This transcript is generated with the help of AI and is lightly edited for clarity.
DEMIS HASSABIS:
AI is going to affect the whole world. It’s going to affect every industry. It’s going to affect every country. It’s going to be the most transformative technology ever, in my opinion. So if that’s true, and it’s going to be like electricity or fire, then I think it’s important that the whole world participates in its design. I think it’s important that it’s not just a hundred square miles of a patch of California. I do actually think it’s important that we get these other inputs, the broader inputs—not just geographically, but also different subjects: philosophy, social scientists, economists, not just the tech companies, not just the scientists, involved in deciding how this gets built and what it gets used for.
REID:
Hi, I’m Reid Hoffman.
ARIA:
And I’m Aria Finger.
REID:
We want to know how, together, we can use technology like AI to help us shape the best possible future.
ARIA:
With support from Stripe, we ask technologists, ambitious builders and deep thinkers to help us sketch out the brightest version of the future, and we learn what it’ll take to get there.
REID:
This is Possible.
REID:
In the 13th century, Sir Galahad embarked on a treacherous journey in pursuit of the elusive Holy Grail. The Grail, known in Christian lore as the cup Christ used in the Last Supper, had disappeared from King Arthur’s table. The Knights of the Round Table swore to find it. After many trials, Galahad’s pure heart allowed him the unique ability to look into the Grail and observe divine mysteries that could not be described by the human tongue.
ARIA:
In 2020, a team of researchers at DeepMind successfully created a model called AlphaFold that could predict how proteins will fold. This model helped answer one of the holy grail questions of biology: How does a long line of amino acids configure itself into a 3D structure that becomes the building block of life itself? In October 2024, three scientists involved with AlphaFold won a Nobel Prize for these efforts. This is just one of the striking achievements spearheaded by our guest today.
REID:
Demis Hassabis is a British artificial intelligence researcher, co-founder, and CEO of the AI company DeepMind. Under his leadership, DeepMind developed AlphaGo, the first AI to defeat a human world champion in Go, and later created AlphaFold, which solved the 50-year protein folding problem. He’s considered one of the most influential figures in AI.
ARIA:
Reid and I sat down for an interview with Demis, in which we talked about everything from game theory to medicine to multimodality and the nature of innovation and creativity.
REID:
Here’s our conversation with Demis Hassabis.
REID:
Demis, welcome to Possible. It was awesome dining with you at Queens. It was kind of a special moment in all kinds of ways. And you know, I think I’m going to start with a question that kind of came from your Babbage Theatre lecture, and also from the fireside chat that you did with Mohamed El-Erian. Which is: Share with us the moment that you went from thinking, ‘chess is the thing that I have spent my childhood doing’ to ‘what I want to do is start thinking about thinking. I want to accelerate the process of thinking and that computers are a way to do that.’ And how did you arrive at that? What age were you? What was that turn into metacognition?
DEMIS HASSABIS:
Well, first of all, thanks for having me on the podcast. Chess for me was, is, where it all started actually, and gaming. And I started playing chess when I was four, very seriously all through my childhood, playing for most of the England junior teams, captaining a lot of the teams. And for a long while, I was going to—my main aim was to become a professional chess player, a Grandmaster—maybe one day, possibly a world champion. And that was my whole childhood, really. Every spare moment, not at school, I was playing, going around the world, playing chess against adults in international tournaments. And then around 11 years old, I sort of had an epiphany, really, that although I love chess—and I still love chess today—is it really something that one should spend their entire life on?
DEMIS HASSABIS:
Is it the best use of my mind? So that was one thing that was troubling me a little bit. But then the other thing was, as we were going to training camps with the England chess team, we started to use early chess computers to try and improve your chess. And I remember thinking that—of course we were supposed to be focusing on improving the chess openings and chess theory and tactics—but actually I was more fascinated by the fact that someone had programmed this inanimate lump of plastic to play very good chess against me. And I was fascinated by how that was done. And I really wanted to understand that and then eventually try and make my own chess programs.
ARIA:
I mean, it’s so funny. I was saying to Reid before this, my 7-year-old’s school just won the New York State Chess Championship. So they have a long way to go before they get to you, but he takes it on faith, like, “Oh yeah, mom, I’m just going to go play ChessKid on the computer. I’ll go play against the computer a few games.” Which of course was sort of a revelation decades ago. And I remember when I was in middle school, it was obviously the Deep Blue versus Garry Kasparov, and this was a man versus machine moment. And one thing that you’ve gestured at about this moment is that it illustrated—in this case, based on Grandmaster data—it was brute force versus a self-learning system. Can you say more about that dichotomy?
DEMIS HASSABIS:
Yeah. Well, look, first of all, I mean, it’s great your son’s playing chess, and I think it’s fantastic. I’m a big advocate for teaching chess in schools as a part of the curriculum. I think it’s fantastic training for the mind, just like doing maths or programming would be. And it certainly affected the way I approach problems, and problem solve, and visualize solutions, and plan. It teaches you all these amazing meta skills—dealing with pressure. So you sort of learn all of that as a young kid, which is fantastic for anything else you’re going to do. And as far as Deep Blue goes, you’re right, most of these early chess programs—and then Deep Blue became the pinnacle of that—were these types of expert systems, which at the time were the favored way of approaching AI. Where actually it’s the programmers that solve the problem—in this case, playing chess.
DEMIS HASSABIS:
And then they encapsulate that solution in a set of heuristics and rules, which guides a kind of brute force search towards, in this case, making a good chess move. And I always had this, although I was fascinated by these AI chess programs—that they could do that—I was also slightly disappointed by them. And actually, by the time I got to Deep Blue—I was already studying at Cambridge in my undergrad—I was actually more impressed with Kasparov’s mind, because I’d already started studying neuroscience, than I was with the machine. Because here was this brute of a machine—all it can do is play chess. And then Kasparov can play chess at the same—roughly—the same level, but also could do all the other things, amazing things that humans can do. And so I thought, doesn’t that speak to the wonderfulness of the human mind?
DEMIS HASSABIS:
And it also, more importantly, means something was missing, very fundamental, from Deep Blue and these expert system approaches to AI. Right? Very clearly. Because Deep Blue did not seem—even though it was a pinnacle of AI at the time—it did not seem intelligent. And what was missing was its ability to learn, learn new things. So for example, it was crazy that Deep Blue could play chess to world champion level, but it couldn’t even play tic-tac-toe. Right? You’d have to reprogram, nothing in the system would allow it to play tic-tac-toe. So that’s odd, right? That’s very different to human Grandmasters, who can obviously play a simpler game trivially. And then also it was not general right? In the way that the human mind is. And I think those are the hallmarks. That’s what I took away from that match, is those are the hallmarks of intelligence, and they were needed if we wanted to crack AI.
REID:
And go a little bit into the deep learning, which obviously is part of the reason why DeepMind’s named the way it is. Because part of, I think, the what would seem to be completely contrarian hypothesis that you guys played out—with self-play and learning system—was that this learning approach was the right way to generate these significant systems. So say a little bit about having the hypothesis, what the trek through the desert looked like, and then what finding the Nile ended up with.
DEMIS HASSABIS:
Yes. Well, look, of course we started DeepMind in 2010 before anyone was working on this in industry. And there was barely any work on it in academia. And we partially named the company DeepMind, the deep part because of deep learning. It was also a nod to Deep Thought in a Hitchhiker’s Guide to the Galaxy, and Deep Blue, and other AI things. But it was mostly around the idea—we would bet on these learning techniques. Deep learning and hierarchical neural networks, they’d just sort of been invented, right? In seminal work by Jeff Hinton and colleagues in 2006. So it’s very, very new. And reinforcement learning, which has always been a speciality of DeepMind. And the idea of learning from trial and error, learning from your experience, right? And then making plans and acting in the world.
DEMIS HASSABIS:
And we combine those two things, really, we sort of pioneered doing that, and we called it deep reinforcement learning—these two approaches. And deep learning to build a model of the environment or what you were doing—in this case a game. And then the reinforcement learning to do the planning, and the acting, and actually accomplish and be able to build agent systems that could accomplish goals—in the case of games is maximizing the score, winning the game. And we felt that that was actually the entirety of what’s needed for intelligence. And the reason that we were pretty confident about that is actually from using the brain as an example, right? Basically those are the two major components of how the brain works. You know, the brain is a neural network. It’s a pattern matching and structure finding system.
DEMIS HASSABIS:
But then it also has reinforcement learning and this idea of planning and learning from trial and error and trying to maximize reward, which is actually in the human brain and the animal brain—a mammal brain—the dopamine system implements that. A form of reinforcement learning called TD [temporal difference] learning. So that gave us confidence that if we pushed hard enough in this direction, even though no one was really doing that, that eventually this should work, right? Because we have the existence proof of the human mind. And of course, that’s why I also studied neuroscience. Because when you’re in the desert—like you say—you need any source of water or any evidence that you might get out of the desert. You know, even a mirage in the distance is a useful thing to understand in terms of giving you some direction when you’re in the midst of that desert. And of course, AI was itself in the midst of that because several times this had failed. The expert system approach basically had reached a ceiling.
REID:
I could easily hog the entire interview, so I’m trying not to [laugh]. So one of the things that the learning system obviously ended up creating was solving what was previously considered an insoluble problem. There were even people who thought that computers couldn’t—like classical computational techniques—couldn’t solve Go, and it did. But not only did it solve Go, but in the classic “move 37”, it demonstrated originality, creativity, that was beyond the thousands of years of Go play, and books, and the hundreds of years of very serious play. What was that moment of “move 37” like for understanding where AI is, and what do you think the next “move 37” is?
DEMIS HASSABIS:
Well, look the reason Go was considered to be, and ended up being, so much harder than chess—so it took another 20 years, us with AlphaGo. And all the approaches that have been taken with chess, these expert systems approaches had failed with Go, right? Basically couldn’t even beat a professional, let alone a world champion. And the reason was two main reasons. One is the complexity of Go is so enormous. One way to measure that is there are 10 to the power 170 possible positions, right? Far more than atoms in the universe. There’s no way you can brute force a solution to Go, right? It’s impossible. But even harder than that, is that it’s such a beautiful, esoteric, elegant game. You know, it’s sort of considered art—an art form in Asia, really. Right?
DEMIS HASSABIS:
And it’s because it’s both beautiful aesthetically, but also it’s all about patterns rather than sort of brute calculation, which chess is more about. And so, even the best players in the world can’t really describe to you very clearly what are the heuristics they’re using. They just kind of intuitively feel the right moves, right? They’ll, they’ll sometimes just say that. “This move, why did you play this move?” “Well, it felt right.” Right? And then it turns out their intuition—if they’re a brilliant player—their intuition is brilliant, fantastic. And it’s an amazingly beautiful and effective move. But that’s very difficult then to encapsulate in a set of heuristics and rules to direct how a machine should play Go. And so that’s why all of these Deep Blue methods didn’t work.
DEMIS HASSABIS:
Now we got around that by having the system learn for itself—what are good patterns, what are good moves, what are good motifs and approaches and what our valuable and high probability of winning positions are. So it learned that for itself through experience. Through seeing millions of games and playing millions of games against itself. So that’s how we got AlphaGo to be better than world champion level. But the additional exciting thing about that is that it means those kinds of systems can actually go beyond what we as the programmers or the system designers know how to do, right? No expert system can do that, because of course, it’s strictly limited by what we already know and can describe to the machine. But these systems can learn for themselves.
DEMIS HASSABIS:
And that’s what we’ve resulted, in “move 37” in game two of the famous world championship match, the challenge match we had against Lee Sedol in Seoul in 2016. And that was a truly creative move. You know, Go has been played for thousands of years. It’s the oldest game humans have invented. And it’s the most complex game. And it’s been played professionally for hundreds of years in places like Japan. And even still, even despite all of that exploration by brilliant human players, this “move 37” was something never seen before. And actually worse than that, it was thought to be a terrible strategy. In fact, if you go and watch the documentary—which I recommend on YouTube, it’s on YouTube now—of AlphaGo, you’ll see the professional commentators nearly fell off their chairs when they saw “move 37” because they thought it was a mistake.
DEMIS HASSABIS:
They thought that the computer operator, Aja, had misclicked on the computer because it was so unthinkable that someone would play that. And then of course, in the end, it turned out a hundred moves later that “move 37″—the stone, the piece that was put down on the board—was in exactly the right place to be decisive for the whole game. So in turn now it’s studied as a great classic of Go, history of Go—that game and that move. And then even more exciting for that is that’s exactly what we hoped these systems would do. Because the whole point of me and my whole motivation, my whole life of working on AI, was to use AI to accelerate scientific discovery. And it’s those kinds of new innovations, albeit in a game, is what we were looking for from our systems.
REID:
And that I think is an awesome rendition of why it is these learning systems are even now doing original discovery. What do you think the next “move 37” might be for opening our minds to what is the way that AI can add a whole lot to the quality of human thought, human existence, human science?
DEMIS HASSABIS:
Well, look, I think there’ll be a lot of “move 37’s” in almost every area of human endeavor. Of course, the thing I’ve been focusing on since then is mostly being how can we apply those types of AI techniques, those learning techniques, those general learning techniques, to science. Big areas of science, I call them root node problems. So problems where if you think of the tree of all knowledge that’s out there in the universe—can you unlock some root nodes that unlock entire branches or new avenues of discovery that people can build on afterwards, right? And for us, protein folding and AlphaFold was one of those, is always top of my list. I have a mental list of all these types of problems that I’ve come across throughout my life and just being generally interested in all areas of science and sort of thinking through which ones would be suitable—would both be hugely impactful—but also suitable for these types of techniques.
DEMIS HASSABIS:
And I think we’re going to see a new golden era of these types of new strategies, new ideas, in very important areas of human endeavor. Now, I would say, one thing to say though, is that we haven’t fully cracked creativity yet, right? So I don’t want to claim that. I think that there are, I often describe as, three levels of creativity, and I think AI is capable of the first two. So the first one would be interpolation. So you give it a million pictures of cats—an AI system, a million pictures of cats—and you say, “Create me a prototypical cat.” And it will just average all the million cat pictures that it’s seen. And that prototypical one won’t be in the training set. So it will be a unique cat, but that’s not very interesting from a creative point of view, right?
DEMIS HASSABIS:
It’s just an averaging. But the second thing would be what I call extrapolation. So that’s more like AlphaGo where you’ve played 10 million games of Go, you’ve looked at a few million human games of Go, but then you come up with—you extrapolate from what’s known—to a new strategy, never seen before, like “move 37”. Okay? So that’s very valuable already. I think that is true creativity. But then there’s a third level, which I call it invention or out-of-the-box thinking, which is, not only can you come up with a “move 37”, but could you have invented Go, right? Or another measure I like to use is if we went back to the time of Einstein in 1900, early 1900s, could an AI system actually come up with general relativity with the same information that Einstein had at the time? And clearly today, the answer is no to those things, right? It can’t invent something, a game as great as Go, and it wouldn’t be able to invent general relativity just from the information that Einstein had at the time. And so there’s still something missing from our systems to get true out of the box thinking. But I think it’ll come, but we just don’t have it yet.
ARIA:
I think so many people outside of sort of the AI realm would sort of be surprised that it all starts with gaming, but that’s sort of gospel for what we’re doing. That’s how we created these systems. And so, switching gears from board games to video games. Can you give us the elevator pitch explanation for what exactly makes an AI that can play StarCraft II—like AlphaStar—so much more advanced and fascinating than the one that can play chess or Go.
DEMIS HASSABIS:
Yeah, with AlphaGo, we sort of cracked the pinnacle of board games, right? So Go is always considered the Mount Everest, if you like, of games—AI for board games. But there are even more complex games, by some measures, if you take on board the most complex strategy games that you can play on computers. And StarCraft II is acknowledged to be the classic of the genre of real-time strategy games. And it’s a very complex game. You’ve got to build up your base, and your units, and other things. So every game is different, right? And the board game’s very fluid, and you’ve got to move many units around in real time. And the way we cracked that was to add this additional level in of a league of agents competing against each other, all seeded with slightly different initial strategies.
DEMIS HASSABIS:
And then you kind of get a sort of survival of the fittest. You have a tournament between them all. There’s a kind of multi-agent set up now, and the strategies that win out in that tournament go to the next epoch, and then you generate some other new strategies around that. And you keep doing that for many generations. You’re kind of both having this idea of self-play that we had in AlphaGo, but you’re adding in this multi-agent competitive, almost evolutionary, dynamic in there. And then eventually you get an agent—or series or a set of agents—that are kind of the Nash distribution of agents that no other strategy dominates them, but they dominate the most number of other strategies. And then you have this kind of Nash equilibrium, and then you pick out the top agents from that. And that succeeded very well with this type of very open-ended kind of gameplay.
DEMIS HASSABIS:
So it’s quite different from what you get with chess or Go where the rules are very prescribed, and the pieces that you get are always the same, and it’s a very ordered game. Something like Starcraft’s much more chaotic. So it’s interesting to have to deal with that. It has hidden information too. You can’t see the whole map at once. You have to explore it. So it’s not a perfect information game, which is another thing we wanted our systems to be able to cope with, is partial information situations, which is actually more like the real world, right? Very rarely in the real world do you actually have full information about everything. Usually you only have partial information and then you have to infer everything else in order to come up with the right strategies.
REID:
And part of the game side of this is—I presume you’ve heard that there’s this theory of Homo Ludens.
DEMIS HASSABIS:
Yes.
REID:
That we’re game players. Is that informing the thinking about how games is both strategic, but also framing for science acceleration. Framing for the serendipity of innovation. Is in addition to the fitness function, the evolution of self-play, the ability to play-scale-compute—are there other deeper elements to the game playing nature that allows this thinking of thinking?
DEMIS HASSABIS:
Well, look, I’m glad you brought up Homo Ludens, and it’s a wonderful book and it basically argues that games playing is actually a fundamental part of being human, right? In many ways, the act of play, what could be more human than that, right? And then of course it leads into creativity, fun. As you know, all of these things get built on top of that. And so I’ve always loved them as a way to practice and train your own mind in situations that you might only ever get a handful of times in real life, but they’re usually very critical. What company to start, what deal to make—things like that. So I think games is a way to practice those scenarios.
DEMIS HASSABIS:
And if you take games seriously, then you can actually simulate a lot of the pressures one would have in decision making situations. And going back to earlier, that’s why I think chess is such a great training ground for kids to learn, because it does teach them about all of these situations. And so of course it’s the same for AI systems too. There was the perfect proving ground for our early AI system ideas, partly because they were invented to be challenging and fun for humans to play. And of course there’s different levels of gameplay. So we could start with very simple games, like Atari games, and then go all the way up to the most complex computer games, like StarCraft, right? And continue to sort of challenge our system. So we were in the sweet spot of the s-curve. So it’s not too easy, it’s trivial, or too hard, you can’t even see if you’re making any progress.
DEMIS HASSABIS:
You want to be in that maximum part of the s-curve where you’re making almost exponential progress. And we could keep picking harder and harder games as our systems got improved. And then the other nice feature about games is because they’re some kind of microcosm of the real world, they’ve usually been boiled down to very clear objective functions, right? So winning the game or maximizing the score is usually the objective of a game. And that’s very easy to specify to a reinforcement learning system or an agent based system. So it’s perfect for hill climbing against, right? And measuring Elo scores, ratings, and exactly where you are. And then finally, of course, you can calibrate yourself against the best human players. So you can calibrate what your agents are doing in their own tournaments.
DEMIS HASSABIS:
In the end, even with the StarCraft agent, we had to eventually challenge a professional Grandmaster at StarCraft to make sure that our systems hadn’t overfitted somehow to their own tournament strategies, right? It actually needed to be—we grounded it with—oh, it can actually be a genuine human Grandmaster StarCraft player. The final thing is, of course, you can generate as much synthetic data as you want with games too. Which is coming into vogue right now again about data limitations, and with large language models, and how many tokens left in the world, and has it read everything in the world. Obviously for things like games you can actually just play the system against itself and generate lots more data from the right distribution.
ARIA:
Can you double click on that for a moment? Like you said, it is in vogue to talk about—are we running out of data, do we need synthetic data? Where do you stand on that issue?
DEMIS HASSABIS:
Well I’ve always been a huge proponent of simulations, and simulations and AI. And it’s also interesting to think about what the real world is, right? In terms of a computational system. And so I’ve always been involved with trying to build very realistic simulations of things. And now of course that interacts with AI because you can have an AI that learns a simulator of some real world system, just by observing that system or all the data from that system. So I think the current debate is to do with these large foundation models now pretty much use the whole internet, right? And so then once you’ve tried to learn from those what’s left, right? That’s all the language that’s out there. Of course, there’s other modalities like video and audio.
DEMIS HASSABIS:
I don’t think we’ve exhausted all of that kind of multimodal tokens, but even that will reach some limit. So then the question becomes can you generate synthetic data? And I think that’s why you’re seeing quite a lot of progress with maths and coding, because in those domains, it’s quite easy to generate synthetic data. Because the problem with synthetic data is are you creating data that is from the right distribution? The actual distribution, right? Does it mimic the real distribution? And also are you generating data that’s correct, right? And of course, for things like maths, for coding, and for play—things like gaming—you can actually test the final data and verify if it’s correct, before you feed it in as input into the training data for a new system. So it’s very amenable, certain areas. In fact, turns out the more abstract areas of human thinking that you can verify and prove that it’s correct. And so therefore that unlocks the ability to create a lot of synthetic data.
REID:
So one of the things that is also in addition to the frequent discussion around data—how do we get more—but one of the questions is, in order to do AI, is it important to actually have it embedded in the world?
DEMIS HASSABIS:
Yeah. Well, interestingly, if we talked about this five years ago, or certainly 10 years ago, I would’ve said that some real world experience, maybe through robotics—usually when we talk about embodied intelligence, we’re meaning robotics, but it could also be a very accurate simulator, right? Like some kind of ultra-realistic game environment—would be needed to fully understand the, say, the physics of the world around you, right? And the physical context around you. And there’s actually a whole branch of neuroscience that is predicated on this. It’s called action in perception. So this is the idea that one can’t actually fully perceive the world unless you can also act in it. And the kinds of arguments go how can you really understand the concept of, of the weight of something, for example, unless you can pick things up and compare them with each other, and then you get this sort of idea of weight. Can you really get that notion just by looking at things?
DEMIS HASSABIS:
It seems hard, right? Certainly for humans. I think you need to act in the world. So this is the idea that acting in the world is part of your learning. You are kind of like an active learner. And in fact, reinforcement learning is like that because the decisions you make give you new experiences, but those experiences depend on the actions you took, but also those are the experiences that you’ll then subsequently learn from. So in a sense, reinforcement learning systems are involved in their own learning process, right? Because they’re active learners. And, and I think you can make a good argument that that’s also required in the physical world. Now, it turns out, I’m not sure I believe that anymore. Because now with our systems, especially our video models—if you’ve seen Veo 2, our latest video models completely state-of-the-art, which we released late last year—and it shocked even me, that even though we are building this thing, that basically by watching a lot of YouTube videos, it can figure out the physics of the world.
DEMIS HASSABIS:
There’s a sort of funny Turing test—in some sense “Turing test”—of video models. Which is, can you chop a tomato? Can you show a video of a knife chopping a tomato with the fingers and everything in the right place, and the tomato doesn’t magically spring back together, or the knife goes through the tomato without cutting it, et cetera. And Veo can do it. And if you think through the complexity of the physics to understand what you’ve got to keep consistent and so on. It’s pretty amazing. It’s hard to argue that it doesn’t understand something about physics and the physics of the world. And it’s done it without acting in the world. And certainly not acting as a robot in the world.
DEMIS HASSABIS:
So, it’s not clear to me there is a limit now with just sort of passive perception. Now, the interesting thing is that I think this has huge consequences for robots as an embodied intelligence as an application. Because the types of models we’ve built—Gemini, and also now Veo, and we’ll be combining those things together at some point in the future—we’ve always built Gemini, our foundation model, to be multimodal from the beginning. And the reason we did that—and we still lead on all the multimodal benchmarks—is because for twofold. One is we have a vision for this idea of a universal digital assistant. An assistant that goes around with you, it’s on the digital devices, but also in the real world, maybe on your phone or a glasses device. And actually helps you in the real world—recommend things to you, help you navigate around, help with physical things in the world like cooking. Stuff like that.
DEMIS HASSABIS:
And for that to work, you obviously need to understand the context that you’re in. It’s not just the language I’m typing into a chat bot. You actually have to understand the 3D world I’m living in, right? I think to be a really good assistant, you need to do that. But the second thing is, of course, it is exactly what you need for robotics as well. And we released our first big Gemini robotics work—which has caused a bit of a stir—and that’s the beginning of showcasing what we can do with these multimodal models that do understand physics of the world with a little bit of robotics fine tuning on top to do with the actions, and the motor actions, and the planning, a robot needs to do. And looks like it’s going to work. So actually now, I think these general models are actually going to transfer to the embodied robotics setting without too much extra special casing, or extra data, or extra effort. Which is probably not what most people, even the top roboticists, would’ve predicted five years ago.
ARIA:
I mean, that’s wild [laugh]. And thinking about benchmarks and what we’re going to need these digital assistants to do. When we look under the hood of these big AI models, there’s—well, some people would say it’s a tension. So the trade offs is thinking time versus output quality. We need them to be fast, but of course we need them to be accurate. And so talk about what is that trade off and how is that going in the world right now?
DEMIS HASSABIS:
Well, look, we sort of pioneered all that area of thinking systems because that’s what our original gaming systems all did, right? Go, AlphaGo. But actually most famously AlphaZero, which was our follow-up system that could play any two-player game. And there you always have to think about your time budget, your compute budget, you’ve got to actually do the planning part, right? So the model you can pre-train, just like we do with our foundation models today. So you can play millions of games offline, and then you have your model of chess, or your model of Go, whatever it is. But at test time, at runtime, you’ve only got one minute to think about your move, right? One minute times however many computers you’ve got running. So that’s still a limited compute budget.
DEMIS HASSABIS:
So what’s very interesting today is there’s [a] trade-off between do you use a more expensive, larger base model, foundation model, right? So in our case we have different size names like Gemini Flash, or Pro, or even bigger, which is Ultra. But those models are more costly to run, so they take longer to run, but they’re more accurate and they’re more capable. So you can run a bigger model with a shorter number of planning steps, or you can run a very efficient, smaller model, that’s slightly less powerful, but you can run it for many more steps, right? And it’s actually, currently what we’re finding is, it’s sort of roughly about equal. But of course what we want to find is the Pareto frontier of that, right? Like actually the exact right trade off of the size of the model and the expense of running that model versus the amount of thinking time you want, and thinking steps, that you are able to do per unit of compute time. And I think that’s actually fairly cutting edge research right now that I think all the leading labs are probably experimenting on. And I think there’s not a clear answer to that yet.
REID:
You know, all the major labs—DeepMind, others—are all working intensely on coding assistants. And there’s a number of reasons. Everything from A: it’s one of the things that accelerates productivity across the whole front. It has a good fitness function. It’s also, of course, one of the ways that everyone is going to be enhanced in productivity is having a software copilot agent for helping. There’s just a ton of reasons. Now, one of the things that gets interesting here is as you’re building these obviously there’s a tendency to start with these computer languages that have been designed for humans. What would be computer languages that would be designed for AIs, or an agentic world? Or designed for this hybrid process of a human plus an AI? Is that a good world to start looking at those computer languages? How would it change our theory of computation, linguistics, et cetera?
DEMIS HASSABIS:
I think we are entering a new era in coding, which is going to be very interesting. And as you say, all the leading labs are pushing on this frontier for many reasons. It’s easy to create synthetic data, so that’s another reason that everyone’s pushing on this vector. And I think we’re going to move into a world where—sometimes it’s called vibe coding—where you’re basically coding with natural language, really. Right? And we’ve seen this before with computers, right? I remember when I first started programming in the eighties we were doing assembler. And then of course, that seems crazy now, like why would you do machine code? You start with C, and then you get Python, and so on.
DEMIS HASSABIS:
And really, one could see is the natural evolution of going higher and higher up the abstraction stack of programming languages, and leaving more and more of the lower level implementation details to the compiler in a sense. And now one could just view this as the natural final step—is, “Well, we just use natural language.” And then everything is high level programming—super high level programming language. And I think eventually that’s maybe what we’ll get to. And the exciting thing there is that, of course, it will make accessible coding to a whole new range of people, creatives, right? Designers, game designers, app writers, that normally would not have been able to implement their ideas without the help of teams of programmers.
DEMIS HASSABIS:
So that’s going to be pretty exciting, I think, from a creativity point of view. But it may also be very good, certainly in the next few years, for coders as well. Because I think—and I think this in general with these AI tools is—I think that the people that are going to get most benefit out of them initially will be the experts in that area who also know how to use these tools in precisely the right way. You know, whether that’s prompting or interfacing with your existing code base. There’s going to be this sort of interim period where I think the current experts who embrace these new tools—whether that’s filmmakers, game designers, or coders—are going to be like superhuman, in terms of what they’re able to do. And I see that with some film directors and film designer friends of mine, who are able to pitch decks, for example, for new film ideas in a day on their own. But it’s a very high quality pitch deck that they can pitch for a $10 million budget for.
DEMIS HASSABIS:
And normally they would’ve had to spend a few tens of thousands of dollars just to get to that pitch deck, which is a huge risk for them. I think there’s going to be a whole new incredible set of opportunities. And then there’s the question of if you think about creative, the creative arts, whether there’ll be new ways of working. Much more fluid. Instead of doing Adobe Photoshop or something, you’re actually co-creating this thing with this fluid responsive tool. And that could kind of feel more like Minority Report or something I imagine. With the interface and there’s this thing swirling around, but it will require people to get used to a very new workflow, to take maximum advantage of that. But I think when they do it’ll be probably incredible for those people. They’ll be like 10 X more productive.
ARIA:
So I want to go back to the world of multimodal that we were talking about before, with robots in the real world. And so right now, most AI doesn’t need to be multimodal in real time because the internet is not multimodal. And for our listeners, that means absorbing many types of input—voice, text, vision—at once. And so can you go deeper in what you think the benefits of truly real time multimodal AI will be? And what are the challenges to get to that point?
DEMIS HASSABIS:
I think first of all, we live in a multimodal world, right? And we have our five sensors and that’s what makes us human. So if we want our systems to be brilliant tools or fantastic assistance, I think in the end, they’re going to have to understand the world—the spatial, temporal world—that we live in. Not just our linguistic, maths world, right? Abstract thinking world. I think that they’ll need to be able to act in, and plan in, and process things, in the real world and understand the real world. I think the potential for robotics is huge. I don’t think it’s had its ChatGPT or its AlphaFold moment yet, say, in science and language, right? Or AlphaGo moment. I think that’s to come. But I think we’re close and as we talked about before, I think in order for that to happen—I think the shortest path I see that happening on now—is these general multimodal models being eventually good enough, and maybe we’re not very far away from that, to sort of install on a robot. Perhaps a human on robot, with the cameras.
DEMIS HASSABIS:
Now there’s additional challenges of you’ve got to fit it locally, maybe on the local chips, to have the latency fast enough and so on. But as we all know, just wait a couple of years and those systems that are state-of-the-art today will fit on a little mobile chip tomorrow. So I think it’s very exciting, multimodal, from that point of view—robotics, assistance, and then finally, I think also for creativity, I think we are the first model in the world—Gemini 2.0, that you can try now in AI Studio—that allows native image generation. So not calling a separate program, a separate model—in our case Imagen 3, which you can try separately—but actually Gemini itself natively coming up in the chat flow of images.
DEMIS HASSABIS:
And I think people seem to be really enjoying using that. So you’re now talking to a multimodal chat bot, right? And so you can get it to express emotions in pictures. Or you can give it a picture and then tell it to modify it, and then continue to work on it with word descriptions. You know, “Can you remove that background? Can you do this?” So this goes back to the earlier thing we said about programming or any of these creative things in a new workflow. I think we’re just seeing the glimpse of that—if you try out this new Gemini 2 experimental model—of how that might look in image creation. And that’s just the beginning. Of course it’ll work with video, and coding, and all sorts of things.
REID:
So in the land of the real world, multimodal, one of the things that frequently people speculate is geolocation of AI work. And obviously in the U.S. we intensely track everything that’s happening on the West Coast. We also intensely track DeepMind. And then somewhat less, Mistral, and others. What’s some of the stuff that’s really key for the world to understand about what’s coming out of Europe? What’s the benefit of having there be multiple major centers of innovation and invention? Not just within the West Coast, but also obviously DeepMind in London, and Mistral and Paris, and others. And what are some of the things for people to pay attention to—why it’s important and what’s happening, especially within the U.K. and European AI ecosystem?
DEMIS HASSABIS:
We started DeepMind in London and are still headquartered here for several reasons. I mean, this is where I grew up. It’s what I know, it’s where I had all the contacts that I had. But the competitive reasons were that we felt that the talent in the U.K. and in Europe—the coming out of universities—was the equivalent of the top U.S. ones. You know, Cambridge—my alma mater—and Oxford, they’re up there with the MITs, and Harvards, and the Ivy Leagues, right? I think they’re always in the top ten there together on the university world tables. But if you—this is certainly true in 2010—say you had a PhD in physics out of Cambridge and you didn’t want to work in finance at a hedge fund in the city, but you wanted to stay in the U.K. and be intellectually challenged, they were not that many options for you, right?
DEMIS HASSABIS:
There are not that many deep tech startups. So we were the first, really, to prove that could be done. And actually we were a big draw for the whole of Europe. So we got the best people from the technical universities in Munich, and in Switzerland, and so on. And for a long while that was a huge competitive advantage. And also salaries were cheaper here than in the west coast. And you weren’t competing against the big incumbents, right? And also it was conducive. The other reason I chose to do that was I knew that AGI—which was our plan from the beginning, solve intelligence and then use it to solve everything else, that was where we articulated our mission statement, and I still like that framing of it—it was a 20 year mission.
DEMIS HASSABIS:
And if you are on a 20 year mission—and we are now 15 years in, and I think we’re sort of on track unbelievably, right? Which is strange for any 20 year mission. But if you don’t want to be too distracted on the way, in a deep technology, deep scientific mission. So one of the issues I find with Silicon Valley—lots of benefits, obviously, contacts, and support systems, and funding, and amazing things, and the amount of talent there, the density of talent—but it is quite distracting, I feel. Like everyone and their dog is trying to do a startup that they think’s going to change the world, but it’s just a photo app or something. And then the cafes are filled with this. Of course, it leads to some great things, but it’s also a lot of noise if one actually wants to commit to a long-term mission that you think is the most important thing ever.
DEMIS HASSABIS:
And you and your staff don’t want to be too distracted and like, “Oh, maybe I could make a hundred million though if I jumped and quickly did this gaming app or something!” Right? And, and I think that’s sort of the milieu that you are in, in the Valley, at least back then. Maybe this is less true now. There’s probably more mission focused, startups now. But I also wanted to prove it could be done elsewhere. And then the final reason I think it’s important is that AI is going to affect the whole world, right? It’s going to affect every industry. It’s going to affect every country. It’s going to be the most transformative technology ever, in my opinion. So if that’s true and it’s going to be like electricity or fire—more impactful than even the internet or mobile—then I think it’s important that the whole world participates in its design. And with the different value systems that we think are out there that are good philosophies, and from democratic values—Western Europe, U.S. I think it’s important that it’s not just a hundred square miles of a patch of California, right?
DEMIS HASSABIS:
I do actually think it’s important that we get these other inputs, the broad inputs, not just geographically, but also—and I know you agree with this Reid—different subjects. Philosophy, social scientists, economists, academia, civil society. Not just the tech companies, not just the scientists, involved in deciding how this gets built and what it gets used for. And I’ve always felt that very strongly from the beginning. And I think having some European involvement and some U.K. involvement at the top table of the innovation is a good thing.
ARIA:
So, Demis, one of the areas of AI that when anyone asks me, “Hey Aria, I know you’re interested in AI, but like it can write my emails? Like, why is it so special?” I just say, “No, think about what it can do in medicine.” I always talk about AlphaFold. I tell them about what Reid is doing. I’m just so excited for those breakthroughs. Can you give us just a little bit—you had the seminal breakthrough in AlphaFold, and what is it going to do for the future of medicine?
DEMIS HASSABIS:
I’ve always felt that—what are the most important things AI can be used for? Right? And I think there are two. One is human health—that’s number one—trying to solve and cure terrible diseases. And then number two is to help with energy sustainability, and climate—the planet. The planet’s health, let’s call it. Right? So there’s humans’ health and then there’s the planet’s health. And those are the two areas that we have focused on in our science group, which I think is fairly unique amongst the AI labs actually, in terms of how much we pushed that from the beginning. And protein folding specifically was this canonical for me. I came across it when I was an undergrad in Cambridge 30 years ago. And it’s always stuck with me as this fantastic puzzle that would unlock so many possibilities.
DEMIS HASSABIS:
You know, the structure of proteins. Everything in life depends on proteins and we need to understand the structure so we know their function. And if we know the function, then we can understand what goes wrong in disease, and we can design drugs and molecules that will bind to the right part of the surface of the protein, if you know the 3D structure. So a fascinating problem. It goes to all of the computational things we were discussing earlier as well. Can you enumerate? Can you see this through this forest of possibilities? You know, all these different ways a protein can fold. Some people estimate—Levinthal very famously in the 1960s estimated—an average protein can fold in 10 to 300 possible ways, right? So how do you enumerate those astronomical possibilities? And yet it is possible with these learning systems.
DEMIS HASSABIS:
And that’s what we did with AlphaFold. And then we spun out a company Isomorphic—and I know Reid’s very interested in this area too, with his new company. If we can reduce the time it takes to discover a protein structure from. It used to take a PhD student, their entire PhD—as a rule of thumb—to discover one protein structure. So four or five years. And there’s 200 million proteins known to science. And we folded them all in one year. So we did a billion years of PhD time in one year, is another way you can think of it. And then gave it to the world freely to use. And two million researchers around the world have used it. And we spun out a new company, Isomorphic, to try and go further downstream now and develop the drugs needed, and try and reduce that time.
ARIA:
I mean, it’s just amazing. I mean Demis, there’s a reason they give you the Nobel Prize. Thank you so much for all of your work in this area. It’s truly amazing.
REID:
And now to rapid fire, is there a movie, song, or book that fills you with optimism for the future?
DEMIS HASSABIS:
There’s lots of movies that I’ve watched that have been super inspiring for me. Things like Blade Runner. It’s probably my favorite sci-fi movie. But maybe it’s not that optimistic. So if you want an optimistic thing, I would say the Culture series by Iain Banks. I think that’s the best depiction of a post-AGI universe where AIs—and you’ve basically got societies of AIs, and humans, and alien species actually, and maximum human flourishing across the galaxy. That’s an amazing, compelling future that I would hope for humanity.
ARIA:
What is a question that you wish people asked you more often?
DEMIS HASSABIS:
The questions I sort of often wonder why people don’t discuss a lot more, including with me, are some of the really fundamental properties of reality, that actually drove me in the beginning when I was a kid, to think about building AI to help us sort of this ultimate tool for science. So for example, I don’t understand why people don’t worry more about what is time? What is gravity? Basically the fundamental fabric of reality which is sort of staring us in the face all the time. All these very obvious things that impact us all the time. And we don’t really have any idea how it works. And I don’t know why that doesn’t trouble people more [laugh]. It troubles me! And I’d love to have more debates with people about those things. But, actually most people, they seem to sort of shy away from those topics.
REID:
Where do you see progress or momentum, outside of your industry, that inspires you?
DEMIS HASSABIS:
That’s a tough one because AI is so general, it’s almost touching—you know, what industry is outside of the AI industry? I’m not sure there’s many. Maybe the progress going on in quantum is interesting. I still believe AI is going to get built first and then will maybe help us perfect our quantum systems. But I have ongoing bets with some of my quantum friends, like Hartmut Neven, on they’re going to build quantum systems first, and then that will help us accelerate AI. So I always keep a close eye on the advances going on with quantum computing systems.
ARIA:
Final question. Can you leave us with a final thought on what is possible over the next 15 years if everything breaks humanity’s way, and what’s the first step to get there?
DEMIS HASSABIS:
Well, what I hope for the next ten, 15, years is what we’re doing in medicine to really have new breakthroughs. And I think maybe in the next ten, 15, years, we can actually have a real crack at solving all disease, right? That’s the mission of Isomorphic. And I think with AlphaFold, we showed what the potential was to do what I like to call science at digital speed. And why couldn’t that also be applied to finding medicines? And so my hope is—ten, 15, years time—we’ll look back on the medicine we have today a bit like how we look back on medieval times and how we used to do medicine then. And that would be, I think, the most incredible benefit we could imagine from AI.
REID:
Possible is produced by Wonder Media Network. It’s hosted by Aria Finger and me, Reid Hoffman. Our showrunner is Shaun Young. Possible is produced by Katie Sanders, Edie Allard, Sara Schleede, Vanessa Handy, Alyia Yates, Paloma Moreno Jimenez, and Melia Agudelo. Jenny Kaplan is our executive producer and editor.
ARIA:
Special thanks to Surya Yalamanchili, Saida Sapieva, Thanasi Dilos, Ian Alas, Greg Beato, Parth Patil, and Ben Relles. And a big thanks to Leila Hajaj, Alice Talbert, and Denese Owusu-Afriyie.