Google DeepMind CEO Demis Hassabis: The Path To AGI, Deceptive AIs, Building a Virtual Cell

1. 0:00 – 2:46

   The Journey Towards AGI

   1. AKAlex Kantrowitz0:00

      Google DeepMind CEO and Nobel laureate Demis Hassabis joins us to talk about the path toward artificial general intelligence, Google's AI roadmap, and how AI research is driving scientific discovery. That's coming up right after this. Welcome to Big Technology Podcast, a show for cool-headed, nuanced conversation of the tech world and beyond. Today, we're at Google DeepMind headquarters in London for what promises to be a fascinating conversation with Google DeepMind CEO Demis Hassabis. Demis, great to see you again. Welcome to the show.

   2. DHDemis Hassabis0:30

      Thanks for having me on the show.

   3. AKAlex Kantrowitz0:32

      Definitely. It's great to be here. So every research house right now is working toward building AI that mirrors human intelligence, human-level intelligence. They call it AGI. Where are we right now in the progression, and how long is it gonna take to get there?

   4. DHDemis Hassabis0:48

      Well, look, I mean, of course, the last, uh, few years has been an incredible amount of progress, um, actually, you know, maybe over the last decade plus. Um, this is what's on everyone's lips right now, and the debate is, is how close are we to AGI? What's the correct definition of AGI? Um, we've been working on this for more than 20-plus years. Um, we've sort of had a consistent view about AGI being a system that's capable of exhibiting all the cognitive capabilities humans can. Um, and I think we're getting, you know, closer and closer, but I think we're still probably, uh, a handful of years away.

   5. AKAlex Kantrowitz1:21

      Okay. And so what is it gonna take to get there?

   6. DHDemis Hassabis1:24

      So the models today are pretty capable. Of course, we've all interacted with the language models, and, and now they're becoming multimodal. I think there are still some missing attributes, things like reasoning, um, hierarchical planning, uh, long-term memory. Um, there's quite a few capabilities that, uh, the current systems, uh, I would say don't have. They're also not consistent across the board. You know, they're very, very strong in some things, but they're still surprisingly weak and flawed in, in other areas. So you'd want, uh, an, an AGI to have pretty consistent, robust behavior across the board, all the cognitive tasks. And I think one thing that's clearly missing, and I always, always had as a benchmark for, for AGI, was the ability for these systems to invent their own hypotheses or conjectures about science, not just prove existing ones. So of course, that's extremely useful already to prove an existing maths conjecture or something like that, or, or play a game of Go to a world champion level, but could a system invent Go? Could it come up with a new Riemann hypothesis? Uh, or could it have come up with relativity, um, back in the days that Einstein did it with the information that he had? And I think today's systems are still pretty far away from having that kind of creative, uh, inventive capability.

   7. AKAlex Kantrowitz2:35

      Okay, so a couple years away till we hit AGI.

   8. DHDemis Hassabis2:37

      I think, um, you know, I, I would say probably, like, three to five years away at least.

   9. AKAlex Kantrowitz2:41

      So if someone were to declare that they've reached AGI in 2025, probably marketing.

2. 2:46 – 5:53

   Current AI Capabilities and Limitations

   1. DHDemis Hassabis2:46

      I think so. I mean, I think there's a lot of, um, uh, hype in the area, of course. Uh, I mean, some of it's very justified. I mean, I would say that, um, AI, AI research today is, um, overestimated in the short term, um, I think probably a bit overhyped at this point, um, but still underappreciated and, um, and some- and very underrated about what it's gonna do in the medium to long term. Um, so it's sort of we're still in that weird kind of space. Uh, and I think part of that is, you know, there's a lot of pe- uh, people that need to do fundraising, a lot of startups and other things, and so I think we're going to have quite a few sort of fairly outlandish and, and, and slightly, uh, exaggerated claims. Um, and you know, I think that's a bit of a shame, actually.

   2. AKAlex Kantrowitz3:30

      Yeah. In the AI, in the AI products-

   3. DHDemis Hassabis3:33

      Mm.

   4. AKAlex Kantrowitz3:33

      What's it gonna look like on the path there? I mean, you've talked about memory again, planning-

   5. DHDemis Hassabis3:37

      Yeah

   6. AKAlex Kantrowitz3:37

      ... um, being better at some of the tasks that it's not excelling at at the moment.

   7. DHDemis Hassabis3:41

      Mm-hmm.

   8. AKAlex Kantrowitz3:41

      So when we're using these AI products, let's say we're using Gemini-

   9. DHDemis Hassabis3:44

      Mm-hmm

   10. AKAlex Kantrowitz3:45

       ... what are some of the things that we should look for in these domains that will make us say, "Oh, okay, it seems like it's, that's a step closer, and that's a step closer"?

   11. DHDemis Hassabis3:52

       Yeah, so I think, um, uh, today's systems, you know, obviously we're very proud of Gemini 2.0. I'm sure we're gonna talk about that, but I feel like, um, they're very useful for still quite niche tasks, right? If you're doing some research, um, perhaps you're summarizing some area of research, incredible. You know, I use NotebookLM and Deep Research all the time to kind of especially, like, um, uh, break the ice on a new area of research that I want to get into or summarize some, you know, uh, maybe a, a fairly mundane set of documents or something like that. So they're extremely good for certain tasks, and then people are getting a lot of value out of them. Um, but they're still not pervasive, in my opinion, in everyday life, like helping me every day with my research, my work, my day-to-day, um, my daily life too. And I think that's where we're going with our products with building things like, um, As- Project Astra. Our vision for a universal assistant is it should be, uh, involved in all aspects of your life and be enriching, helpful, and, and making that more efficient. And I think part of the reason is these systems are still fairly brittle, partly because they are quite flawed still, and they're not, uh, AGIs, and, and you have to be quite specific, for example, with your prompts, or they... You need a lot of, there's quite a lot of skill there in, in coaching or guiding these systems to be useful and to stick to, uh, uh, the areas they're good at. And, and a true AGI system shouldn't be that difficult to, uh, to coax. It should be much more straightforward, you know, just like talking to another human.

   12. AKAlex Kantrowitz5:16

       Yeah. And then on the reasoning front, you said that's another thing that's missing. I mean, that's, everybody's talking about reasoning right now.

   13. DHDemis Hassabis5:20

       Yeah.

   14. AKAlex Kantrowitz5:21

       So how does that end up getting us closer to artificial general intelligence?

   15. DHDemis Hassabis5:25

       Right. So reasoning and, and mathematics and other things, and there's a lot of progress on maths and coding and so on. But let's take maths for example. You have systems, uh, uh, some systems that we work on like AlphaProve, AlphaGeometry, that are getting, you know, silver medals in maths Olympiads, which is fantastic. But on the other hand, some of our systems, those same systems, are still making some fairly basic mathematical errors, right? Um, for, for various reasons. Um, like the classic, you know, counting the number of Rs in strawberries and so on, and, and, and the word strawberry and

3. 5:53 – 8:55

   Mathematics and Reasoning in AI

   1. DHDemis Hassabis5:53

      so on. And, and, um, is 9.11 bigger than 9.9? Uh, and so, and things like that. And, and, and of course you can fix those things, and we are, and everyone's improving on those systems. But we shouldn't really be seeing those kinds of flaws in a system that is that capable in other domains, in more narrow domains of doing, you know, Olympiad level mathematics. So there's something still a little bit missing, in my opinion, about the robustness of, uh, these systems. And then that's, I think that speaks to the generality of these systems. A truly general system would not have those sorts of weaknesses. It would be very, very strong, maybe even better than the best humans in some things, like playing Go or doing mathematics, but it, it would be overall consistently good.

   2. AKAlex Kantrowitz6:35

      Now, can you talk a little bit about how these systems are attacking math problems? Because-

   3. DHDemis Hassabis6:38

      Yeah

   4. AKAlex Kantrowitz6:38

      ... you know, I think the general understanding of these systems is, the LLMs, is they encompass all the world's knowledge, and then they predict what-

   5. DHDemis Hassabis6:45

      Yeah

   6. AKAlex Kantrowitz6:45

      ... you know, somebody might answer if they-

   7. DHDemis Hassabis6:46

      Yeah

   8. AKAlex Kantrowitz6:46

      ... were asked a question. But it's kind of different, different when you're working step by step through-

   9. DHDemis Hassabis6:50

      Yes

   10. AKAlex Kantrowitz6:50

       ... an algorithm.

   11. DHDemis Hassabis6:51

       Yeah.

   12. AKAlex Kantrowitz6:51

       Or through a math problem.

   13. DHDemis Hassabis6:52

       Yes. That's not enough, of course. Like, you know, just understanding the world's information and, and then trying to sort of almost compress that into your memory, that's not enough for solving a, a novel maths problem or novel, novel conjecture. Um, so there, you know, we start needing to bring in, I think we talked about this last time, more kind of like AlphaGo planning ideas-

   14. AKAlex Kantrowitz7:11

       Mm

   15. DHDemis Hassabis7:11

       ... into the mix with these large foundation models, which are now beyond just language. They're multimodal, of course. Um, and there, the, what, what you need to do is you need to have, um, your system, uh, uh, uh, uh, not just pattern matching roughly what it's seeing, which is the model, but also planning and being able to kind of go over, uh, uh, uh, that plan. We, you know, we, we, we, we, revisit that branch and then go into a different direction until you find the right, uh, uh, criteria or the right match to the criteria that you're looking for. And that's very much like the, the kind of games playing AI agents that we used to build for Go, chess, and so on. They had those, um, aspects, and I think we've got to bring them back in, but now working in a more general way on these general models, not just a narrow domain like games. Um, and I think that also, that approach of a model guiding a search or planning process so it's efficient, uh, works very well with mathematics as well. I mean, you can sort of turn maths into a kind of game-like search.

   16. AKAlex Kantrowitz8:08

       Right. And I want to ask about math. Like-

   17. DHDemis Hassabis8:10

       Mm

   18. AKAlex Kantrowitz8:10

       ... once these models get math right, is that generalizable? Because I think there was, like, a whole hubbub, uh-

   19. DHDemis Hassabis8:16

       Yeah

   20. AKAlex Kantrowitz8:16

       ... when people first learned about reasoning systems, and they're like, "Oh, this is, like, this is gonna be a problem."

   21. DHDemis Hassabis8:21

       Yeah.

   22. AKAlex Kantrowitz8:21

       "These, these models are getting smarter than we can control."

   23. DHDemis Hassabis8:23

       Mm.

   24. AKAlex Kantrowitz8:23

       Because if they can do math, then they can do-

   25. DHDemis Hassabis8:25

       Mm-hmm

   26. AKAlex Kantrowitz8:25

       ... X, Y, and Z.

   27. DHDemis Hassabis8:26

       Mm-hmm.

   28. AKAlex Kantrowitz8:26

       So is that generalizable, or is it, like, we're gonna teach them how to do math, and they can just do math?

   29. DHDemis Hassabis8:30

       I think for now, uh, uh, the jury's out on that. I mean, I feel like it's a clearly a c- a capability you want of a general AGI system. Uh, it can be very powerful in itself. Obviously, mathematics is, is extremely general in itself. Um, but it's not clear, you know, maths and even coding and games, these are areas, they're quite special, uh, uh, uh, areas of, of knowledge because you can verify if the answer

4. 8:55 – 12:08

   Building World Models for AI

   1. DHDemis Hassabis8:55

      is correct, right? In all of those, uh, domains, right? The math, you know, the final answer the AI system puts out, you can check whether that maths, uh, that solves the, the, the, the conjecture or the problem. So but most things in, in the general world, which is messy-

   2. AKAlex Kantrowitz9:10

      Mm-hmm

   3. DHDemis Hassabis9:10

      ... and ill-defined, um, do not have easy ways to verify whether you've done something correct. So that, that puts a limit on these self-improving systems if they want to go beyond these areas of high, you know, maybe very highly defined spaces like mathematics, coding, or, or games.

   4. AKAlex Kantrowitz9:26

      So how are you trying to solve that problem?

   5. DHDemis Hassabis9:28

      Well, you, you, you know, you've got to first of all, you've got to build, um, general models, mo- world models, we call them, uh, to understand the world around you, the physics of the world, um, the dynamics of the world, the spa- spatial temporal dynamics of the world, and so on, and the structure of, of the real world we live in. And of course, um, you need that for a universal assistant. So Project Astra is our project built on Gemini to do that, to understand, you know, objects and the, and the, and the context around us. I think that's important if you want to have an assistant. But also robotics requires that too. Of course, robots are physically embodied AIs, and they need to understand their environment, the, the physical environment, the, the physics of the world. So we're building those, uh, types of models. Um, and also you can, you can also use them in simulation to understand game environments, so that's another way to bootstrap more data for, to, to understand, uh, you know, the, the physics of a world. Um, but the issue at the moment is that those models are not 100% accurate, right? So they, you know, maybe they're accurate 90% of the time, or even 99% of the time. Um, but the problem is, if you start using those models to plan, maybe you're planning 100 steps in the future with that model. Even if you only have a 1% error in what the model's telling you, that's gonna compound over 100 steps to the point where you'll be in a, i- you know, you'll kind of get almost a random answer.

   6. AKAlex Kantrowitz10:46

      Mm-hmm.

   7. DHDemis Hassabis10:46

      Uh, uh, and so that makes the planning very difficult. Whereas with maths, with gaming, with code coding, you can verify each step. Are you still grounded, uh, to reality, um, and is the final answer, uh, mapped to what you're expecting? And, and so, um, I think part of the answer is to, is to make the, the world models more and more sophisticated, uh, and more and more accurate and, and, um, and not hallucinate and all of those kinds of things. So you get, you know, you, the errors are, are really minimal. Another approach is to, um- Um, plan not at each sort of a, a, a linear time step, but actually do what's called hierarchical planning. Another thing we used to- we've done a lot of research on in the past, and I think it's gonna come back into vogue, where you plan at different levels of temporal abstraction. So instead of... That, that could, that could also alleviate the need for your model to be super, super accurate because you're not planning over hundreds of time steps, you're planning over only a handful of time steps, but at different levels of abstraction.

   8. AKAlex Kantrowitz11:41

      How do you build a world model? Because, you know, I, I always thought it was gonna be like, uh, send robots out into the world-

   9. DHDemis Hassabis11:46

      Yeah

   10. AKAlex Kantrowitz11:46

       ... and have them figure out how the world works. But one thing that surprised me is with these video generation tools.

   11. DHDemis Hassabis11:51

       Yes.

   12. AKAlex Kantrowitz11:51

       You would think that if the AI didn't have a good world model, then nothing would really fit together when they try to figure out how the world works-

   13. DHDemis Hassabis11:58

       Yeah

   14. AKAlex Kantrowitz11:58

       ... as they show you these videos, like Veo 2, for instance. But they actually get the physics pretty right.

   15. DHDemis Hassabis12:03

       Yeah.

   16. AKAlex Kantrowitz12:03

       So can you get a world model just by showing an AI a video? Do you have to be out in the world?

   17. DHDemis Hassabis12:07

       Yeah.

5. 12:08 – 14:45

   The Role of Active Learning in AI

   1. AKAlex Kantrowitz12:08

      How's this gonna work?

   2. DHDemis Hassabis12:08

      It's interesting and actually been pretty surprising, I think, to the extent of how far these models can go without being out in the world, right, as you say. So Veo 2, our latest video model, which is actually in- surprisingly, uh, uh, accurate on things like physics. You know, uh, there's this, uh, this great demo that someone created of like, uh, chopping a tomato with a knife, right?

   3. AKAlex Kantrowitz12:29

      Mm-hmm.

   4. DHDemis Hassabis12:29

      And, and, and getting the slices of the tomato just right, and the fingers and all of that. And, um, Veo's the first model that can do that. You know, if you look at other competing models, they often, the tomato sort of randomly comes back together or, or the-

   5. AKAlex Kantrowitz12:41

      The person has 17 fingers

   6. DHDemis Hassabis12:42

      ... or the, the finger sort of-

   7. AKAlex Kantrowitz12:42

      Yeah

   8. DHDemis Hassabis12:42

      ... yeah, exactly, splits from the knife. Um, so those things are, if you think about it really hard, you've got to understand consistency across frames, all of these things. And it turns out that y- you know, you can do that by using enough data and, and viewing that. Um, I think these systems will get even better if they're supplemented by some real world data, like collected by an acting robot, or even potentially in very realistic simulations where you have avatars that, um, act in the world too.

   9. AKAlex Kantrowitz13:10

      Mm-hmm.

   10. DHDemis Hassabis13:10

       So I think that's the next big step, actually, for agent-based systems, is to go beyond world models. Can you collect enough data where the agents are also acting in the world and making plans and achieving tasks? Um, and, uh, I think for that you will need, uh, uh, not just passive observation, you will need actions, a- active, uh, participation.

   11. AKAlex Kantrowitz13:31

       I think you just answered my next question-

   12. DHDemis Hassabis13:32

       Mm

   13. AKAlex Kantrowitz13:32

       ... which is if I, if you develop AI that can reasonably plan and have, and reason about the world-

   14. DHDemis Hassabis13:38

       Mm

   15. AKAlex Kantrowitz13:38

       ... and has a model of how the world works, it can.

   16. DHDemis Hassabis13:41

       Mm.

   17. AKAlex Kantrowitz13:41

       And it seems like that's the answer. It can be an agent that can go out and do things for you.

   18. DHDemis Hassabis13:44

       Yes, exactly. And I think that's, that's, that's what will unlock robotics.

   19. AKAlex Kantrowitz13:48

       Right.

   20. DHDemis Hassabis13:48

       I think that's also what will then allow, uh, this notion of a universal assistant that can help you in your daily life across both the digital world and the real world. Um, that's what, that's the, that's the thing we're missing. Um, and I think that's gonna be incredibly powerful and useful, uh, tool.

   21. AKAlex Kantrowitz14:04

       You can't get there then by just scaling up the current models and building, you know, hundreds of thousand or million GPU clusters like Elon's doing right now, and that's not gonna be the path to AGI.

   22. DHDemis Hassabis14:15

       Um, well, look, I actually think it... So my view is a bit more nuanced than that, is like that, that, the scaling approach is absolutely working. Of course, that's where we've, why we've got to where we have now. Um, one can argue about are we getting diminishing returns or we S curve-

   23. AKAlex Kantrowitz14:28

       What do you think about that question?

   24. DHDemis Hassabis14:29

       ... or sigmoid. Well, well, my view is that we are getting substantial returns, but not, but it's slowing.

   25. AKAlex Kantrowitz14:35

       Okay.

   26. DHDemis Hassabis14:35

       So vis-a-vis, but, but it would, would have to. I mean, it's, it's not just continuing to be exponential, but that doesn't mean the scaling's not working. It's absolutely working and we're still getting, you know,

6. 14:45 – 17:52

   Creativity and Invention in AI

   1. DHDemis Hassabis14:45

      as you see Gemini 2 over Gemini 1.5. And by the way, the other thing that was working with the scaling is also making efficiency gains on the smaller sized models. So the, the cost or the size per performance is, is, is, is radically improving under the hood as well, which is very important for, for scaling, you know, the adoption of these systems. Um, but yeah, so, so, you know, you've got, you've got the scaling part, um, and that's absolutely needed to build more and more sophisticated world models. Um, but then I think we are missing, or, or we need to reintroduce some ideas on the planning side, uh, memory side, the searching side, uh, uh, uh, the reasoning to build on top of the model. The model itself is not enough to be an AGI. You need, uh, uh, this other capability for it to, to act in the world and solve problems for you. And, a- and then there's still the additional question mark of the, of the invention piece and the creativity piece.

   2. AKAlex Kantrowitz15:38

      Mm-hmm.

   3. DHDemis Hassabis15:38

      True creativity be- you know, beyond, uh, mashing together, uh, what's already known, right? So, uh, and that's, uh, also unknown yet. If, if something new is required or, again, if existing techniques will eventually scale to that. I can see both arguments, and I think from my perspective it's an empirical question. We just gotta push both the scaling and the invention part to the limit. And, and fortunately at, at Google DeepMind, we have, you know, a big enough group, we, we can invest in both those things.

   4. AKAlex Kantrowitz16:06

      So Sam Altman recently said, uh, something that caught people's eye. He said, "We are now confident we know how to build AGI as we have tradit- traditionally understood it." It just seems by listening to what you're saying, that you feel the same way.

   5. DHDemis Hassabis16:18

      Well, it depends what we n- you know, I think the way you said that was quite ambiguous, right? So in the sense of like, "Oh, we're building it right now and here's the ABC to do it." What I would say, and if this what he was meaning, I would agree with it, is that we, we roughly know the zones of techniques that are required, what's probably missing, which bits need to be put together. But, um, that's still a credible amount of research, in my opinion, that needs to be done to get that all to work, even if that was the case. And that's, and I think there's a 50% chance we are, uh, missing some new techniques.

   6. AKAlex Kantrowitz16:49

      Hmm.

   7. DHDemis Hassabis16:49

      You know, maybe we need one or two more transformer-like breakthroughs. And I, and I'm, I think I'm genuinely uncertain about that, so that's why I say 50%. So I mean, I wouldn't be surprised either way, if we got there with existing techniques and things we already knew, but put them together in the right way and scaled that up, or if it turned out one or two things were missing.

   8. AKAlex Kantrowitz17:08

      So let's talk about creativity for a moment.

   9. DHDemis Hassabis17:09

      Mm.

   10. AKAlex Kantrowitz17:09

       I mean, you brought it up a couple times here-

   11. DHDemis Hassabis17:11

       Yes

   12. AKAlex Kantrowitz17:11

       ... that the models are gonna have to be creative. They're gonna have to learn how to invent.

   13. DHDemis Hassabis17:15

       If we wanna call it AGI, in my opinion.

   14. AKAlex Kantrowitz17:16

       For, which is where everybody's trying to go.

   15. DHDemis Hassabis17:19

       Yes.

   16. AKAlex Kantrowitz17:19

       I was rewatching the AlphaGo documentary.

   17. DHDemis Hassabis17:22

       Oh, yeah.

   18. AKAlex Kantrowitz17:22

       And the algorithms make a-

   19. DHDemis Hassabis17:25

       Yes

   20. AKAlex Kantrowitz17:25

       ... creative move.

   21. DHDemis Hassabis17:26

       They do.

   22. AKAlex Kantrowitz17:26

       Uh, move...

   23. DHDemis Hassabis17:28

       37.

   24. AKAlex Kantrowitz17:28

       37.

   25. DHDemis Hassabis17:29

       Yes.

   26. AKAlex Kantrowitz17:29

       I just had it. Okay.

   27. DHDemis Hassabis17:30

       Yes.

   28. AKAlex Kantrowitz17:30

       Thank you. Uh, that's interesting, 'cause it was a couple years ago. The w- the algorithms were already being creative.

   29. DHDemis Hassabis17:36

       Yes.

   30. AKAlex Kantrowitz17:36

       Why have we not really seen creativity from large language models? I mean, this is, to me, I think the greatest disappointment that people have-
7. 17:52 – 20:55

   Understanding Human Ingenuity through AI

   1. DHDemis Hassabis17:52

      Yeah. Well, look, so what... And I should probably write this up, but what I sometimes talk about in talks ever since the AlphaGo match, which is now, you know, eight plus years ago, amazingly, right? That happened.

   2. AKAlex Kantrowitz18:01

      Yeah.

   3. DHDemis Hassabis18:01

      That was probably... The reason that was such a watershed moment for AI was, uh, first of all, there was the Everest of, of, of, of you know, um, uh, cracking Go, right? Which was always considered to be one of the holy grails of AI. Um, so we did that. Um, second thing was the way we did it, which was these learning systems that were generalizable, right? Eventually became AlphaZero, and, and so on, even when play any two-player game and so on. Uh, and then the third thing was this move 37. So not only did it win 4-1, it beat Lee Sedol, the great Lee Sedol, 4-1, it also played original moves. But, so I, I have three categories of, of, of, of originality or creativity. The most basic kind of mundane form is just interpolation, which is like averaging of what you see. So if I said to a system, you know, "Come up with a new picture of a cat," and it's seen a million cats, and it produces s- some kind of average of all the ones it's seen, in theory, that's an original cat because it-- you won't find the average in the, the specific examples, but it's a pretty boring ex- You know, it's not really very creative. I wouldn't call that creativity. That's the lowest level. Next level is what AlphaGo exhibited, which is extrapolation. So here's all the games humans have ever played. It's played another million games on top of-- you know, 10 million games on top of that. Um, and now it comes up with a new strategy in Go that, um, no human has ever seen before. That's move 37, right?

   4. AKAlex Kantrowitz19:16

      Mm-hmm.

   5. DHDemis Hassabis19:16

      Revolutionizing Go, even though we've played it for thousands of years. So that's pretty incredible, and that could be very useful in science, and that's why I got very excited about that and started doing things like AlphaFold, 'cause clearly extrapolation beyond what we already know was in the training set, um, could be extremely useful. So that's already very valuable and, and I think truly creative. But there's one level above that, that humans can do, which is invent Go. Can you invent me a game if I g- that, you know, if I specify it to an abstract level, you know, takes five minutes to learn the rules, uh, but a lifetime-

   6. AKAlex Kantrowitz19:47

      Mm

   7. DHDemis Hassabis19:47

      ... to many lifetimes to master. It's beautiful aesthetically, encompasses some sort of mystical part of the universe in it that, the, the, the... It's beautiful to look at, uh, it... But you can play a game in a human afternoon in two hours, right? That's the, that's-- That would be a high-level specification of Go. And then somehow the system's gotta come up with a game that's as elegant and as beautiful and, and perfect as Go. Now, we can't do that. Now, the, the question is, why? Is it that we don't know how to specify that type of goal to our systems at the moment? Um, what's the objective function? It's very amorphous. It's very abstract. So I'm not sure if it's just we need higher level, uh, more abstracted, uh, uh, layers in our systems, um, building more and more abstract models so we can talk to it in this way, give it those kind of amorphous goals. Or is there a missing capability actually about that, that we still have, human intelligence has, that are still missing from our systems? And again, I'm unsure about that, which, which way that is. I can see arguments both ways, and we'll try both.

   8. AKAlex Kantrowitz20:48

      But I think the thing that people are upset a- or, or not upset-

   9. DHDemis Hassabis20:51

      Yeah

   10. AKAlex Kantrowitz20:51

       ... but people are disappointed by is they don't even see a move 37 in today's LLMs.

8. 20:55 – 29:03

   Deception in AI Systems

   1. DHDemis Hassabis20:55

      Right. Well, and that's because-

   2. AKAlex Kantrowitz20:55

      What's going on there?

   3. DHDemis Hassabis20:56

      Okay, so, well, that's because I don't think we have... So if you look at AlphaGo, and I'll give you an example of there, which, which maps to today's LLMs. So, um, you can run AlphaGo and AlphaZero, our chess program, general two-player program, without the search and the reasoning part on top. You can just run it with the model.

   4. AKAlex Kantrowitz21:14

      Mm.

   5. DHDemis Hassabis21:14

      So what you say is to the model, "Come up with the first Go move you can think of in this position that's most, the most pattern matched, most likely good move." Okay? And it can do that, and it'll play a reasonable game. But it will only be around, uh, master level or p- possibly grandmaster level. It won't be world champion level, and it certainly won't come up with, um, original moves. That, for that, I think you need, um, the search component to get you beyond where the model knows about, which is m- mostly summarizing existing knowledge to some new part of the tree of knowledge, right? So you can use the search to get beyond what the model currently understands, and that's where I think you can get, uh, new ideas like, you know, move 37.

   6. AKAlex Kantrowitz21:57

      What's it searching, the web?

   7. DHDemis Hassabis21:59

      No. So, well, it, it, it depends on, uh, uh, uh, uh, what the domain is. Searching that, that knowledge tree. Uh, so in, obviously in Go, it was searching Go moves beyond what the model kn- knew. Um, I think for language models, it will be searching the world model for new parts, configurations in the world, um, that are useful. So of course-

   8. AKAlex Kantrowitz22:18

      Fascinating

   9. DHDemis Hassabis22:18

      ... that's so much more complicated-

   10. AKAlex Kantrowitz22:19

       Yeah

   11. DHDemis Hassabis22:20

       ... which is why we haven't seen it yet. Um, but I think the agent-based systems that are coming will be capable of Move 37 type things.

   12. AKAlex Kantrowitz22:27

       So are we setting too high of a bar for AI? Because I'm curious if you've learned anything about humanity doing this work.

   13. DHDemis Hassabis22:32

       Yeah.

   14. AKAlex Kantrowitz22:32

       It seems like we almost give too much of a premium on humanity or pe- gen- individual people's ingenuity, where like a lot of us, like we're kinda taking stuff, we spit it out, like our society really works in memes.

   15. DHDemis Hassabis22:44

       Yeah.

   16. AKAlex Kantrowitz22:44

       Like we have a cultural thing and it gets translated. So, um, what do you, what have you learned about like the nature of humans from doing the work with the AIs?

   17. DHDemis Hassabis22:51

       Well, look, I, I think humans are incredible and, and, and especially the best humans in the best domains. I love watching any sports person or, or, or talented musician or games player at the top of their game, the, the abso pinnacle of human performance. It's always incredible no matter what it is.

   18. AKAlex Kantrowitz23:07

       Mm-hmm.

   19. DHDemis Hassabis23:07

       Um, so I think as a species we're amazing. Uh, individual- individually we're also kind of amazing what everyone can do with their brain so generally, right? Deal with new technologies. I mean, I'm always fascinated by how we just adapt to these things sort of almost effortlessly as a society and as individuals. Um, so that speaks to the power and the generality of our minds. Um, now th- the reason I have set the bar like that, and I don't think it's a question of like can we get economic worth out of these systems, I think that's already coming very soon, but that's not what... AGI shouldn't be, uh, uh, I think we should tr- treat AGI with scientific integrity, not just move goalposts for commercial reasons or whatever it is, hype and so on. And there, the, the, the definition of that was always having a system that was, you know, if we think about it theoretically, that was capable of being as powerful as a Turing machine. So Alan Turing, one of my all-time scientific heroes, you know, he described a Turing machine which underpins all modern computing, right? Uh, a- as a system that can simulate any other comp- c- can compute anything that's computable. So we know and we have the theory there that if an AI system is Turing powerful, it's called, if it can simulate a Turing machine, then it's able to calculate anything in theory that is calc- is, is computable. And the human brain is probably some sort of Turing machine, at least that's what I believe. Uh, and so, um, in order for our, to know our... And, and that, I think that's what AGI is, is a system that's truly general and in theory could be applied to anything. And, and the only way we'll know that is if we, um, it exhibits all the cog- cognitive capabilities that humans have, assuming that human, the human mind is a type of Turing machine, or is at least as powerful as a Turing machine. So that's my, always been my sort of bar. Um, it seems like people are trying to rebadge things as that as being what's called ASI, artificial super intelligence, but I think that's beyond that. That's after you have that system and then it starts going beyond, uh, in certain domains, what humans are capable of, uh, potentially inventing themselves.

   20. AKAlex Kantrowitz25:05

       Okay, so when I see everybody making the same joke on the same topic on Twitter, it's-

   21. DHDemis Hassabis25:09

       Yeah

   22. AKAlex Kantrowitz25:09

       ... and I say, "Oh, that's just us being LLMs-"

   23. DHDemis Hassabis25:12

       Yeah

   24. AKAlex Kantrowitz25:12

       ... I think I'm selling humanity a little short.

   25. DHDemis Hassabis25:14

       Uh, well, we'll, we'll, we'll, we'll s-

   26. AKAlex Kantrowitz25:16

       [laughs]

   27. DHDemis Hassabis25:16

       Yes. I guess so. I guess so.

   28. AKAlex Kantrowitz25:17

       Okay.

   29. DHDemis Hassabis25:18

       Yeah.

   30. AKAlex Kantrowitz25:18

       I wanna ask you about deceptiveness.
9. 29:03 – 30:12

   The Dual Nature of AI: Opportunities and Risks

   1. AKAlex Kantrowitz29:03

      oh.

   2. DHDemis Hassabis29:03

      Well, look, of course, you know, these are... it's impressive capabilities and, and, and, and the, the, the, the, you know, the, the negatives are things like deception, but the positives would be things like inventing, you know, new materials, accelerating science. You need that kind of a- a- ability to problem solve-

   3. AKAlex Kantrowitz29:19

      Mm

   4. DHDemis Hassabis29:19

      ... and get around, you know, uh, issues that are, are blocking progress. Um, but of course you want that only in the positive direction, right? So those are exactly the kinds of capabilities. I mean, they are v- you know, uh, it's kind of mind-blowing we're talking about those, those possibilities. But also at the same time, uh, there's risk and it's scary. So I think both the things are true.

   5. AKAlex Kantrowitz29:40

      Wild.

   6. DHDemis Hassabis29:41

      Yeah.

   7. AKAlex Kantrowitz29:41

      All right, let's talk about product quickly.

   8. DHDemis Hassabis29:42

      Sure.

   9. AKAlex Kantrowitz29:43

      Um, one of the things that your colleagues have told me about you is you're very good at scenario planning what's gonna happen in the future.

   10. DHDemis Hassabis29:49

       Yeah.

   11. AKAlex Kantrowitz29:49

       It's sort of an exercise that happens-

   12. DHDemis Hassabis29:50

       Mm

   13. AKAlex Kantrowitz29:50

       ... within DeepMind. Uh, what do you think is gonna happen with the web? Because obviously the web is so important to Google.

   14. DHDemis Hassabis29:56

       Yeah.

   15. AKAlex Kantrowitz29:56

       I had an editor that told me, he's like, "Oh, you're gonna speak with Demis. Ask him what happens when we stop clicking." Right?

   16. DHDemis Hassabis30:01

       Yeah.

   17. AKAlex Kantrowitz30:01

       We're clicking through the web at all times.

   18. DHDemis Hassabis30:02

       Yeah.

   19. AKAlex Kantrowitz30:02

       The, there are a rich, uh, corpus of websites that we use. Um, if we're all just dialoguing with AI, then maybe we don't click anymore. So what do you, what is your scenario plan for what happens to

10. 30:12 – 32:35

    The Future of the Web: Agent-Based Interactions

    1. AKAlex Kantrowitz30:12

       the web?

    2. DHDemis Hassabis30:12

       Well, look, I think there's, um, it's gonna be, there's gonna be a very interesting phase in the next few years on the web and, and the way we, we interact with websites and apps and so on. Um, you know, if everything becomes more agent-based, then I think we're gonna want our assistants and our agents to do a lot of the work and a lot of the mundane work, um, that we currently do, right? Um, you know, fill in forms, make payments, uh, a, you know, book tables, this kind of thing. So, you know, I think that we're gonna end up with, uh, probably uh, a sit- a, a kind of economics model where agents talk to other agents and negotiate things between themselves and then give you back the results, right? And you'll have the service providers with agents as well that are offering services and maybe there's some, uh, uh, uh, bidding and cost and things like that involved, uh, and efficiency. And then I hope from a user perspective, you know, you have this assistant that's super capable that you can just like a, a brilliant, uh, uh, uh, a human assistant, personal assistant, and can take care of a lot of the mundane things for you. And I think if you follow that through, that does imply a lot of changes to, uh, the structure of, of the web and the way we currently use the web.

    3. AKAlex Kantrowitz31:22

       It's a lot of middlemen.

    4. DHDemis Hassabis31:23

       Yeah, sure. But there'll be many other... I think there'll be incredible other opportunities that will appear, economic and otherwise, uh, based on this, this change, but I, I think it's gonna be a big disruption. Yeah.

    5. AKAlex Kantrowitz31:34

       And what about information?

    6. DHDemis Hassabis31:37

       Well, uh, I, I mean, finding information, I think, uh, you'll still need the reliable sources. I think you'll have assistants that are, um, able to synthesize and, and, and, and help you kind of understand that information. Um, I think education's gonna be revolutionized by AI. Um, so, uh, a- again, I, I hope that, uh, uh, these assistants will, will be able to, uh, more efficiently gather information for you and perhaps, you know, what I dream of is, again, assistants that take care of a lot of the mundane things, perhaps replying to, you know, everyday emails and other things so that you have, you protect your own mind and brain space from this bombardment we're getting today from social media and emails and so on, and texts and so on. So it actually blocks deep work and, and being in flow and things like that, which I, I value very much. So I would quite like these assistants to take away, uh, a, a lot of the, uh, the mundane aspects of, of admin that we do every day.

11. 32:35 – 35:42

    AI Assistants: Transforming Personal and Work Life

    1. AKAlex Kantrowitz32:35

       What's your best guess to, as to what type of relationships we're gonna have with our AI agents or AI assistants? So there's on one hand, you could have a dispassionate agent that's just, like, really good at getting stuff done for you.

    2. DHDemis Hassabis32:45

       Yeah.

    3. AKAlex Kantrowitz32:45

       On the other hand, like, it's already clear that people are, like, falling in love with these bots.

    4. DHDemis Hassabis32:48

       Yeah.

    5. AKAlex Kantrowitz32:49

       There was a New York Times article last week about someone who's fallen in love-

    6. DHDemis Hassabis32:52

       Right

    7. AKAlex Kantrowitz32:52

       ... with ChatGPT, like, for real fallen in love.

    8. DHDemis Hassabis32:54

       Okay. [laughs]

    9. AKAlex Kantrowitz32:54

       And I had, uh, the CEO of Replika on the show-

    10. DHDemis Hassabis32:57

        Yeah

    11. AKAlex Kantrowitz32:57

        ... a couple weeks ago.

    12. DHDemis Hassabis32:57

        Yeah.

    13. AKAlex Kantrowitz32:58

        And she said that they are regularly invited to marriages, uh-

    14. DHDemis Hassabis33:01

        Wow

    15. AKAlex Kantrowitz33:01

        ... of people who are marrying their replikas.

    16. DHDemis Hassabis33:03

        Okay.

    17. AKAlex Kantrowitz33:03

        And they're moving into this more assistive-

    18. DHDemis Hassabis33:05

        Yeah

    19. AKAlex Kantrowitz33:05

        ... space. So do you think that when we, when we start interacting with something-

    20. DHDemis Hassabis33:09

        Mm

    21. AKAlex Kantrowitz33:09

        ... that knows us so well-

    22. DHDemis Hassabis33:10

        Mm

    23. AKAlex Kantrowitz33:10

        ... uh, that helps us with everything we need-

    24. DHDemis Hassabis33:13

        Yeah

    25. AKAlex Kantrowitz33:13

        ... is it gonna be, like, a third type of relationship where it's not necessarily a friend, not a lover, but it's gonna be a deep relationship, don't you think?

    26. DHDemis Hassabis33:19

        Yeah, it's gonna be really interesting. I think the way I'm modeling that, first of all, is, uh, two, at, at least two domains, first of all, which is your, your personal life and then your work life, right? So I think you'll have this notion of virtual workers or something. Maybe we'll have a set of them or managed by a, a, a m- you know, a, a, a, a lead assistant that does a lot of the, uh, uh... helps us be way more productive at work, you know, or, or whether that's email, across workspace or whatever that is. So we're really thinking about that. Then there's the personal side where, you know, we're talking about earlier about all these, um, uh, booking holidays for you, arranging things, mundane things for you, sorting things out. Um, and then, uh, that makes your life more efficient. I think it can also enrich your life, so recommend you things that, amazing things, that it knows you as well as you know yourself. Um, so those two I think are definitely gonna happen. And then I think there is a, a, a philosophical discussion to be had about is there a third space where these things start becoming so integral to your life they become more like companions. I think that's possible too. We've seen that a little bit in gaming. So you may have seen we had the little prototypes of Astro working in and Gemini working with, like, being almost a game companion, commenting on you, almost like as if you had a friend looking at a game you're playing and recommending things to you and advising you, but also maybe just playing along with you, and it's, it's, it's very fun. Um, so I, I, I haven't, you know, quite through, thought through all the implications of that, but they're gonna be big. Uh, and I'm sure there is gonna be demand for companionship and other things. Maybe the good side of that is it'll help with loneliness and these sorts of things. But there's also, you know, uh, I think it's gonna be, uh, uh, and it's gonna have to be really carefully thought through by society whether, uh, you know, what directions we wanna take that in.

    27. AKAlex Kantrowitz35:01

        I mean, my personal opinion is that, that it's the most underappreciated part of AI right now.

    28. DHDemis Hassabis35:05

        Mm.

    29. AKAlex Kantrowitz35:05

        And that- People are just gonna form such deep relationships with these bots as they get better 'cause, like, I know it's a meme in AI that this is the worst it's ever gonna be.

    30. DHDemis Hassabis35:14

        Yeah.
12. 35:42 – 39:21

    Project Astra: The Next Generation of AI Assistants

    1. AKAlex Kantrowitz35:42

       It is almost an always-on AI assistant.

    2. DHDemis Hassabis35:44

       Mm-hmm.

    3. AKAlex Kantrowitz35:44

       You can, like, hold your phone.

    4. DHDemis Hassabis35:46

       Yeah.

    5. AKAlex Kantrowitz35:46

       It's, it's currently just a prototype-

    6. DHDemis Hassabis35:47

       Yes

    7. AKAlex Kantrowitz35:48

       ... or not publicly released.

    8. DHDemis Hassabis35:49

       Yes.

    9. AKAlex Kantrowitz35:49

       But you can hold your phone, and it will see what's going on in the room.

    10. DHDemis Hassabis35:51

        Yeah.

    11. AKAlex Kantrowitz35:51

        So if-- I could basically-- I've seen you do this on your show.

    12. DHDemis Hassabis35:54

        Yeah.

    13. AKAlex Kantrowitz35:54

        Or not you personally, but somebody-

    14. DHDemis Hassabis35:55

        Yeah

    15. AKAlex Kantrowitz35:55

        ... on your team. Uh, you could say, "Okay, where am I?" And it'll be like, "Oh, you're in-"

    16. DHDemis Hassabis35:58

        Yeah

    17. AKAlex Kantrowitz35:58

        ... "a podcast studio."

    18. DHDemis Hassabis35:59

        Yes.

    19. AKAlex Kantrowitz35:59

        You know, anything. Okay, so it could have this contextual awareness.

    20. DHDemis Hassabis36:02

        Yes.

    21. AKAlex Kantrowitz36:03

        Uh, can that work without smart glasses? Because it's really annoying to hold my phone up.

    22. DHDemis Hassabis36:07

        Yes.

    23. AKAlex Kantrowitz36:07

        So, like, when are, when are we gonna see Google smart glasses-

    24. DHDemis Hassabis36:10

        Yeah

    25. AKAlex Kantrowitz36:10

        ... with this technology embedded?

    26. DHDemis Hassabis36:11

        It, they're, they're coming. So we, we teased it in some of our early prototypes, so that we're mostly prototyping on, on, on phones currently 'cause they have more processing power. But we're-- of course, Google's always been a leader in, in glasses.

    27. AKAlex Kantrowitz36:23

        Google Glass.

    28. DHDemis Hassabis36:23

        Yeah, and exactly.

    29. AKAlex Kantrowitz36:24

        Just a little too early.

    30. DHDemis Hassabis36:25

        Yeah, maybe a little too early. And now I actually think, and we're, they're super excited, that team, is that, you know, maybe this assistant is the killer use case that glasses has always been looking for. And I think it's quite obvious when you, when you start using Astra in your daily life, which we have in with trusted testers at the moment and in kind of beta form, um, there are many use cases where it would be so useful to use it, but it's a bit-- it's, it's inconvenient that you're holding the phone. So one example is while you're cooking, for example, right?
13. 39:21 – 43:25

    Decoding Biology: The Virtual Cell Project

    1. AKAlex Kantrowitz39:21

       All right, science. So, um, you worked on basically, uh, decoding all protein folding with AlphaFold.

    2. DHDemis Hassabis39:27

       Yes.

    3. AKAlex Kantrowitz39:28

       You won the Nobel Prize for that. Not to skip over the thing that you won the Nobel Prize for-

    4. DHDemis Hassabis39:32

       Yes

    5. AKAlex Kantrowitz39:32

       ... but I wanna talk about what's on the roadmap-

    6. DHDemis Hassabis39:33

       Sure

    7. AKAlex Kantrowitz39:34

       ... which is that you have, um, an interest in mapping a virtual cell.

    8. DHDemis Hassabis39:38

       Yes.

    9. AKAlex Kantrowitz39:38

       Uh, what is that?

    10. DHDemis Hassabis39:40

        Yes.

    11. AKAlex Kantrowitz39:40

        And what does it get us?

    12. DHDemis Hassabis39:41

        Yeah. Well, so if you think about what we did with AlphaFold was essentially solve the problem of the, the, the stru- the, the finding the structure of a protein. Um, proteins, everything in life depends on proteins, right? Everything in your body. Um, so that's the kind of static picture of a protein. But the thing about biology is really it's you only understand what's going on in biology if you understand the dynamics and the interactions between the different things in, in a cell. And so a virtual cell project is about building a simulation, an AI simulation of a full working cell. I'd probably start with something like a yeast cell because of, uh, the simplicity of the, of the yeast organism. And, um, and you have to build up there. So the next step is with AlphaFold three, for example, we started doing pairwise interactions between proteins and ligands and proteins and DNA proteins and RNA. And then the next step would be modeling a whole pathway, maybe a cancer pathway or something like that that would be helpful with, for solving a disease, and then finally a whole cell. And the reason that's important is you would be able to hypothesis, make hypotheses and test those hypotheses about making some change, some nutrient change or injecting a drug into the cell and then s- seeing what happens to the, how the cell responds. Um, and at the moment, of course, you have to do that painstakingly in a wet lab. But imagine if you could do it a thousand, a million times faster in silico first, and only at the last step do you do a validation in the wet lab. So instead of doing the search in, in, in the wet lab, which is millions of times more expensive and time-consuming than the validation step, you just do the search part, um, in silico. So it's aga- it's sort of translating again what we did in the games environments, uh, but here in the sciences and the biology. So you, you build a model and then you use that to do the reasoning and the search over, and then the predictions are, you know, at least better than m- maybe they're not perfect, but they're useful enough to, um, to be useful for experimentalists to, to validate against.

    13. AKAlex Kantrowitz41:41

        And the wet lab is within people?

    14. DHDemis Hassabis41:44

        Yeah, so the wet lab, uh, you'd, you'd, you'd still need a final step with the, with the wet lab to prove the, uh, what the predictions were actually valid. So you know, you, but you wouldn't have to do all of the work to get to that prediction in, in, in the wet lab. So you'd just get, "Here's the prediction. If you put this chemical in, this should be the change," right? And then you'd just do that one experiment. So and then after that, of course, you still have to have clinical trials if you're talking about a drug. You would still need to test that properly through the clinical trials and so on, and test it on humans for efficacy and so on. That, I also think, could be improved with AI, that whole clinical trial.

    15. AKAlex Kantrowitz42:18

        Mm-hmm.

    16. DHDemis Hassabis42:18

        That also takes many, many years. But this, that would be a different technology from the virtual cell. The virtual cell would be p- would be helping the discovery phase for drug discovery.

    17. AKAlex Kantrowitz42:26

        Just like I have a idea for a drug, throw it in the virtual cell-

    18. DHDemis Hassabis42:29

        Cell

    19. AKAlex Kantrowitz42:29

        ... and then-

    20. DHDemis Hassabis42:29

        See what it does. Yeah, and maybe eventually it's a liver cell or, or a brain cell or something like that, so you have different cell models. And then, you know, at least 90% of the time, it give, it's giving you back what would really happen.

    21. AKAlex Kantrowitz42:40

        That'd be incredible.

    22. DHDemis Hassabis42:41

        Yeah.

    23. AKAlex Kantrowitz42:41

        How, how long do you think that's gonna take to figure out?

    24. DHDemis Hassabis42:43

        Um, I think that would be, like, um, maybe five years from now.

    25. AKAlex Kantrowitz42:46

        Okay.

    26. DHDemis Hassabis42:46

        Yeah, yeah. So I, I have a kind of five-year project, and a lot of the AlphaFold, the old AlphaFold team are working on that.

    27. AKAlex Kantrowitz42:52

        Yeah, I was asking your team here, so you've-

    28. DHDemis Hassabis42:54

        Yeah, John Jumper and-

    29. AKAlex Kantrowitz42:54

        ... you figured out, yeah.

    30. DHDemis Hassabis42:55

        Yeah.
14. 43:25 – 45:57

    Genomics and AI: Understanding Genetic Mutations

    1. AKAlex Kantrowitz43:25

       Yeah.

    2. DHDemis Hassabis43:25

       Yeah.

    3. AKAlex Kantrowitz43:25

       No, I've seen it a couple times in my family, and-

    4. DHDemis Hassabis43:27

       Yeah

    5. AKAlex Kantrowitz43:28

       ... if we can ensure that doesn't happen, it's just, it, it's a-

    6. DHDemis Hassabis43:30

       One of the best things we could use AI for-

    7. AKAlex Kantrowitz43:32

       Yeah, I agree

    8. DHDemis Hassabis43:32

       ... in my opinion.

    9. AKAlex Kantrowitz43:32

       Yeah.

    10. DHDemis Hassabis43:33

        Yeah.

    11. AKAlex Kantrowitz43:33

        Yeah, it's ter- a terrible way to see somebody-

    12. DHDemis Hassabis43:35

        Yeah

    13. AKAlex Kantrowitz43:35

        ... uh, decline.

    14. DHDemis Hassabis43:36

        Yeah.

    15. AKAlex Kantrowitz43:36

        So.

    16. DHDemis Hassabis43:37

        Yeah.

    17. AKAlex Kantrowitz43:37

        It's important work. Um, uh, in addition to that, there's the genome.

    18. DHDemis Hassabis43:41

        Yes.

    19. AKAlex Kantrowitz43:42

        And so the Human Genome Project sort of-

    20. DHDemis Hassabis43:43

        Yes

    21. AKAlex Kantrowitz43:44

        ... I was like, "Okay, so they decoded the whole genome. There's no more work to do there."

    22. DHDemis Hassabis43:47

        Yeah.

    23. AKAlex Kantrowitz43:47

        Like, just same way that you decoded proteins with Fold.

    24. DHDemis Hassabis43:50

        Yeah.

    25. AKAlex Kantrowitz43:50

        But it turns out that actually the, we just have, like, a bunch of letters-

    26. DHDemis Hassabis43:54

        Yeah

    27. AKAlex Kantrowitz43:54

        ... when it's decoded. And so now you're working to use AI to translate what those letters mean?

    28. DHDemis Hassabis43:58

        Yes. So yeah, we have lots of cool work, uh, on, on genomics and, um, uh, uh, uh, trying to figure out if mutations are going to be harmful or, or benign, right? Uh, most mutations to your DNA are, are, are harmless, uh, but of course some are pathogenic and you wanna know which ones they are. So our first, uh, systems, um, are, are, ha- have, are the best in the world at predicting that. Um, and then, um, uh, uh, the next step is to look at, uh, uh, sit- situations where the disease isn't caused just by one genetic mutation, but maybe a series of them in concert, and obviously that's a lot harder. Like, and a lot of more complex diseases that we haven't made progress with are probably not due to a single mutation, right? That's more, like, rare childhood diseases, things like that. Um, so there, you know, we need to... I think AI is the perfect tool, uh, to, to, to sort of, um, uh, uh, uh, try and figure out what these weak interactions, uh, are like, right? How they maybe, um, uh, uh, kind of compound on top of each other. Um, and so maybe the statistics are not very obvious, but a, an AI system that's able to kinda spot patterns would be able to figure out there is some connection here.

    29. AKAlex Kantrowitz45:12

        And so we talk about this a lot in terms of disease, but also I wonder what happens in terms of, uh, making people superhuman.

    30. DHDemis Hassabis45:18

        Mm.
15. 45:57 – 48:25

    The Future of Aging: Extending Healthy Lifespan

    1. DHDemis Hassabis45:57

       very interesting, and I'm, I'm pretty sure AI will be extremely useful in helping us find answers to those questions too.

    2. AKAlex Kantrowitz46:03

       You know, I see memes come across my Twitter feed, and maybe I need to change the stuff I'm recommended.

    3. DHDemis Hassabis46:07

       Yeah.

    4. AKAlex Kantrowitz46:07

       But it's often like if you will live to 2050, you're not gonna die.

    5. DHDemis Hassabis46:11

       Yeah.

    6. AKAlex Kantrowitz46:11

       Uh, what do you think the potential max lifespan is for a person?

    7. DHDemis Hassabis46:16

       Well, look, I know those, a lot of those folks in aging research very well. I think it's very interesting the pioneering work they, they do. Um, I think there's nothing good about go- getting old and your body decaying. I think it's, you know, if anyone who's seen that up close with their relatives, it's a pretty hard thing to go through, right, as a family or, or the, or the person of course. And, um, and, and so I think anything we can alleviate human suffering and, and extend healthy lifespan is a good thing. Um, you know, the natural limit seems to be about 120 years old, but, um, from what we know. You know, if you look at the oldest people that, that, that are ve- that, that are lucky enough to, to, to live to that age. So there's, you know, it's, it's, it's a, it's a, an area I follow quite closely. Uh, I don't have any, I g- I guess, new insights that, uh, are not already known in that. Um, but I do, uh, I, I would be surprised if there, if that's, if that's the limit, right? Because there's a sort of two steps to this. One is curing all diseases one day, which I think we're gonna do with Isomorphic and the work we're doing there, our, our spinout, our drug discovery spinout. Um, but then that's not enough to probably get you past 120 because there's some sort of, then there's the question of just natural systemic decay, right? Aging, in other words. So, uh, not specific disease, right? Uh, often those people that live to 120, they don't seem to die from a specific disease. It's just sort of just general atrophy. Um, so then you're gonna need something more like rejuvenation, where you, you, you rejuvenate your cells or you, you know, maybe s- stem cell research. You know, companies like Altos are, are, are working on these things, resetting the, the cell clocks. Um- Seems like that could be possible. But again, I feel like it's so complex 'cause biology's such a complicated and emergent system, you need a, in my view, you need AI to help to, to be able to crack any- anything, and anything close to that.

    8. AKAlex Kantrowitz48:04

       Very quickly on-

    9. DHDemis Hassabis48:05

       Yeah

    10. AKAlex Kantrowitz48:05

        ... material science. I don't wanna leave here without talking about the fact that you've discovered many new materials or potential materials.

    11. DHDemis Hassabis48:10

        Yeah.

    12. AKAlex Kantrowitz48:11

        Uh, the stat I have here is, known to humanity, uh, recently were 30,000 stable-

    13. DHDemis Hassabis48:17

        Yeah

    14. AKAlex Kantrowitz48:17

        ... uh, materials. You've discovered 2.2 million with a new AI program.

    15. DHDemis Hassabis48:24

        Yeah.

    16. AKAlex Kantrowitz48:24

        Uh,

16. 48:25 – 53:21

    Material Science: Discovering New Materials for the Future

    1. AKAlex Kantrowitz48:25

       just dream a little bit.

    2. DHDemis Hassabis48:27

       Yes.

    3. AKAlex Kantrowitz48:27

       And, 'cause we don't know what all those materials can do.

    4. DHDemis Hassabis48:29

       No.

    5. AKAlex Kantrowitz48:29

       We don't know what, you know-

    6. DHDemis Hassabis48:30

       Yes

    7. AKAlex Kantrowitz48:30

       ... whether they'll be able to handle being out of, like, a frozen box or whatever.

    8. DHDemis Hassabis48:35

       Yes.

    9. AKAlex Kantrowitz48:35

       Uh, dream materials-

    10. DHDemis Hassabis48:37

        Yeah

    11. AKAlex Kantrowitz48:37

        ... for you to find in that set of-

    12. DHDemis Hassabis48:39

        Yeah

    13. AKAlex Kantrowitz48:39

        ... new materials.

    14. DHDemis Hassabis48:40

        Well, I mean, we're working really hard on materials. To me, it's, like, the next, uh, one of the next sort of, uh, big, uh, impacts we can have, like, at the level of AlphaFold, really, in biology, but this time in chemistry and materials. You know, I dream of, uh, one day discovering a room temperature superconductor. That's-

    15. AKAlex Kantrowitz48:55

        So what will that do? Because that's another big meme that people talk about.

    16. DHDemis Hassabis48:58

        Yeah. Well, then-

    17. AKAlex Kantrowitz48:59

        So what do you... Yeah.

    18. DHDemis Hassabis49:00

        Well, it would help with the energy crisis and climate crisis because, um, if you had sort of cheap, uh, superconductors, you know, then you can transport energy from one place to another without any loss of that energy.

    19. AKAlex Kantrowitz49:11

        Mm-hmm.

    20. DHDemis Hassabis49:11

        Right? So you could potentially put solar panels in the Sahara Desert, and then just have a, the, the, the, the superconductor tr- you know, uh, funneling that into Europe where it's needed. At the moment, you would just lose a ton of the power to heat and other things on the way. So then you need other technologies, like batteries and other things, to store that 'cause you can't, you can't just pipe it to the place that you want, uh, without, without, without being incredibly inefficient. So, uh, but also materials could help with things like batteries too, like, to come up with the optimal battery. I don't think we have the optimal battery designs, um, that maybe we can do things like a combination of materials and, and, and proteins. We can do things like carbon capture, you know, modify, uh, uh, algae or other things to, to do carbon capture, uh, better than, um, uh, uh, our artificial systems. Um, I mean, even the, one of the most famous and most important chemical, chemical processes, the Haber process to make fertilizer and ammonia, you know, to take nitrogen out of the air, um, was, was, was something that allows modern civilization. Uh, but there might be many other, uh, chemical processes that could be catalyzed in that way if we knew what the right catalyst and the right material was. Um, so, uh, I think it's gonna be, will be one of the most impactful technologies ever is to, to, to basically have a in silico design of materials. So we've done step one of that where we showed we can come up with new stable materials, but we need a way of testing the properties of those materials.

    21. AKAlex Kantrowitz50:37

        Right.

    22. DHDemis Hassabis50:37

        'Cause no lab can test 200,000, you know, tens of thousands of materials or millions of materials at the moment. So we have to, that's, that's the hard part is to, is to do the testing.

    23. AKAlex Kantrowitz50:46

        You think it's in there, the room temperature superconductor?

    24. DHDemis Hassabis50:48

        I think, um, well, I heard that we, we actually think there are some superconductor materials. I, I, I doubt they're room temperature ones, though. But, uh-

    25. AKAlex Kantrowitz50:56

        Okay

    26. DHDemis Hassabis50:56

        ... I think at some point, if, if it's possible with physics, um, an AI system will one day find it.

    27. AKAlex Kantrowitz51:02

        So that's one use. Uh, the two other uses I could imagine-

    28. DHDemis Hassabis51:04

        Mm

    29. AKAlex Kantrowitz51:04

        ... probably people interested in this type of work, toy manufacturers-

    30. DHDemis Hassabis51:08

        Mm
17. 53:21 – 54:57

    Visions of Superintelligence: A Philosophical Perspective

    1. DHDemis Hassabis53:21

       Yes. Yeah. Well, look, I think for there, you, you, you, two things there. One is, um, I think a lot of the best sci-fi can, we can look at as, as interesting models to debate about what kind of, uh, uh, galaxy or, or, or universe do we want to, or world do we want to, to, to, to move towards. And the one I've always liked most is actually the Culture series by Iain Banks. Um, I started reading that back in the '90s and, and I think that is a picture, it's, it's like 1,000 years into the future, but it's in a post AGI world where there are AGI systems coexisting with human society and also alien society, and we, we've, humanity's basically maximally flourished and spread to the galaxy. And, um, I, I, I, that, that I think is a great vision of, um, how the thing, things might go if in, in the, in the positive case. So, um, I'd sort of hold that up. Um, I think the other thing we're gonna need to do is, as I mentioned earlier about the under- under-appreciating still what's gonna come in the longer term, I think there is a need for some great philosophers to... You know, where are they?

    2. AKAlex Kantrowitz54:29

       Mm-hmm.

    3. DHDemis Hassabis54:29

       The great next philosophers, the, the equivalents of Kant or Wittgenstein or even Aristotle. Um, uh, I think we're going to need that to, uh, to, to help navigate society to that next step, 'cause I think the, you know, A- AGI and artificial superintelligence is going to change, um, uh, humanity and the human condition.

    4. AKAlex Kantrowitz54:48

       Demis, thank you so much for doing this. Great to see you in person, and I hope to do it again soon.

    5. DHDemis Hassabis54:52

       Thank you. Thank you very much.

    6. AKAlex Kantrowitz54:53

       All right, everybody, thank you for listening, and we'll see you next time on Big Technology Podcast.

Episode duration: 54:57