Demis Hassabis: DeepMind - AI, Superintelligence & the Future of Humanity | Lex Fridman Podcast #299

1. 0:00 – 1:01

   Introduction

   1. LFLex Fridman0:00

      The following is a conversation with Demis Hassabis, CEO and co-founder of DeepMind, a company that has published and built some of the most incredible artificial intelligence systems in the history of computing, including AlphaZero that learned all by itself to play the game of Go better than any human in the world, and AlphaFold2 that solved protein folding, both tasks considered nearly impossible for a very long time. Demis is widely considered to be one of the most brilliant and impactful humans in the history of artificial intelligence and science and engineering in general. This was truly an honor and a pleasure for me to finally sit down with him for this conversation, and I'm sure we will talk many times again in the future. This is the Lex Fridman Podcast. To support it, please check out our sponsors in the description. And now, dear friends, here's Demis Hassabis.

2. 1:01 – 8:27

   Turing Test

   1. LFLex Fridman1:01

      Let's start with a bit of a personal question. Am I an AI program you wrote to interview people until I get good enough to interview you?

   2. DHDemis Hassabis1:11

      Well, I'll be impressed if, if you were. I'll be impressed with myself if you were. I don't think we're quite up to that yet, but, uh, maybe you're from the future, Lex.

   3. LFLex Fridman1:18

      If you did, would you tell me? Is that a- is that a good thing to tell a language model that's tasked with interviewing that it is in fact, um, AI?

   4. DHDemis Hassabis1:27

      Maybe we're in a kind of meta Turing test. Uh, probably, probably it would be a good idea not to tell you so it doesn't change your behavior, right?

   5. LFLex Fridman1:33

      This is a kind of-

   6. DHDemis Hassabis1:34

      Heisenberg uncertainty principle situation.

   7. LFLex Fridman1:37

      Yeah. (laughs)

   8. DHDemis Hassabis1:37

      If I told you, you'd behave differently.

   9. LFLex Fridman1:38

      Yeah.

   10. DHDemis Hassabis1:39

       Maybe that's what's happening with us, of course.

   11. LFLex Fridman1:41

       This is a benchmark from the future where they replay 2022 as a year before AIs were good enough yet, and now we want to see-

   12. DHDemis Hassabis1:51

       (laughs)

   13. LFLex Fridman1:51

       ... is it gonna pass?

   14. DHDemis Hassabis1:52

       Exactly. (laughs)

   15. LFLex Fridman1:54

       If I was such a program, would you be able to tell, do you think? So to the Turing test question, you've, you've talked about the benchmark for solving intelligence. What would be the impressive thing? You've talked about winning a Nobel Prize, an AI system winning a Nobel Prize, but I still return to the Turing test as a compelling test, the spirit of the Turing test as a compelling test.

   16. DHDemis Hassabis2:16

       Mm-hmm. Yeah, the Turing test, of course, it's been unbelievably influential, and Turing's one of my all-time heroes. But I think if you look back at the 1950 paper, his original paper, and read the original, you'll see I don't think he meant it to be a rigorous formal test. I think it was more like a thought experiment, almost a bit of philosophy he was writing if you look at the style of the paper. And you can see he didn't specify it very rigorously. So for example, he didn't specify the knowledge that the expert or judge would have. Um, not, you know, how much time would they have to investigate this? So these are important parameters if you were gonna make it a, a true sort of formal test. Um, and you know, some, by some measures, people claim the Turing test passed several, you know, a decade ago. I remember someone claiming that with a, with a kind of very bog standard normal, uh, uh, logic model, um, because they pretended it was a, it was a kid. So the, the judges thought that the machine, you know, was, was a, was a child. So, um, that would be very different from an expert AI person, uh, interrogating a machine and knowing how it was built and so on. So I think, um, you know, we should probably move away from that as a, as a formal test, and move more towards a, a general test where we test the AI capabilities on a range of tasks and see if it reaches human level or above performance on maybe thousands, perhaps even millions of tasks eventually, and cover the entire sort of cognitive space. So I think, um, for its time, it was an amazing thought experiment. And also 1950s, obviously, it was barely the dawn of the computer age, so of course, he only thought about text. And now, um, we have a lot more different inputs.

   17. LFLex Fridman3:54

       So yeah, maybe the better thing to test is the generalizability, so across multiple tasks, but, uh, I think it's also possible, as, as systems like GATO show, that eventually that might map right back to language. So you might be able to demonstrate your ability to generalize across tasks by then communicating your ability to generalize across tasks, which is kind of what we do through conversation anyway when we jump around. Ultimately, what's in there in that conversation is not just you moving around knowledge. It's you moving around, like, these entirely different modalities of understanding that ultimately map to your ability to, to, uh, operate successfully in all of these domains, which you can think of as tasks.

   18. DHDemis Hassabis4:42

       Yeah, I think certainly we as humans use language as our main generalization communication tool, so I think we end up thinking in language and expressing our solutions in language. Um, so it's gonna be a very powerful, uh, uh, mode in which to, uh, explain, you know, the system to explain what it's doing. Um, but I don't think it's the only, uh, uh, modality that matters. So I think there's gonna be a lot of, you know, there's, there's a lot of different ways to express, uh, capabilities, uh, o- other than just language.

   19. LFLex Fridman5:15

       Yeah, visual.

   20. DHDemis Hassabis5:16

       Yeah.

   21. LFLex Fridman5:16

       Robotics, body language. Um, yeah, actions, the interactive aspect of all that. That's all part of it.

   22. DHDemis Hassabis5:24

       But what's interesting with GATO is that it's, uh, it's, it's, it's sort of pushing prediction to the maximum in terms of, like, you know, uh, mapping arbitrary sequences to other sequences and sort of just predicting what's gonna happen next. So prediction seems to be, um, fundamental to intelligence.

   23. LFLex Fridman5:41

       And what you're predicting doesn't so much matter.

   24. DHDemis Hassabis5:44

       Yeah. It seems like you can generalize that quite well. So obviously language models predict the next word. Um, GATO predicts potentially any, uh, action or any token. Uh, and it's just the beginning really. It's our most general agent one could call it so far. But, um, you know, that itself can be scaled up massively more than we've done so far, and obviously we're in the, in the middle of doing that.

   25. LFLex Fridman6:04

       But the big part of solving AGI is creating benchmarks that help us get closer and closer, sort of, uh, creating benchmarks that test its generalizability. And it's just still interesting that this fella, Alan Turing, was one of the first, and probably still one of the only people that was trying, maybe philosophically, but was trying to formulate a benchmark that could be followed. It is... Even though it's, it's fuzzy, it's still sufficiently rigorous to where you can run that test. And I still think something like the Turing test will in- at the end of the day be the thing that truly impresses other humans, so that you can have a close friend who's an AI system. For that friend to be a good friend, they're going to have to be able to, uh, play StarCraft, and they're gonna have to do all of these tasks. Um, get you a beer, so the, the robotics tasks, uh, play games with you, use language, humor, all of those kinds of things, but that ultimately can boil down to language. It feels like in- not in terms of the AI community, but in terms of the actual impact of general intelligence on the world, it feels like language will be the place where it truly shines.

   26. DHDemis Hassabis7:18

       I think so, because it's such an important kind of input-output for us. I think you're right. I think the Turing test that what... the, the kind of, the, the philosophy behind it, which is the idea of can, can a machine mimic the behaviors of, uh, uh, a human in man- And I would say wider than just language and text. Um, then, you know, in terms of actions and everything else, creativity, all these things, then, um, if it can sort of match or exceed human cognitive capabilities, then I think we have a, you know, a true intelligence. So I th- from that perspective, you're right, I think he did formulate the right, uh, kind of, uh, setup.

   27. LFLex Fridman7:56

       I just... I think there'll be a kind of humor in the AI systems of the future looking back to this conversation-

   28. DHDemis Hassabis8:02

       (laughs)

   29. LFLex Fridman8:03

       (laughs) ... and thinking about the Turing test, and, uh, also thinking about by that time, they would know which year they were finally able to sort of cross the threshold of human-level intelligence a- and think how funny it is that we humans were still confused about this whole problem.

   30. DHDemis Hassabis8:21

       (laughs) Absolutely.
3. 8:27 – 30:02

   Video games

   1. LFLex Fridman8:27

      Anyway, so g- going back to your, to your journey, when did you fall in love with programming first?

   2. DHDemis Hassabis8:33

      Well, it was pretty, uh, it was pretty young age actually, so, um, you know, I started off, uh... Actually games was my first love, so starting to play chess when I was around four years old, and then, um, it was actually with winnings from a chess competition that I managed to buy my first chess computer when I was about eight years old. It was a ZX Spectrum, which was hugely popular in the UK at the time, and, uh, it was amazing machine because I think it tr- trained a whole generation of programmers in the UK because it was so accessible. You know, you literally switched it on and there was the basic prompt, and you could just get going. And, um, my parents didn't really know anything about computers so... but because it was my money from a chess competition, I could, I could say I, I, I wanted to buy it. Uh, and then, you know, I just went to bookstores, got books on programming, and, um, started typing in, you know, the, the programming code. And, and then of course, um, once you start doing that, you start adjusting it, and then making your own games, and that's when I fell in love with computers and realized that they were a very magical device. Um, in a way I kind of... I don't... would have been able to explain this at the time, but I felt that they were sort of almost a magical extension of your mind. I always had this feeling, and, and I've always loved this about computers, that you can set them off doing something, some task for you, you can go to sleep, come back the next day, and it's solved. Um, you know, that feels magical to me. So I mean, all machines do that to some extent, they all enhance our natural capabilities. Obviously cars make us... allow us to move faster than we can run, but this was a machine to extend the mind, and, uh, and then of course, AI is the ultimate expression of what a machine may be able to do or learn. So, um, very naturally for me, that thought extended into, into AI quite quickly.

   3. LFLex Fridman10:16

      Do you remember the, uh, the programming language that was first started in?

   4. DHDemis Hassabis10:20

      Yeah.

   5. LFLex Fridman10:20

      Was it spec- special to the machine or was it something general?

   6. DHDemis Hassabis10:22

      No, it was just a basic... It was just a...

   7. LFLex Fridman10:23

      It was BASIC?

   8. DHDemis Hassabis10:23

      I think it was just BASIC, uh, on the ZX Spectrum, I don't know what specific form it was, and then later on I got a Commodore Amiga, which, um-

   9. LFLex Fridman10:30

      Nice.

   10. DHDemis Hassabis10:30

       ... uh, was a fantastic machine. Uh-

   11. LFLex Fridman10:32

       Now you're just showing off.

   12. DHDemis Hassabis10:33

       So yeah. Well, lots of my friends had Atari STs and I, I, I managed to get an Amiga so it was a bit more powerful and, uh, and that was incredible, and used to do, um, programming in Assembler and, and, uh, also Amos BASIC, this, this, this specific form of BASIC. It was incredible actually. It's where I learned all my coding skills.

   13. LFLex Fridman10:51

       And when did you fall in love with AI? So when did you first start to gain an understanding that you can not just write programs that do some mathematical operations for you while you sleep, but something that's akin to b- bringing an entity to life? Sort of a, a thing that can figure out something more complicated than a, th- than a simple mathematical operation?

   14. DHDemis Hassabis11:15

       Yeah. So there was a few stages for me, all, all while I was very young. So first of all, as I was trying to improve at playing chess, so I was captaining various England junior chess teams, and at the time, when I was about, you know, maybe 10, 11 years old, I was gonna become a professional chess player. That was my first thought. Um, and-

   15. LFLex Fridman11:32

       So that dream was there, to-

   16. DHDemis Hassabis11:33

       Sure, sure.

   17. LFLex Fridman11:34

       ... to try to get to the highest levels of chess.

   18. DHDemis Hassabis11:35

       Yeah. So I was, um... You know, I got to... When I was about 12 years old, I got to master standard and I was second-highest rated player in the world to Judith Polgar, who obviously ended up being an amazing chess player and, uh, uh, world women's champion, and when I was trying to improve at chess, well, you know, what you do is you... Obviously, first of all, you're trying to improve your own thinking processes, so, uh, that leads you to thinking about thinking. How is your brain coming up with these ideas? Why is it making mistakes? How can you... uh, how can you improve that thought process? But the second thing is that you... It was just the beginning, this was like in the, in the eight... early '80s, mid '80s, of chess computers. If you remember, they were physical boards like the one we have in front of us, and p- you'd press down the, you know, the s- the squares, and I think Kasparov had a branded version of it that I, I, I got, and, um, you, uh... you know, you're used to the- they're not as strong as they are today, but they were, they were pretty strong, and you used to practice against them, um, uh, to try and improve your openings and other things. And so I remember, I think I probably got my first when I was around 11 or 12, and I remember thinking, "Um, this is amazing." You know, "How, how has someone programmed, uh, uh, this, this chess board to play chess?" Uh, and, uh, it was very formative book I bought which was called The Chess Computer Handbook by David Levy.

   19. LFLex Fridman12:49

       (laughs)

   20. DHDemis Hassabis12:49

       This game came out in 1984 or something, so I must have got it when I was about 11, 12. And it explained fully how these chess programs were made. And I remember my first AI program being, uh, programming my Amiga. Uh, it couldn't...... it wasn't powerful enough to play chess. I couldn't write a whole chess program, but I, I wrote a program for it to play Othello-

   21. LFLex Fridman13:06

       Mm-hmm.

   22. DHDemis Hassabis13:06

       ... or Reversi, it's sometimes called, I think, in the US, and, uh, so a slightly simpler game than chess. But I used all of the principles that chess programs had, alpha-beta search, all of that, and that was my first AI program. I remember that very well. I was around 12 years old. So that, that, that brought me into AI, and then the second part was later on, uh, when I was around 16, 17, and I was writing games professionally, designing games, uh, writing a game called Theme Park, which, um, had AI as a core gameplay component as part of the simulation, um, and it sold, you know, millions of copies around the world, and people loved the way that the AI, even though it was relatively simple by today's AI standards, um, was, was reacting to the way you as the player played it. So it was called a sandbox game, so it was one of the first types of games like that along with SimCity, and it meant that every game you played was unique.

   23. LFLex Fridman13:55

       Is there something you could say, just on a small tangent, about really impressive AI from a game design human enjoyment perspective, really impressive AI that you've seen in games, and maybe what does it take to create an AI system? Like how hard of a problem is that? So a million questions-

   24. DHDemis Hassabis14:15

       Yeah.

   25. LFLex Fridman14:16

       ... with, with just as a brief tangent.

   26. DHDemis Hassabis14:18

       Well, look, I think, um, games, uh, games have been significant in my life for three reasons. So first of all, to, to... I was playing them and training myself on games, uh, when I was a kid. Then, uh, I went through a phase of designing games and writing AI for games. So all the games I, I professionally wrote, uh, had AI as a core component, and that was mostly in the, in the 90s, and, um, the reason I was doing that in games industry was, at the time, the games industry, I think, was the cutting edge of technology. So whether it was graphics with people like John Carmack and Quake and those kind of things, or, or AI, I think, uh, actually all the action was going on in games, and, and we see, and we're still reaping the benefits of that even with things like GPUs, which, uh, you know, I find ironic. It was obviously invented for graphics, computer graphics, but then turns out to be amazingly useful for AI. It just turns out everything's a matrix multiplication appe- appears (laughs) you know, in the wh- in the whole world.

   27. LFLex Fridman15:10

       Yes.

   28. DHDemis Hassabis15:11

       So, um, so I think games at the time had the most cutting edge AI, and a lot of the, the, the games, uh, uh, we, we... You know, I was involved in writing, so there was a game called Black and White, which was one game I was involved with in the early stages of, which I still think is the most, um, impressive, uh, example of reinforcement learning in a computer game. So in that game, you know, you trained a little pet animal, uh, and, uh-

   29. LFLex Fridman15:34

       It's a brilliant game.

   30. DHDemis Hassabis15:35

       Yeah, and it sort of learnt from how you were treating it.
4. 30:02 – 32:13

   Simulation

   1. LFLex Fridman30:02

      do you think we're living in a simulation?

   2. DHDemis Hassabis30:03

      (laughs) Yes. Well, so okay, so I, I-

   3. LFLex Fridman30:06

      We're gonna jump around from the absurdly philosophical to the-

   4. DHDemis Hassabis30:09

      Sure.

   5. LFLex Fridman30:10

      ... to the technical.

   6. DHDemis Hassabis30:10

      Sure. Very, very happy to. So I think, uh, my answer to that question is a little bit complex because, uh, there is simulation theory, which obviously Nick Bostrom I think famously first proposed. Um, and, uh, I don't quite believe it in, in the, in that sense, so, um, in the s- in the sense that, uh, are we in some sort of computer game or have our descendants somehow recreated, uh, uh, Earth in the, you know, 21st century and, and, and some for some kind of experimental reason? I think that, um, but I do think that we, that, that, that we might be... that the best way to understand physics and the universe is from a computational perspective. So understanding it as an information universe, and actually information being the most fundamental unit of, uh, uh, reality, rather than matter or energy. So a physicist would say, you know, matter or energy, you know, E equals MC squared, these are the things that are, are, are the fundamentals of the universe. I'd actually say information, um, which of course itself can be... can specify energy or matter, right? Matter is actually just, you know, we're, we're just... out the way our bodies and all the molecules in our body arrange is information. So I think information may be the most fundamental way to describe the universe, and therefore you could say we're in some sort of simulation because of that. Um, but I do- I don't... I'm not, I'm not really a subscriber to the idea that, um, you know, they, these are sort of throwaway billions of simulations around. I think this is actually very critical and possibly unique, this simulation. Um-

   7. LFLex Fridman31:41

      This particular one?

   8. DHDemis Hassabis31:42

      Yes.

   9. LFLex Fridman31:42

      So, but, and y- you just mean-... tre- treating the universe as a computer that's processing and modifying information is, is a good way to solve the problems of physics, of chemistry-

   10. DHDemis Hassabis31:55

       Yeah.

   11. LFLex Fridman31:55

       ... of biology-

   12. DHDemis Hassabis31:56

       Yes.

   13. LFLex Fridman31:57

       ... and perhaps of humanity and so on?

   14. DHDemis Hassabis31:59

       Yes. I think, uh, understanding physics i- in terms of information theory, uh, might be the best way to, to really, uh, understand what's going on here, uh-

   15. LFLex Fridman32:09

       From our understanding of a universal Turing machine,

5. 32:13 – 37:13

   Consciousness

   1. LFLex Fridman32:13

      from our understanding of a computer, do you think there's something outside of the capabilities of a computer that is present in our universe? You have a disagreement with Roger Penrose about-

   2. DHDemis Hassabis32:23

      Yes.

   3. LFLex Fridman32:23

      ... the nature of consciousness. He d- he thinks that consciousness is more than just a computation.

   4. DHDemis Hassabis32:29

      Mm-hmm.

   5. LFLex Fridman32:29

      Uh, do you think all of it, the whole shebang is a com- can be, can be a computation?

   6. DHDemis Hassabis32:34

      Yeah, I've had many fascinating debates with, uh, uh, Sir Roger Penrose, and obviously he's, he's famously, and- and I read, you know, Emperor's of the New Mind and, um, and his books, uh, his classical books, uh, and they, they were pretty influential in the- you know, in the 90s. And, um, he believes that there's something more, you know, something quantum that is needed to explain consciousness in the brain. Um, I think about what we're doing actually, at DeepMind and what my career is being, we're almost like Turing's champion. So we are pushing Turing machines or classical computation to the limits. What are the limits of what classical computing can do? Now, um, and at the same time I've also studied neuroscience to see, and that's why I did my PhD in, was to see... also to look at, you know, is there anything quantum in the brain from a neuroscience or biological perspective? And, um, and so far, I think most neuroscientists and most mainstream biologists and neuroscientists would say there's no evidence of any quantum, uh, systems or effects in the brain. As far as we can see, it's- it can be mostly explained by classical, uh, classical theories. So, uh, and then so there's sort of the, the search from the biology side, and then at the same time, there's the raising of the water, uh, at the bar from what classical Turing machines can do. Uh, uh, and, and- and you know, including our new AI systems. And, uh, as you alluded to earlier, um, you know, I think AI, uh, especially in the last decade plus has been a continual story now of surprising, uh, events, uh, and surprising successes knocking over one theory after another of what was thought to be impossible. You know, from Go to protein folding and so on. And so I think, um, I would be very hesitant to bet against how far the, uh, universal Turing machine and classical computation paradigm can go. And- and my betting would be that all of certainly what's going on in our brain, uh, can probably be mimicked or- or approximated on a, on a classical machine. Um, not, you know, not requiring something metaphysical or- or quantum.

   7. LFLex Fridman34:38

      And we'll get there with some of the work with AlphaFold which I think begins the journey of modeling this beautiful and complex world of biology.

   8. DHDemis Hassabis34:47

      Mm-hmm.

   9. LFLex Fridman34:48

      So you think all the magic of the human mind comes from this, just a few pounds of mush? Of bi- of biological com- computational mush that's akin to some of the neural networks, not directly but in spirit, that, uh, DeepMind has been working with?

   10. DHDemis Hassabis35:06

       Well, look, I think it's, um, you say it's a few... you know, of course it's... this is the- I think the biggest miracle of the universe, is that, um, it is just a few pounds of mush in our skulls-

   11. LFLex Fridman35:14

       Yeah.

   12. DHDemis Hassabis35:15

       ... and yet it's also... our brains are the most complex objects in the... that we know of in the universe. So there's something profoundly beautiful and amazing about our brains, and I think that it's an incredibly, uh, uh, incredible eff- efficient machine and- and, uh, uh, uh... and it's, uh, it's, you know, p- phenomenon basically. And I think that building AI, one of the reasons I wanna build AI and I've always wanted to is, I think by building an intelligent artifact like AI and then comparing it to the human mind, um, that will, uh, help us unlock the uniqueness and the true secrets of the mind that we've always wondered about since the dawn of history. Like consciousness, dreaming, uh, creativity, uh, uh, em- emotions. What are all these things? Right? We've, we've, we've wondered about them since, since the dawn of humanity and I think one of the reasons and, you know, I love philosophy and philosophy of mind is, we- we've found it difficult is there haven't been the tools for us to really, other than introspection-

   13. LFLex Fridman36:13

       Mm-hmm.

   14. DHDemis Hassabis36:13

       ... to- from very clever people in- in- in history, very clever philosophers, to really investigate this scientifically. But now suddenly we have a plethora of tools. Firstly, we have all the neuroscience tools, fMRI machines, single cell recording, all of this stuff, but we also have the ability, computers and AI, to build, uh, intelligent systems. So I think that, um, uh, you know, I- I think it is amazing what the human mind does and, um, and- and I- I'm kind of in awe of it really and, uh, and I think it's amazing that with our human minds we're able to build things like computers and- and actually even, you know, think and investigate about these questions. I think that's also a testament to the human mind.

   15. LFLex Fridman36:52

       Yeah. The universe built the human mind that now is building computers that help us understand both the universe and our own human mind.

   16. DHDemis Hassabis37:01

       That's right. That's exactly it. I mean, I think that's one... you know, one could say we- we are... maybe we're the mechanism by which the universe is going to try and understand itself.

   17. LFLex Fridman37:10

       Yeah. (laughs)

   18. DHDemis Hassabis37:12

       (laughs)

   19. LFLex Fridman37:12

       Ah, it's beautiful.

6. 37:13 – 50:53

   AlphaFold

   1. LFLex Fridman37:13

      So let's, uh, let's go to the basic building blocks of biology that I think is another angle at which you can start to understand the human mind, the human body, which is quite fascinating, which is from the basic building blocks, start to simulate, start to model how from those building blocks you can construct bigger and bigger and more complex systems. Maybe one day the entirety of the human biology. So here's another problem that thought to be impossible to solve, which is protein folding, and AlphaFold or specific AlphaFold2, uh, did just that. It solved protein folding. I think it's one of the biggest breakthroughs.... uh, certainly in the history of structural biology, but, uh, in general in, in science. Um, maybe from a high level, what is it and how does it work?

   2. DHDemis Hassabis38:04

      Mm-hmm.

   3. LFLex Fridman38:04

      And then we can ask some fascinating-

   4. DHDemis Hassabis38:07

      Sure. (laughs)

   5. LFLex Fridman38:07

      ... questions after.

   6. DHDemis Hassabis38:08

      Sure. Um, so maybe, uh, to explain it, uh, to people not familiar with protein folding is, you know, I first of all explain proteins, which is, you know, proteins are essential to all life. Every function in your body depends on proteins. Som- sometimes they're called the workhorses of biology. And if you look into them, and I've, you know, obviously as part of AlphaFold I've been researching proteins and, and, and pr- and structural biology for the last few years. You know, they're amazing little bio nano-machines, proteins. They're incredible, if you actually watch little videos of how they work, animations of how they work. And, um, proteins are specified by their genetic sequence, called their amino acid sequence, so you can think of it as their, their genetic, uh, make-up. And then in the body, uh, in, in nature, they- when they- when- whe- they fold up into a 3D structure. So you can think of it as a string of beads, and then they fold up into a ball. Now the key thing is, you want to know what that 3D structure is, um, because the structure, the 3D structure of a protein, uh, is what, uh, helps to determine what does it do, the function it does in your body. Uh, and also if you're interested in drug- drugs or, or disease, you need to understand that 3D structure, because if you want to target something with a drug compound a- about to block th- something the protein's doing, uh, you need to understand where it's gonna bind on the surface of the protein. So obviously for t- in order to do that, you need to s- n- understand the 3D structure.

   7. LFLex Fridman39:26

      So the structure is mapped to the function?

   8. DHDemis Hassabis39:28

      The structure is mapped to the function, and the structure is obviously somehow specified by the, by the amino acid sequence. And that's the p- in essence the protein folding problem, is, can you, just from the amino acid sequence, the one-dimensional, uh, string of letters, can you i- uh, immediately computationally predict the 3D structure?

   9. LFLex Fridman39:46

      Right.

   10. DHDemis Hassabis39:47

       And this has been a, a grand challenge in biology for over 50 years, so I think it was first articulated by Christian Anfinsen, a Nobel Prize winner in 1972, uh, as part of his Nobel Prize-winning lecture. And he just speculated this should be possible, to go from the amino acid sequence to the 3D structure, but he didn't say how. So it's a- i- i- I, you know, I've been- it's been described to me as, as equivalent to Fermat's Last Theorem-

   11. LFLex Fridman40:11

       Yeah.

   12. DHDemis Hassabis40:11

       ... but for biology, right?

   13. LFLex Fridman40:12

       Y- you should... As somebody that, uh, very well might win the Nobel Prize in the future, but outside of that, y- you should do more of that kind of thing. In the margins just put random things-

   14. DHDemis Hassabis40:22

       (laughs) Yeah, rand- exactly.

   15. LFLex Fridman40:22

       ... that will take like 200 years to solve.

   16. DHDemis Hassabis40:24

       Set people off for 200 years.

   17. LFLex Fridman40:26

       It should be possible-

   18. DHDemis Hassabis40:27

       Exactly.

   19. LFLex Fridman40:27

       ... and just don't give any details.

   20. DHDemis Hassabis40:29

       Exactly. I think everyone sh- exactly. It should be... I'll, I'll have to remember that for future. So yeah, so he set off, you know, with this one throwaway remark, just like Fermat, you know, he, he set off this whole 50-year, uh, uh, uh, uh, field really of c- of computational biology. And, and they had, you know, they got stuck. They hadn't really got very far with doing this, and, and, um, until now, until AlphaFold came along, this is done experimentally, right? Very painstakingly. So the rule of thumb is... And you have to, like, crystallize the protein, which is really difficult. Some proteins can't be crystallized, like membrane proteins. And then you have to use very expensive electron microscopes or X-ray crystallography machines. Really painstaking work to get the 3D structure and visualize a 3D structure. So the rule of thumb in, in, in experimental biology is that it takes one PhD student their entire PhD to do one protein. Uh, and with AlphaFold2, we're able to predict the 3D structure in a matter of seconds. Um, and so we were, you know, over Christmas, we did the whole human proteome, all, every protein in the human body, all 20,000 proteins. So the human proteome is like the equivalent of the human genome, but on proteome space. And, uh, and sort of revolutionized really what, uh, uh, uh, structural biologists can do, because now, um, they don't have to worry about these painstaking experimentals, you know, should they put all of their effort in or not. They can almost just look up the structure of their proteins like a Google search.

   21. LFLex Fridman41:53

       And so there's a dataset on which it's trained and how to map this amino acid sequence. First of all, it's incredible that a protein, this little chemical computer is able to do that computation itself in a some kind of distributed way and do it very quickly.

   22. DHDemis Hassabis42:07

       Mm-hmm.

   23. LFLex Fridman42:07

       That's a weird thing, and they evolved that way 'cause, you know, in the beginning, I mean, that's a great invention, just the protein itself of Earth.

   24. DHDemis Hassabis42:15

       Yes. I mean-

   25. LFLex Fridman42:15

       A- and then they- there's I think probably a, a history of, like, uh, they evolved, uh, to have many of these proteins and those proteins figure out how to be computers themselves s- in such a way that you can create structures that can interact in complexes with each other in order to form high level functions. I mean, it's a weird system that they've figured it out.

   26. DHDemis Hassabis42:35

       Well, for sure. I mean, we, you know, maybe we should talk about the origins of life too, but proteins themselves I think are magical and incredible, uh, uh, uh, as I said, little, little bio nano-machines, and, um, and, and actually Le- Levinthal, who was another scientist, uh, a c- a contemporary of Anfinsen, uh, he, he, he coined this Levinthal, what became known as Levinthal's paradox, which is exactly what you're saying. He calculated roughly a prot- an average protein, which is maybe 2,000 amino acids, uh, uh, bases long, is, um, is, is- can fold in maybe 10 to the power of 300 different conformations. So there's 10 to the power of 300 different ways that protein could fold up, and yet somehow, in nature, physics solves this, "solves" this in a matter of milliseconds. So proteins fold up in your body in, you know, sometimes in, in fractions of a, of a second.

   27. LFLex Fridman43:25

       Mm-hmm.

   28. DHDemis Hassabis43:25

       So h- physics is somehow solving that search problem.

   29. LFLex Fridman43:29

       And just to be clear-

   30. DHDemis Hassabis43:30

       Yeah.
7. 50:53 – 1:03:12

   Solving intelligence

   1. LFLex Fridman50:53

      let's try to think through this because you're in a very interesting position where DeepMind is a place of some of the most, uh, brilliant ideas in the history of AI, but it's also a place of brilliant engineering. (inhales deeply) So how much of solving intelligence, this big goal for DeepMind, how much of it is science? How much is engineering? So how much is the algorithms? How much is the data?

   2. DHDemis Hassabis51:15

      Mm-hmm.

   3. LFLex Fridman51:16

      How much is the hardware compute infrastructure? How much is it the software compute infrastructure?

   4. DHDemis Hassabis51:22

      Yeah.

   5. LFLex Fridman51:23

      Um, what else is there? How much is the human infrastructure?

   6. DHDemis Hassabis51:26

      (laughs)

   7. LFLex Fridman51:27

      And, like, just the humans interact in certain kinds of ways-

   8. DHDemis Hassabis51:30

      Sure.

   9. LFLex Fridman51:30

      ... you know, all the s- space of all those ideas, and how much is maybe, like, philosophy? How much... What's the key? If, um, uh, if, if you were to sort of look back, like, if we go forward 200 years and look back, what was the key thing that solved intelligence? Is it the ideas or the engineering?

   10. DHDemis Hassabis51:48

       Well, I think it's a, I think it's a combination. Uh, first of all, of course, it's a combination of all those things, but the, the ratios of them changed over, over time.

   11. LFLex Fridman51:54

       (laughs) Yeah.

   12. DHDemis Hassabis51:55

       So, so, um-

   13. LFLex Fridman51:56

       For sure.

   14. DHDemis Hassabis51:56

       ... even in the last 12 years, so we started DeepMind in 2010, which is hard to imagine now because 2010, it's only 12 short years ago, but nobody was talking about AI. Uh, uh, you know, I don't know if you remember back to your MIT days, uh, th- you know, no one was talking about it. I w- I did a postdoc at MIT back around then, and it was sort of thought of as a, "Well, look, we know AI doesn't work. We tried this hard in the '90s" at places like MIT.

   15. LFLex Fridman52:16

       Yes.

   16. DHDemis Hassabis52:17

       Mostly lo- using, using logic systems and old-fashioned sort of good old-fashioned AI we would call it now. Um, people like Minsky and, and, and, and Patrick Winston and you know all these characters, right?

   17. LFLex Fridman52:26

       Yeah.

   18. DHDemis Hassabis52:26

       And used to debate a few of them, and they used to think I was mad thinking about that some new advance could be done with learning systems, and-

   19. LFLex Fridman52:32

       Yes.

   20. DHDemis Hassabis52:32

       ... um, I was actually pleased to hear that because at least you know you're on a unique track at that point.

   21. LFLex Fridman52:37

       (laughs)

   22. DHDemis Hassabis52:37

       Right? Even if every, all of your s- you know, professors are telling you you're mad.

   23. LFLex Fridman52:41

       Yeah, that's true. That's true.

   24. DHDemis Hassabis52:41

       And of course, in industry, uh, you couldn't, we couldn't get, you know... It was difficult to get two cents together, uh, and which is hard to imagine now as well given that it's the biggest sort of buzzword in, in VCs and, and, and fundraising's easy and all these kind of things today. So back in 2010, it was very difficult, and what we, the reason we started then, and, and Shane and I used to discuss, um, uh, uh, what were the sort of founding tenets of DeepMind? And it was ver- various things. One was, um, algorithmic advances, so deep learning. You know, Geoff Hinton and co. had just, had just sort of invented that in academia, but no one in industry knew about it. Uh, we, we love reinforcement learning. We thought that could be scaled up. But also understanding about the human brain had advanced, um, quite a lot, uh, in the, in the decade prior with fMRI machines and other things. So we could get some good hints about architectures and algorithms and, and, and sort of, um, representations maybe that the brain uses, so as, at a systems level, not at a implementation level. Um, and then the other big things were compute and GPUs, right? So we could see, uh, a compute was gonna be really useful, and it got to a place where it become commoditized mostly through the games industry, and, and that could be taken advantage of. And then the final thing was also mathematical and theoretical definitions of intelligence, so things like AIXI, AIXI, which, uh, Shane worked on with his supervisor Marcus Hutter, which is a sort of theoretical, uh, proof really of universal intelligence, um, which is actually a reinforcement learning system.

   25. LFLex Fridman54:07

       Mm-hmm.

   26. DHDemis Hassabis54:07

       Um, uh, in the limit. I mean, it assumes infinite compute and infinite memory in the way Tu- you know, like a Turing machine proof.

   27. LFLex Fridman54:12

       Yes.

   28. DHDemis Hassabis54:12

       But I was also waiting to see something like that too to s- uh, you know, like Turing machines, uh, and, and computation theory that people like Turing and Shannon came up with underpins modern computer science.

   29. LFLex Fridman54:23

       Mm-hmm.

   30. DHDemis Hassabis54:24

       Um, uh, you know, I was waiting for a theory like that to sort of underpin AGI research. So when I, you know, met Shane and saw he was working on something like that, you know, that to me was a sort of final piece of the jigsaw. So in the early days, I would say that, uh, ideas were the most important, uh, you know, and for us it was deep reinforcement learning, scaling up deep learning, um, of course we've seen transformers. So huge leaps I would say, you know, three or four from, for- if you think from 2010 till now, uh, huge evolutions, things like AlphaGo. Um, and, um, and, and maybe there's a few more still needed. But as we get closer to AI, AGI, um, I think it'll- engineering becomes more and more important, and data, because scale and of course the, the recent, you know, results of GPT-3 and all the big language models and large models, including our ones, uh, has shown that scale is a, is... and large models are clearly gonna be a necessary, but perhaps not sufficient part of an AGI solution.
8. 1:03:12 – 1:13:18

   Open sourcing AlphaFold & MuJoCo

   1. LFLex Fridman1:03:12

      you, you mentioned you've open-sourced AlphaFold, uh, and even the data involved. To me personally, also really happy and a big thank you for open-sourcing MuJoCo, uh, the physics simulation engine that's, um, that's often used for robotics research and so on. So I think that's a pretty gangster move. Uh, so what- (laughs) what's the- (laughs) ... What's... I mean, this... uh, very few companies or people would do that kind of thing. What's the philosophy behind that?

   2. DHDemis Hassabis1:03:41

      You know, it's a case-by-case basis, and in both those cases, we felt that was the maximum benefit to humanity to do that and, and the scientific community, in one case, the robotics, uh, physics community with MuJoCo, so-

   3. LFLex Fridman1:03:53

      You purchased it and then open-sourced it.

   4. DHDemis Hassabis1:03:54

      We purchased it for to op- yes, we purchased it for the express principle to open-source it.

   5. LFLex Fridman1:03:58

      (laughs) . That's awesome.

   6. DHDemis Hassabis1:03:59

      So, um, so, uh-

   7. LFLex Fridman1:04:01

      (laughs) .

   8. DHDemis Hassabis1:04:01

      ... you know, hope people appreciate that. It's great to hear that, that you do.

   9. LFLex Fridman1:04:04

      Yeah.

   10. DHDemis Hassabis1:04:04

       And then the second thing was... an- mostly we did it because the person building it, who's, uh, uh, uh, would not, uh, was not able to cope with supporting it anymore 'cause it was, it got too big for, for him. He's an amazing professor, uh, who, who built it in the first place. So we helped him out with that. And then with AlphaFold, it's even bigger, I would say. And I think in that case, we decided that there were so many downstream applications of AlphaFold, um, that we couldn't possibly even imagine what they all were. So the best way to accelerate, uh, drug discovery and also fundamental research would be to, to, um, give all that data away and, and, and the, and the, and the system itself. Um, you know, it, it's been so gratifying to see what people have done that within just one year, which is a short amount of time in science. And, uh, it's being used by over 500,000 researchers have used it. We think that's almost every biologist in the world. I think there's roughly 500,000 biologists in the world, professional biologists, have used it to, to look at their, um, proteins of interest. Um, we've seen amazing fundamental research done. So, uh, a couple of weeks ago, front cover of... there was a whole special issue of Science, uh, including the front cover, which had the nuclear pore complex on it, which is one of the biggest proteins in the body. The nuclear pore complex is a protein that governs all the nutrients going in and out of your cell nucleus. So it's, they're, they're like little hole- gateways that open and close to let things go in and out of your cell nucleus. So they're really important. Um, but they're huge, 'cause they're massive donut ring-shaped things. And they've been looking to try and figure out that structure for decades, and they have lots of, you know, experimental data, but it's too low resolution, there's bits missing, and they were able to, like a giant Lego jigsaw puzzle, use AlphaFold predictions plus experimental data and combined those two independent sources of information. Uh, actually four different groups around the world were able to put it together at the sa- uh, more or less simultaneously using AlphaFold predictions. So that's been amazing to see. And pretty much every pharma company, every drug company executive I've spoken to has said that their teams are using AlphaFold to accelerate whatever drugs, uh, uh, uh, they're- they're trying to discover. So I think the knock-on effect has been i- enormous in terms of, uh, uh, the impact that, uh, AlphaFold has made.

   11. LFLex Fridman1:06:15

       And it's probably bringing in... it's creating biologists. It's bringing more people into the field, um, b- both on the excitement and both on the technical skills involved, and, um, it's almost like a, a gateway drug to biology.

   12. DHDemis Hassabis1:06:28

       Yes, it is.

   13. LFLex Fridman1:06:29

       You follow- (laughs)

   14. DHDemis Hassabis1:06:29

       And to get more computational people involved, too-

   15. LFLex Fridman1:06:31

       Exactly.

   16. DHDemis Hassabis1:06:31

       ... hopefully. And, and I think for us, you know, the next stage, as I said, you know, in future, we have to have other considerations, too. We're building on top of AlphaFold and these other ideas I discussed with you about protein-protein interactions and, and genomics and other things, and not everything will be open source. Some of it we'll, we'll do commercially, 'cause that will be the best way to actually get the most resources and impact behind it. In other ways, so oth- some other projects we'll do nonprofit style. Um, and also we have to consider, for future things as well, safety and ethics as well, like bo- you know, synthetic biology. There are... you know, there is dual use, and we have to think about that as well. With AlphaFold, we, you know, we consulted with 30 different bioethicists and, and other people expert in this field to make sure it was safe before, um, we released it. So there'll be other considerations in the future. But for right now, you know, I think AlphaFold is a, a kind of a, a gift from us to, to, to the scientific community.

   17. LFLex Fridman1:07:20

       So I'm pretty sure that something like AlphaFold, uh, would be part of Nobel Prizes in the future, but us humans, of course, are horrible with credit assignment, so we'll, of course, give it to the humans. Um, do you think there will be a day when AI system can't be denied that it earned that Nobel Prize? Do you think we will see that in the 21st century?

   18. DHDemis Hassabis1:07:47

       It depends what type of AIs we end up building, right? Whether they're, um...... you know, goal-seeking agents, who specifies the goals, uh, who comes up with the hypotheses, who, you know, who determines which problems to tackle, right? So, I think-

   19. LFLex Fridman1:08:01

       And tweets about it, announcement of the results.

   20. DHDemis Hassabis1:08:02

       Yes. And tweets announcing the re- results, exactly, is part of it. Um, so I think right now, of course, it's, it's, i- i- it's, it's amazing human ingenuity that's behind these systems, and then the system, in my opinion, is just a tool. You know, it would be a bit like saying with Galileo and his telescope, you know, the ingenuity... the, the, the credit should go to the telescope. I mean, it's clearly Galileo building the tool which he then uses. So, I still see that in the same way today, even though th- these tools learn for themselves. Um, they are, I think of, I think of things like AlphaFold and the, the things we're building as the ultimate tools for science and for, for acquiring new knowledge, to help us as scientists acquire new knowledge. I think one day, there will come a point where an AI system may solve or come up with something like general relativity off its own bat, not just by averaging everything on the internet or averaging everything on PubMed. Although, that would be interesting to see what that would come up with. Um, so that, to me, is a bit like our earlier debate about creativity, you know? Inventing Go, rather than just coming up with a good Go move. And, um, so I think, uh, solving... I think to, to, you know... if we wanted to give it the credit of like a Nobel type of thing, then it would need to invent Go, uh, and sort of invent that new conjecture out of the blue, um, rather than being specified by the, the, the human scientist or the human creator. So I think right now, that's... it's definitely just a tool.

   21. LFLex Fridman1:09:26

       Although it is interesting how far you get by averaging everything on the internet, like you said, because, you know, a lot of people do see science as you're always standing on the shoulders of giants. And the question is, how much are you really reaching up above the shoulders of giants? Maybe it's just assimilating different kinds of results of the past with, ultimately, this new perspective that gives you this breakthrough idea. But e- that idea may not be novel in the way that we... can't be already discovered on the internet. Maybe the Nobel Prizes of the next 100 years are already all there on the internet to be discovered.

   22. DHDemis Hassabis1:10:03

       They, they could be, they could be. I mean, I think, um... this is one of the big mysteries, I think, is that, uh, uh... I, I... first of all, I believe a lot of the big new breakthroughs that are gonna come in the next few decades, and even in the last decade, uh, are gonna come at the intersection between different subject areas, where, um, there'll be some new connection that's found between what seemingly were disparate areas. Uh, and one can even think of DeepMind as a... as I said earlier, as a, as a sort of interdisciplinary between neuroscience ideas and, and AI engineering ideas, uh, originally. And so, um, so I think there's that. And then one of the things we can't imagine today is... and w- one of the reasons I think people... we were so surprised by how well large models worked is that, actually, it's very hard for our human minds, our limited human minds, to understand what it would be like to read the whole internet, right? I think we can do a thought experiment, and I used to do this, of like, "Well, what if I read the whole of Wikipedia? Uh, what would I know?" And I think our minds can just about comprehend maybe what that would be like. But th- the whole internet is beyond comprehension. So I think we just don't understand what it would be like to h- be able to hold all of that in mind potentially, right? And then f- uh, active a- at once, and then maybe what are the connections that are available there? So I think, no doubt, there are huge things to be discovered just like that. But I do think there is this other type of creativity, of true spark of new knowledge, new idea, never thought before about, can't be averaged from things that are known, um, that really... of course, everything come, you know... nobody creates in a vacuum, so there must be clues somewhere, but just a unique way of putting those things together. I think some of the greatest scientists in history have displayed that, I would say. Although it's very hard to know, going back to their time, what was exactly known, uh, when they came up with those things.

   23. LFLex Fridman1:11:48

       Although, I, uh, the... you're, you're making me really think, because just the thought experiment of deeply knowing a hundred Wikipedia pages-

   24. DHDemis Hassabis1:11:56

       Mm-hmm.

   25. LFLex Fridman1:11:57

       I don't think I can, um... I've been really impressed by Wikipedia, and for d- for technical topics.

   26. DHDemis Hassabis1:12:03

       Yeah.

   27. LFLex Fridman1:12:03

       So if you know a hundred pages or a thousand pages, I don't think we can visu- truly comprehend what's... what kind of intelligence that is.

   28. DHDemis Hassabis1:12:13

       Yeah.

   29. LFLex Fridman1:12:13

       That's a pretty powerful intel-... if you know how to use that in... integrate that information correctly-

   30. DHDemis Hassabis1:12:18

       Yes.
9. 1:13:18 – 1:17:22

   Nuclear fusion

   1. LFLex Fridman1:13:18

      you have a paper on nuclear fusion. Uh, magnetic control of tachonic plasmas through deep reinforcement learning. So you, uh, you're seeking to solve nuclear fusion with deep RL, uh, so it's doing control of high temperature plasmas. Can you explain this work? And, uh, can AI eventually solve nuclear fusion?

   2. DHDemis Hassabis1:13:36

      (laughs) It's been very fun last year or two, uh, and very productive because we've been ticking off a lot of my, uh, dream projects, if you like-

   3. LFLex Fridman1:13:43

      Nice.

   4. DHDemis Hassabis1:13:43

      ... of things that I've collected over the years of, of areas of science that I would like to, I think could be very transformative if we helped accelerate and are, are really interesting problems, scientific challenges in and of themselves. And-

   5. LFLex Fridman1:13:56

      So this is energy?

   6. DHDemis Hassabis1:13:57

      So, energy. Yes, exactly. So, energy and climate. So we talked about disease and biology as being one of the biggest pla- places I think AI can help with. I think energy and climate, uh, is another one. And so maybe they would be my top two. Um, and fusion is one, one area I think AI can help with. Now, fusion has many challenges, mostly physics, material science, and engineering challenges as well to, to build these massive fusion reactors and contain the plasma. And what we try to do whenever we go into a new field, i- uh, to apply our systems, is we look for, um, we talk to domain experts. We try and find the best people in the world to collaborate with. Um, in this case, in fusion, we, we collaborated with EPFL in Switzerland, the Swiss Technical Institute, who are amazing. They have a test reactor that they were willing to let us use, which, you know, I double-checked with the team, we were gonna use carefully and safely.

Episode duration: 2:10:38