simple operations. They're non-linear, but they collect inputs, they weight them and then they give an output that depends on that weighted input. And the question is, how do you change those weights to make the whole thing do something good?
It seems like a fairly simple question.
What collaborations do you remember from from that time?
The main collaboration I had at Carnegie Mellon was with someone who wasn't at Carnegie Mellon. I was interacting a lot with Terry Sinofsky, who was in Baltimore. Johns Hopkins, and about once a month, either he would drive to Pittsburgh or I would drive to Baltimore. It's 250 miles away, and we would spend a weekend together working on Boltzmann machines.
That was a wonderful collaboration. We were both convinced it was how the brain worked. That was the most exciting research I've ever done, and a lot of technical results came out that were very interesting. But I think it's not how the brain works. I also had a very good collaboration with, Peter Brown, who was a very good statistician, and he worked on speech recognition at IBM.
And then he came as a more mature student to Carnegie Mellon just to get a PhD. but he already knew a lot. He taught me a lot about speech, and he, in fact taught me about Hidden Markov models. I think I learned more from him, and then he learned from me. That's the kind of student you want.
And when he told me about Hidden Markov Models, I was doing backprop with hidden layers, and then one called Hidden Layers, then and I decided that name they use, hidden Markov Models, is a great name for variables that you don't know what they're up to. and so that's where the name hidden in neural nets came from. Me and Peter decided that was a great name for the Hidden Valley, hidden as a neural nets.
but I learned a lot from Peter about speech.
Take us back to when Elia showed up, at your, at your office.
I was in my office. I was probably on a Sunday. and I was programing, I think. And there was a knock on the door. Not just any knock, but it went kind of an urgent knock. So I went and answered the door. And this was this young student there, and he said he was cooking fries over the summer, but he'd rather be working in my lab.
And so I said, well, why don't you make an appointment and we'll talk. And so we just said, how about now? And that sort of was just character. So we talked for a bit and I gave him a paper to read, which was the nature paper and backpropagation and we made another meeting for a week later, and he came back and he said, I didn't understand it, and I was very disappointed.
I thought he seemed like a bright guy, but it's only the chain rule. It's not that hard to understand. And he said, oh no, no, I understood that. I just don't understand why you don't give the gradient to essentially a sensible function optimizer, which took us quite a few years to think about. and it kept on like that.
We then he had very good. His intuitions about things were always very good.
What do you think had enabled those, those intuitions for for Ilya?
I don't know, I think he always thought for himself. He was always interested in AI from a young age. he's obviously good at math. So. But it's very hard to know.
And what was that collaboration between the two of you? Like? What part would you play and what part would Delia play?
It was a lot of fun. I remember one occasion when we were trying to do a complicated thing with producing maps of data, where I had a kind of mixture model, so you could take the same bunch of similarities and make two maps so that in one map, bank could be close to greed, and another map bank could be close to river.
because in one map you can't have it close to both, right? Because River and Guido and so we'd have a mixture maps and we were doing it in Matlab. And this involves a lot of reorganization of the code to do the right matrix multiplies. And then you got fed up with that. So he came one day and said, I'm going to write an interface for Matlab.
So I program in this different language. And then I have something that just converts it into Matlab. And I said, no, LDA. that'll take you a month to do. We got to get on with this project. Don't get diverted by that. But then he said, it's okay. I did it this morning.
that's that's quite, quite incredible. And throughout those, those years, the biggest shift wasn't necessarily just the algorithms but but also the, the skill. How did you sort of view that skill? over, over the years in.
You got that intuition very early. So Ilya was always preaching that, you just make it bigger and it'll work better. And I always thought that was a bit of a copout. You're going to have to have new ideas to it turns out he was basically right. New ideas help things like Transformers helped a lot, but it was really the scale of the data and the scale of the computation.
And back then we had no idea computers would get like a billion times faster. We thought maybe 100 times faster. We were trying to do things by coming up with clever ideas, would have just solved themselves if we'd had bigger scale of the data and computation. In about 2011, Ilya and another graduate student called James Martins and I had a paper using character level prediction.
so we took Wikipedia and we tried to predict the next HTML character, and that worked remarkably well. And we were always amazed at how well it worked. And that was using a fancy optimizer on GPUs. And we could never quite believe that it understood anything, but it looked as though it understood, and that just seemed incredible.
Can you take us through how are these models trained to predict the next word and why is it the wrong way of thinking about them?
Okay, I don't actually believe it is the wrong way. So in fact, I think I made the first neural net language model that used embeddings and backpropagation. So it's very simple. Data just triples. And it was turning each symbol into an embedding. Then having the embeddings interact to predict the embedding of the next symbol. And then from that predict the next symbol.
And then it was backpropagation through that whole process to learn these triples. And I showed it could generalize, about ten years later, Yoshua Bengio used a very similar that we can show to work with real text. And then about ten years after that, linguist started believing in embeddings. It was a slow process. The reason I think it's not just predicting the next symbol is, if you ask, well, what does it take to predict the next symbol?
Particularly if you ask me a question, and then the first word of the answer is the next symbol? you have to understand the question. So I think by predicting the next symbol, it's very unlike old fashioned auto complete, old fashioned auto complete, you store sort of triples of words. And then if you saw a pair of words, you see how often different words came.
Third. And that way you could predict the next symbol. And that's what most people think auto complete is like. It's no longer a tool like that. to predict the next symbol, you have to understand what's been said. So I think you're forcing it to understand by making it predict the next symbol. And I think it's understanding in much the same way we are.
So a lot of people will tell you these things aren't like us. they're just predicting the next symbol. They're not reasoning like us, but actually, in order to predict the next symbol, it's going to have to do some reasoning. And we see now that if you make big ones without putting in any special stuff to do reasoning, they can already do some reasoning.
And I think as you make them bigger, they're going to be able to do more and more reasoning.
Do you think I'm doing anything else than predicting the next symbol right now?
I think that's how you're learning. I think you're predicting the next video frame. you're predicting the next sound. but I think that's a pretty plausible theory of how the brain's learning.
What enables this models to learn such a wide variety of of.
What these big language models are doing is they're looking for common structure. And by finding common structure, they can encode things using the common structure. And that's more efficient. So let me give you an example. If you asked Gpt3 for why is a compost heap like an atom bomb? Most people can't answer that. Most people haven't thought they think atom bombs and compost heap.
They're very different things. But GPT four will tell you, well, the energy scale is very different and the timescales are very different. But the thing that's the same is that when the compost heap gets hotter, it generates heat faster. And when the atom bomb produces more neuron neutrons, it produces more neutrons faster. And so it gets the idea of a chain reaction.
And I believe it's understood that both forms of chain reaction, it's using that understanding to compress all that information into its weights. And if it's doing that, then it's going to be doing that for hundreds of things where we haven't seen the analogies yet, but it has. And that's where you get creativity from, from seeing these analogies between apparently very different things.
And so I think GPT four is going to end up when it gets bigger. Being very creative. I think this idea that it's just regurgitating what it's learned, just pastiche together, text is learned already. That's completely wrong. It's going to be even more creative than people.
I think you'd argue that it won't just repeat the human knowledge we've developed so far, but could also progress beyond that. I think that's something we haven't quite seen yet. We've started seeing some examples of it, but to a to a large extent, we're sort of still at the current level of of science. What do you think will enable it to go beyond that?
Well, we've seen that in more limited contexts, like if you take AlphaGo in that famous competition released, you know, there was move 37 where AlphaGo made a move that all the experts said must have been a mistake. But actually later they realized it was a brilliant move. so that was created within that limited domain. I think we'll see a lot more of that as these things get bigger.
The difference with, AlphaGo as well was that it was using reinforcement learning that that subsequently sort of enabled it to to go beyond the current state. So it started with imitation learning, watching how humans play the game, and then it would, through Self-play, develop way beyond that. Do you think that's the missing component of the current?
I don't know, I think that may well be a missing component. Yes, that the the self-play in alphas, in AlphaGo and AlphaZero, are a large part of what could make these creative moves, but I don't think it's entirely necessary. So there's a little experiment I did a long time ago where you you're training a neural net to recognize handwritten digits.
I love that example, the mNIST example. And you give it training data. Well, half the answers are wrong. and the question is, how well will it learn? And you make half the answers wrong once and keep them like that. So it can't average away the wrongness by just seeing the same example, but with the right answer. Sometimes the wrong, and sometimes when it sees that example.
How far for the examples? Or it's easy, for example, the answer is always wrong. And so the training data has 50% error. But if you train a backpropagation it gets down to 5% error or less. In other words, from badly labeled data, it can get much better results. It can see that the training data is wrong, and that's how smart students can be smarter than their advisor.
their advisor tells them all this stuff, and for half of what their advisor tells them, they think no rubbish and they listen to the other half and then they end up smarter than the advisor. So these big neural nets can actually do they can do much better than their training data. And most people don't realize that.
So how do you expect this model to add reasoning into them? So I mean, one approach is you had sort of the heuristics on on top of them, which a lot of the research is doing now where you have some notion of thought, you just feed back its reasoning, into itself. And another way would be in the model itself, as you scale it up, what's your intuition around that?
So my intuition is that as we scale up these models, they get better at reasoning. And if you ask how people work, roughly speaking, we have these intuitions and we can do reasoning and we use reasoning to correct our intuitions. Of course we use the intuitions during the reasoning to do the reasoning, but it's the conclusion of the reasoning conflicts with our intuitions.
We realized that intuitions need to be changed. That's much like in AlphaGo or AlphaZero, where you have an evaluation function, that just looks at the board and says, how good is that for me? But then you do the Monte Carlo rollout and now you get a more accurate idea, and you can revise your evaluation function so you can train it by getting it to agree with the results of reasoning.
And I think these large language models have to start doing that. They have to start training their raw intuitions about what should come next by doing reasoning and realizing that's not right. And so that way they can get more training data than just mimicking what people did. And that's exactly why AlphaGo could do this creative move. 37 it had much more training data because it was using reasoning to check out what the right next move should have been.
And what do you think about multimodality? So we spoke about this analogies. And often the analogies are way beyond what we could see. It's discovering analogies that are far beyond the humans. And maybe abstraction levels that we'll never be able to to, to understand. Now, when we introduce images to that and video and sound, how do you think that will cinch the models?
And, how do you think it extends the analogies that it will be able to make?
I think it'll change it a lot. I think it'll make it much better at understanding spatial things. For example, from language alone, it's quite hard to understand some spatial things, although remarkably, Gpt3 for can do that even before it was multimodal. but when you make it multimodal, if you have it both doing vision and reaching out and grabbing things, it'll understand object much better.
If it can pick them up and turn them over and so on. So although you can learn an awful lot from language is easier to learn if you multimodal and in fact you then need less language. And there's an awful lot of YouTube video for predicting the next frame, so. Or something like that. So I think these multimodal models are clearly going to take over.
you can get more data that way. They need less language. So the a point that you could learn a very good model from language alone, but it's much easier to learn it from a multimodal system.
And how do you think it will impact the models listening?
I think it'll make it much better at reasoning about space, for example, reasoning about what happens if you pick objects up. If you actually try picking objects up, you're going to get all sorts of training data that's going to help.
Do you think the human brain evolved to work well with with language, or do you think language evolved to work well with the human brain?
I think the question of whether language evolved to work with the brain or the brain evolved to work with language. I think that's a very good question. I think both happened. I used to think we would do a lot of cognition without needing language at all. now I've changed my mind a bit. So let me give you three different views of language.
and how it relates to cognition. There's the old fashioned symbolic view, which is cognition consists of having strings of symbols in some kind of cleaned up logical language where there's no, bigotry and applying rules of inference. And that's what cognition is. It's just these symbolic manipulations on things that are like strings of language symbols. so that's one extreme view.
An opposite extreme view is no, no. Once you get inside the head, it's all vectors. So symbols come in. You convert those symbols into big vectors, and all the stuff inside is done with being vectors. And then if you want to resample, you produce symbols again. So there was a point in machine translation about 2014 when people were using neural recurrent neural nets.
And words will keep coming in and then have a hidden state, and they keep accumulating information in this hidden state. So when they got to the end of a sentence that have a big hidden vector that captured the meaning of that sentence, it could then be used for producing the sentence in another language that was called a thought vector.
And that's the sort of second view of language. You convert the language into a big vector that's nothing like language. And that's what cognition is all about. But then there's a third view, which what I believe now, which is that you take these symbols and you convert the symbols into embeddings, and you use multiple layers of that. So you get these very rich embeddings, but the embeddings are still tied to the symbols in the sense that you've got a big vector for this symbol and a big vector for that symbol.
And these vectors interact to produce the vector for the symbol for the next word. And that's what understanding is. Understanding is knowing how to convert the symbols into these vectors, and knowing how the elements of the vector should interact to predict the vector for the next symbol. That's what understanding is, both in these big language models and in our brains.
And that's an example which is sort of in between your staying with the symbols, but you're interpreting them as these big vectors. And that's where all the work is and all the knowledge is in what vectors you use and how the elements of those vectors interact. Not in symbolic rules. but it's not saying that you get away from the symbol altogether.
It's saying you turn the symbols into big vectors, but you stay with that surface structure of the symbols. And that's how these models are working. And that's and I seem to me a more plausible model of human thought to.
You were one of the first folks to get the idea of using cheap to use. And, I know Yannis and loves you, for that, back in 2009, you mentioned that you told Jensen that this could be a quite good idea. for for training. Training neural nets take us back to that early intuition of of using GPUs for for training neural nets.
So actually, I think in about 2006, I had a former graduate student called Rick Zaleski with a very good computer vision guy, and I talked him at a meeting. He said, you know, you ought to think about using graphics processing cards because they're very good at matrix multiplies. And what you're doing is basically all matrix multiplies. So I thought about that for a bit.
And then we learned about these Tesla systems that had four GPUs in. And initially we just got gaming GPUs and discovered they made things go 30 times faster. And then we bought one of these Tesla systems with four GPUs. And we did speech on that. And it worked very well. And then in 2009, I gave a talk at NIPS and I told a thousand machine learning researchers, you should all go and buy Nvidia GPUs.
They're the future. You need them for doing machine learning. And I actually, then sent mail to Nvidia saying, I told a thousand machine learning researchers to buy your books, could you give me a free one? And they said, no, actually, they didn't say no. They just didn't reply. but when I told Jensen this story later on, he gave me a free one.
That's, that's very, very good. I think what's interesting is, as well is sort of how GPUs has evolved alongside, the so where do you think we should go, go next and in the, in the compute.
So my last couple of years at Google, I was thinking about ways of trying to make analog computation. So instead of using like a megawatt, we could use like 30W like brain. And we could run these big language models in analog hardware. And I never made it work. And but I started a really appreciating digital computation. So if you're going to use that low power analog computation, every piece of hardware is going to be a bit different.
And the idea is the learning is going to make use of the specific properties of that hardware. And that's what happens with people. All our brains are different. so we can't then take the weights in your brain and put them in my brain. The hard way is different. The precise properties of the individual neurons are different. The learning you may has learned to make use of all that.
And so we're mortal in the sense that the weights in my brain are no good for any other brain. When I die, those weights are useless. we can get information from one to another rather inefficiently by I produce sentences and you figure out how to change your weights. So you would have said the same thing. That's called distillation.
But that's a very inefficient way of communicating knowledge. And with digital systems, they're immortal, because once you've got some weights, you can throw away the computer, just store the weights on a tape somewhere, and I build another computer, put those same weights in, and if it's digital, it can compute exactly the same thing as the other system did.
So digital systems can share weights, and that's incredibly much more efficient if you've got a whole bunch of digital systems and they each go into a tiny bit of learning and they start with the same way, they do a tiny bit of learning, and then they share their weights again. they all know what all the others learned.
We can't do that. And so they're far superior to us in being able to share knowledge.
A lot of the ideas that have been deployed in the field are very old school ideas. it's the ideas that have been around the neuroscience for forever. What do you think you sort of left to, to, to apply to the systems that we develop.
So one big thing that we still have to catch up with neuroscience on is the timescales for changes. So in many of the neural nets there's a fast timescale for changing activities. So input comes in the activities, the embedding vectors all change. And then there's a slow timescale which is changing the weights. And that's long term learning. And you just have those two timescales in the brain.
There's many timescales in which weights change. So for example, if I say an unexpected word like cucumber and I five minutes later you put headphones on, there's a lot of noise and there's very faint words. You'll be much better at recognizing the word cucumber because I said it five minutes ago. So where is that knowledge in the brain?
And that knowledge is obviously in temporary changes to synapses. It's not neurons of going cucumber, cucumber, cucumber. You don't have enough neurons for that. It's in temporary changes to the weights. And you can do a lot of things with temporary weight changes fast, what I call fast weights. We don't do that in these neural models. And the reason we don't do it is because if you have temporary changes to the weights that depend on the input data, then you can't process a whole bunch of different cases at the same time.
At present, we take a whole bunch of different strings, we start and stick them together, and we process them all in parallel, because then we can do matrix. Matrix multiplies, which is much more efficient. And just that efficiency is stopping us using fast weights. But the brain clearly uses fast weights for temporary memory. And there's all sorts of things you can do that way that we don't do at present.
I think that's one of the biggest things we have done. I was very hopeful that things like Graphcore, if they went sequential and did just online learning, then they could use fast weights. but that hasn't worked out yet. I think it'll work out eventually when people are using conductance for weights.
How has knowing how this models work and knowing how the brain works impacted the way you think?
I think there's been one big impact, which is at a fairly abstract level, which is that for many years, people were very scornful about the idea of having a big random neural net and just giving you a lot of training data, and it would learn to do complicated things if you talked to statisticians or linguists or most people in AI, they say that's just a pipedream.
There's no way you're going to learn to really complicated things without some kind of innate knowledge, without a lot of architectural restrictions. It turns out that's completely wrong. You can take a big random neural network, and you can learn a whole bunch of stuff just from data. so the idea of stochastic gradient descent to adjust the repeatedly adjusts the weights using a gradient that will learn things and will learn big, complicated things.
That's been validated by these big models. And that's a very important thing to know about the brain. It doesn't have to have all this innate structure. Now, obviously it's got a lot of innate structure, but it certainly doesn't need innate structure for things that are easily learned. And so the sort of idea coming from Chomsky that you want, you won't learn anything complicated like language unless it's all kind of wide in already and just matures.
That idea is now clearly nonsense.
I'm sure some kid would appreciate you calling his ideas nonsense.
Well, I think actually, I think a lot of Chomsky's political ideas are very sensible, and I'm always struck by how how come someone with such sensible ideas about the Middle East could be so wrong about linguistics?
What do you think would make these models simulate consciousness of of humans more effectively? But imagine you had the AI system that you've spoken to in your entire life, and instead of that being like shattered today, that sort of deletes the memory of the conversation and you start fresh all of the time. Okay? It had self reflection at some point you pass away and you tell that to to the assistant.
Do you think.
It's not me? Somebody else tells that thesis.
Yeah. You would, it would be difficult for you to tell that to this extent. do you think that assistant would would fail at that point?
Yes. I think they can have feelings too. So I think just as we have this inner theater model for perception, we have an inner theater model for feelings, the things that I can experience, but other people can't. I think that model is equally wrong. So I think so far as I say, I feel like punching Gary on the nose, which I often do.
Let's try an abstract thought away from the idea of an inner theater. What I'm really saying to you is, if it weren't for the inhibition coming from my frontal lobes, I perform an action. So when we talk about feelings, we're really talking about, actions we would perform if it weren't for, constraints and that. Really? That's really.
What feelings are there actions we would do if it weren't for constraints? so I think you can give the same kind of explanation for feelings, and there's no reason why decisions can't have feelings. In fact, in 1973, I saw a robot have an emotion. So in Edinburgh, they had a robot with two grippers like this that could assemble a toy car.
If you put the pieces separately on a piece of green belt. but if you put them in a pile, if vision wasn't good enough to figure out what was going on. So this group is getting like whack. And it knocked them. So they're scattered. And then you can put them together. If you saw that in a person, you say it was crossed with the situation because they didn't understand it, so it destroyed it.
That's profound. You, we we spoke previously described sort of humans and as analogy missions. What do you think has been the most powerful analogies that you've found throughout your life?
Oh, throughout my life. whew. I guess probably an a sort of weak analogy. This influenced me a lot is, the analogy between religious belief and between belief and symbol processing. So when I was very young, I was confronted, I came from an atheist family and went to school and was confronted with religious belief, and it just seemed nonsense to me.
It still seems nonsense to me. and when I saw symbol processing as an explanation why people worked, I thought it was just the same nonsense. I don't think it's quite so much nonsense now, because I think actually we do do simple processing. It's just we do it by giving these big embedding vectors to the symbols. But we are actually symbol processing, but not at all in the way people thought where you match symbols.
And the only thing a symbol has is it's identical to another symbol or it's not identical. That's the only property a symbol has. We don't do that at all. We use the context to give embedding vectors to symbols, and then use the interactions between the components of these embedding vectors to do thinking. But there's a very good researcher at Google called Fernando Pereira who said, yes, we do have symbolic reasoning and the only symbolic we have is natural language.
Natural language is a symbolic language, and we reason with it. And I believe that now.
You've done some of the most meaningful, research in the history of, of computer science. Can you walk us through, like how do you select the right problems to, to work on?
Well, first let me correct you. Me and my students have done a lot of the most meaningful things, and it's mainly been a very good collaboration with students and my ability to select very good students. And that came from the fact there were very few people doing neural nets in the 70s and 80s and 90s and 2000s. And so the few people doing neural nets got to pick the very best students.
So that was a piece of luck. But my way of selecting problems is basically say, well, you know, when scientists talk about how they work, they have theories about how they work, which probably doesn't have much to do with the truth. But my theory is that I look for something where everybody's agreed about something and it feels wrong just as a slight intuition, something wrong about it.
And then I work on that and see if I can elaborate why it is. I think it's wrong, and maybe I can make a little demo with a small computer program that shows that it doesn't work the way you might expect. So let me take one example. most people think that if you add noise to a neural net is going to work worse.
if, for example, each time you put a training example through you, make half of the neurons be silent, it'll work worse. Actually, we know it'll generalize better if you do that, and you can demonstrate that, in a simple example. That's what's nice about computer simulation. You can show, you know, this idea you had that adding noise is going to make it worse and sort of dropping it off the neurons will make it work worse, which you will in the short term.
But if you train it with like that, in the end, do the work better. You can demonstrate that with a small computer program, and then you can think hard about why that is and how it stops big, elaborate adaptations. but that I think that that's my method of working. Find something that sounds suspicious and work on it, and see if you can give a simple demonstration of why it's wrong.
What sounds suspicious to you now?
Well, that we don't use fast. Wait, sounds suspicious that we only have these two timescales. That's just wrong. That's not at all like the brain. And in the long run, I think we're going to have to have many more timescales. So that's an example that.
And if you had, if you had your group of students today and they came to you and they said the ham in question that we talked about previously, you know, what's the most important problem in your field? What would you suggest that the take on and work on on next we spoke about reasoning, time, skills. What would be the sort of the highest priority problem that that you'd give them?
For me, right now, it's the same question I've had for the last like 30 years or so, which is does the brain do backpropagation? I believe the brain is getting gradients. If you don't get gradients, you're learning is just much worse than if you do get gradients. But how is the brain getting gradients? And is it somehow implementing some approximate version of backpropagation?
Or is it something completely different technique? That's a big open question. And if I kept on doing research, that's what I would be doing research on.
And when you look back at your career now, you've been right about so many things. But what were you wrong about that you wish you sort of spent less time pursuing a certain direction?
Okay, this is two separate questions. One is what were you wrong about? And two, do you wish you'd less spent less time on it? I think I was wrong about both machines and I'm glad I spent a long time on it. There are much more beautiful theory of how you get gradients than backpropagation. Backpropagation is just ordinary and sensible and it's just the chain rule.
Both machines is clever and it's a very interesting way to get gradients, and I would love for that to be how the brain works, but I think it isn't.
Did you spend much time imagining what would happen? Post the systems developing is as well. Did you ever have an idea that, okay, if we could make the systems work really well, we could, you know, democratize education, we could make knowledge way more accessible. we could solve some tough problems in medicine. Or was it more to you about understanding the Britain?
Yes. I sort of feel scientists ought to be doing things that are going to help society. But actually, that's not how you do your best research. You do your best research when it's driven by curiosity. You just have to understand something. much more recently, I've realized these things could do a lot of harm, as well as a lot of good, and I've become much more concerned about the effects they're going to have on society.
But that's not what was motivating. I just wanted to understand how on earth can the brain learn to do things? That's what I want to know. And I sort of failed as a side effect of that failure. We got some nice engineering, but.
Yeah, it was a good, good, good value for the world. If you take the lands of the things that could go really right. What what do you think are the most promising applications?
I think health care is clearly, a big one. with health care, there's almost no end to how much health care society can absorb. If you take someone old, they could use five doctors full time. so when I gets better than people are doing things, you'd like you to get better in areas where you could do with a lot more of that stuff.
And we could do with a lot more doctors. If everybody had three doctors of their own, that would be great, and we're going to get to that point. so that's one reason why health care is good. There's also just a new engineering developing new materials, for example, for better solar panels or for superconductivity or for just understanding how the body works.
there's going to be huge impacts that those are all going to be good things. What I worry about is bad actors using them for bad things. We've facilitated people like Putin or Z or Trump using AI for killer robots, or for manipulating public opinion or for mass surveillance. And those are all very worrying.
Things are you ever concerned that slowing down the field could also slow down the positives?
Oh, absolutely. And I think there's not much chance that the field will slow down, partly because it's international. And if one country slows down, the other countries are going to slow down. So there's a race clearly between China and the US and neither is going to slow down. So yeah, I did I mean there was this petition saying we should slow down for six months.
I didn't sign it just because I thought it was never going to happen. I maybe should have signed it because even though it was never going to happen, it made a political point. It's often good to ask for things you know you can't get just to make a point. but I didn't think we're going to slow down.
And how do you think that it will impact the AI research process, having, this assistance. So I think.
It'll make it a lot more efficient. AI research will get a lot more efficient when you've got these assistance to help you program, but also help you think through things and probably help you a lot with equations, too.
Have you reflected much on the process of selecting talent? Has that been mostly intuitive to you? Like when Ilya shows up at the door, you feel, this is a smart guy, let's work together.
So for selecting talent, sometimes you just know. So after talking to you for not very long, he seemed very smart and then talking was a bit more clear. He was very smart and had very good intuitions as well as being good at math. So that was a no brainer. There's another case where I was at a NIPS conference.
we had a poster and I, someone came up and he started asking questions about the poster, and every question he asked was a sort of deep insight into what we'd done wrong. and after five minutes, I offered him a postdoc position that guy was David McKee, who was just brilliant. And it's very sad he died, but he was.
It was very obvious you'd want him. other times it's not so obvious. one thing I did learn was that people are different. There's not just one type of good student. so there's some students who aren't that creative but are technically extremely strong and will make anything work. There's other students who aren't technically strong but are very creative.
Of course you want the ones who are both, but you don't always get that. But I think actually in the lab you need a variety of different kinds of graduate student. But I still go with my gut intuition that sometimes you talk to somebody and they're just very, very they just get it. And those are the ones you want.
What do you think is the reason for some folks having better intuition that they just have better training data than, than others? Or how can you develop your intuition?
I think it's partly they don't stand for nonsense. So here's a way to get bad intuitions. Believe everything you're told that's fatal. You have to be able to. I think here's what some people do. They have a whole framework for understanding reality. And when someone tells them something, they try and sort of figure out how that fits into their framework.
And if it doesn't, they just reject it. And that's a very good strategy. people are trying to incorporate whatever they're told, end up with a framework that's sort of very fuzzy and sort of can believe everything, and that's useless. So I think actually having a strong view of the world and trying to manipulate incoming facts to fit in with your view, obviously can lead you into deep religious belief and fatal flaws and so on, like my belief in Boltzmann machines.
but I think that's the way to go. If you've got good intuitions you can trust, you should trust them. If you've got bad intuitions, it doesn't matter what you do, so you might as well transform.
The very, very good, very good point. When when you look at, the types of research that's, that's, that's being done to do you think we're putting all of our eggs in one basket and we should diversify our ideas a bit more in the field? Or do you think this is the most promising direction? So let's go all in on it.
I think having big models and training them on multimodal data, even if it's only to predict the next word, is such a promising approach that we should go pretty much all in on it. Obviously there's lots and lots of people doing it, and there's lots of people doing apparently crazy things, and that's good. but I think it's fine for like most of the people to be following this path because it's working very well.
Do you think that the learning algorithms matter that much, or is it just a skill? Are there basically millions of ways that we could we could get to human level intelligence, or are there sort of a select few that we need to discover?
Yeah. So this issue of whether particular learning algorithms are very important or whether this a great variety of learning algorithms that will do the job, I don't know the answer. It seems to me, though, that back propagation, there's a sense in which is the correct thing to do, getting the gradient so that you change your parameter to make it work better.
That seems like the right thing to do, and it's been amazingly successful. There may well be other learning algorithms. There are alternative ways of getting that same gradient. All that are getting the gradient or something else and that also work. I think that's all open and very interesting issue now about whether there's other things you can try and maximize that will give you good systems.
And maybe the is doing that because it's easier. But backprop is in a sense the right thing to do. And we know that doing it works really well.
one last question. When when you look back at your sort of decades of research, what are you what are you most proud of? Is it the students? Is it the research? What what makes you most proud of? When you look back at at your life's work.
The learning algorithms are Boltzmann machines. So the learning algorithm in both machines is beautifully elegant. It's maybe hopeless in practice. but it's the thing I enjoyed most developing that rotary, and it's what I'm proudest of. even if it's wrong.
What questions do you spend most of your time thinking about now? Is it,
What should I watch on Netflix?