11:17
weak and easy questions. Please try to answer in one sentence, sometimes one word. Oh, wow. That's hard. Um, Fineman or Newton. Oh, wow. That's even harder. Dennis, I don't know if I told you, but my mother recently got some health scan results. It was a massive video file, and we had to wait for weeks for the evaluation, and we were really anxious. And then I thought, wait, Gemini, long context, you know, why not give it a look? and it analyzed the scan and said don't worry it's it's fine and later the doctor verified that fantastic that it's it nailed it and I am really grateful for that
and I was about to say thank you. Mhm. Uh but in the meantime you also released Gemma 4 free local AI that can likely do the same with a little compassion. Just wanted to say these are a gift to humanity and thank you so much for doing this for the little man. Fantastic. Well, I'm I'm glad to hear that. It's it's it's and I'm glad to hear your mom's fine, but it's uh we've had a lot of anecdotes like that of people using Gemini for health reasons and actually in some cases life savings. So uh advice so I think it's an incredible use case. So I'm glad it helped.
Yeah. Thank you very much. Now Jensen Hang of Nvidia says that he uses LLM as a confidant for decision- making. What about you? Do you use Gemini for things beyond research? Yes. Um, I don't use it quite as a confidant yet, although maybe at some point I will. I use it a lot for brainstorming. So, you know, project ideas, project names, uh, uh, you know, think creative ideas. I quite like it as a kind of sparring partner for that. Uh, and that's probably the main use. And then I also use uh for and to you know look at summarize some new area of research some new body of research I'm not so expert in but I want to get a
quick take on you know the main key points. Do you use it to get your ideas criticized like fire up deep think you know mathematical Olympia and now criticize my idea? Yes. Uh I have a yes I mostly use it for uh helping me kind of think through some of the steps I've been thinking of as well. I mean I guess you could call it critiquing it, but it's I try it in a more collaborative frame rather than maybe I should try it with more like be harsher come up with the flaws in this. Um but I definitely use it as a kind of sparring partner.
Amazing. Now, when I heard you got the Nobel Prize, I thought finally some proper recognition for the theme park AI. Exactly. Took a while. Yes. And then I heard John Jumper say something that really stuck with me. He said that he's looking forward to seeing someone use your alpha technique to invent something to win the prize. I thought, let's call it the second order Nobel. Yeah. So, do you expect that to happen? Second order Nobel. That's a really interesting idea. I am I think it's it's possible um given the number of researchers who are using Alphafold, you
know, over 3 million at this point and they're all doing incredibly important work and impactful work. So, uh yeah, I guess that John's right that may happen at some point. That would be an amazing moment. Amazing. Now, you have a system called co-scientist that can invent new things. Can you tell me about it? Yeah, you can think of CO scientist as a sort of um fine-tuned version of Gemini that's specifically uh with extra tools and extra harnesses on top that uh is specific for helping with hypothesis generation, helping you analyze data, helping you uh summarize literature as well. So, it's the beginnings of a kind of almost like having a great research
assistant that's helping you in your daily work. It's incredible. By the way, the new hypothesis generator, they also tried it. Oh, great. Gave it some ideas. Yes. And then it was interesting because it wants you to narrow down the idea because you just say something and then just the result was absolutely amazing like wait the eight hours and then I was like this this is incredible and I tried it on my original area ray tracing global illumination very little training data very few people do that. So that's I think it's pretty cool to see it perform on that not just on and it came back with some interesting
sensible ideas helped you. Okay. I wish I had more time. Me too. This is the problem is actually finding the time now to do that. We almost need I feel like we almost need really good AI assistants to help deal with our admin work. We have to do all of that stuff. So we have more time for using co-scientist. Right. This is my dream too. on April 20th, 2025, you said, "I think one day maybe we can cure all disease with the help of AI, maybe within the next decade. I don't see why not." When I saw that, I got so excited. I made a website for that. Oh, cool. cure all disease.com. Okay. And I have some data in there and I wrote, "Check back often because I will update it with new data."
Fantastic. and never updated. Yeah. Well, we're working hard. Uh, no. Yeah, but it won't be it won't be uh a gradual thing. It'll be more like Alpha Fold is the way I'm thinking about it. We're building isomeorphic labs and also our spin out and also uh at Deep Mind as well in our science group um more and more tools. You can think of it as building a platform. you know alpha fold's one of the components of that the you know um advanced versions of alpha fold but as you know protein structure is only one component it's an important component but it's only one step in the drug discovery process so we're building you can think another half dozen to a dozen alpha fold level models that are on different parts
of the drug discovery process and then we got to put that all together and of course test it on um some uh some disease profiles which is what we're doing now in pre-clinical stage um and then once we've proven that out which I think will take a few more years then we may have a an engine that can be applied to you know a platform that can be applied to almost any disease area that's the hope so a bit like with alpha fold 2 you know you get it accurate enough and then suddenly you can fold all 200 million proteins in one year so it's going to be more like that so you may not have any updates on your website for a few years and then suddenly I hope there will be some big breakthrough
cruise where then you'll have you know you won't be able to you have to update it every day. Okay. I think I've heard something similar from Watson's talk. He said that the process is going to be exponential. Mhm. And then at year five they said that well you're only at I don't know 8%. What are you doing? And he said no that's fantastic. I mean if it's doubling every year we are way ahead. Yes. He may have been by the genome project probably or the human genome project. That's exactly right. And we've already seen this once with Alphault. So, um, I we need to rep, you know, replicate that success.
Obviously, it's much more complex. Um, and I also meant by that as well that we would potentially have a platform that could come up with those uh potential cures. You still would need to test them in the clinic and go through that process. That could still take more time. Speed that up. I think actually AI could also help with that, too. Like stratify patients, maybe predict dosages better, all of those things. So I actually think AI could probably speed that up too. Um so there's two parts you know to drug. You got to do the drug discovery process and then you've got to do the clinical trials and right now we're focusing on the first part but I think AI could also help with the second part. That's when
we would then have a step change in human health. So with that I think we are on track right so cure all disease in 9 years. You know, I won't say specifically roughly that kind of time frame, but I think in the next 10 to 20 years, there's I don't see any laws of physics that prevent that, right? Yeah. Think entropy. Exactly. And when I read drug discovery papers, it's it's not my area, but I get really excited and then I ask an expert and they say, "Yes, exactly what you said that's just one step and there's a thousand other steps." Yes.
So I'm wondering what can we now do with AI that we couldn't before in drug development? Well, I think what the kind of things we're working on is um so there's obviously there's advanced versions of alphafold. You can imagine um we can predict lots of interactions now. So not just static picture of a protein, proteins interacting with other proteins, proteins interacting with molecules. Um and then the next the hard parts are then predicting what is it going to do in the body. So when you know where does it bind to are there you know what's the adne properties absorption and toxicity all of these things which ends up being side effects. So we've got to build models that predict all of these different things um and then put that all kind of all
together. Uh as well as like kind of biochemistry models of like what exact compound would you design and how is it going to be made and how and where how will it bind to the particular pocket or whatever it is that you're targeting in the in the protein. Mhm. My impression is that there's a lot of things in drug discovery that we cannot do at the moment and we will be able to do with AI like as of and whatever follows. What are the things that we cannot do at the moment and you see that's maybe not changing. I don't know maybe something about regulations that you know that evades physics basically. regulations obviously that's a bit outside the scope of AI to help with but
what I hope with there you know if you think about regulations FDA these types of things obviously they're incredible important part of the process but they could be sped up like we were saying earlier with um more evidence that once a few AI designed drugs I think get through the whole uh regulatory process then you might have enough data say there's 10 AI designed drugs and nine of them worked right then you would be able to back test How accurate really were your models in their predictions? Which ones can you trust, which ones can't you? And then that would be evidence to the regulator. I think that some of those process could be sped up uh and maybe some of the steps could be uh skipped or replaced with something
more efficient. But I think it's going to require evidence first of all that a few of these AI assisted or AI designed drugs get all the way through to actually help the patient in the probably traditional way and then we can change you know uh maybe think about changing the way that's done maybe similarly as what we had with the mRNA vaccines right yes potentially right so you know there was a new technique there was this emergency need obviously with COVID and then that accelerated uh the tests and the trials on those things. Um, you know, that's that's one potential way it could go. But I don't think there's I mean, this is something about human health for the next, you know, next 10 centuries, right? We don't need to rush it uh in the next 5 to
10 years. But I just think what I see is such exciting technology coming down the road. Um, I just think anything may be possible. That's incredible. Now, as a former player, can you tell me a bit about your new partnership with the EVO Night game? Oh yes. Oh, you a former player. Oh yeah. Okay. Fantastic. Yeah. Well, it's an awesome game and uh I think its community is very special actually. You know, very rare. The community sort of builds the game too really with the way they build their alliances and factions. Um it's incredibly deep game uh you know, massive multiplayer. It's a whole universe really that's created not just by the game designers but also by the players. And uh it has a function
economy. um and really interesting kind of dynamic story lines. So, we think that uh we've admired uh their work for a long time. I know Hillilmar, the CEO of the company very well. We're we're good friends going way back to uh when I was still doing games and we think it's a fantastic uh sort of sandbox if you like for testing out some interesting AI ideas that will help the game play and um you know be interesting kind of to test out how it interacts with economies and the story lines and player alliances and things like that. So it's going to be an interesting proving ground. I think fascinating for the game and the community and uh and also like they're always been at the cutting edge really. So this is continuing that tradition and
then for us it's also continues in the tradition of deep mind of using games as a as a safe kind of proving ground. Would the concept work in a way that some AI agents would be embedded playing with the players? Yeah, potentially or maybe assisting the players. I think we're we're still early days of exploring that. We want to imagine also maybe it's almost like a games master that drives the story line as well. That is so cool. Look, I cannot wait to see an AI bot get scammed in Jeta. So that's you'll have to you have to start playing again. Yeah. And by the way, if you need someone to double your money.
Yeah, I know a guy. You're the guy. Okay. Look, all the others, those are scams. This one guaranteed. We'll make sure our AI agent comes to you for that one. Now, can you tell me a bit about the things that the co-scientist has invented already? My understanding is that it's not directly writing papers. Scientists are writing up the papers and that's right. So, it's it's more like an assistant today rather than autonomously discovering things. You know, maybe that's the next step. Um, but right now it's assisting uh scientists in, you
know, and mathematicians in their work. Um I hope we'll be announcing some things pretty soon but earlier versions of these types of systems were doing things like finding more efficient matrix multiplication or like we used it for uh an alpholve and other tools as well for basically improving like computer science algorithms. So, kind of turning uh invention on itself to make itself more efficient, which is pretty cool. And I think we obviously we're just putting it out there now and we're just scratching the surface of what I hope will be we'll see the potential of it is going to be.
Now, wouldn't this be the ultimate touring test of AI? Like you have the Einstein test where you stop it at 1901 or somewhere else. See if it can special relativity. Yes. Now, what about the Einstein test squared assuming bigger than one? Uh so the cut off is today like invent something new that's like a Nobel Prize worthy thing. That's obviously what we would want and but in order to prove that's the case we would this Einstein test that I've I've been saying about you know if he took it back to 1901 with a knowledge cutoff could it have invented it you know Einstein's anus Morabus of you know 1905 where he came up with you
know four amazing papers all breakthroughs in their different areas right including special relativity. So if it was able to do that then presumably you could turn that model obviously trained on all modern day physics and then ask it for you know something better than string theory right and then maybe you should take quite seriously what it came back what it came what it came back with right so that's that so that would be the kind of back test to show that kind of system those types of uh techniques would be ready to actually invent new science I'm I'm I'm going to try to put in this one because I really want you to hear about Now I'm I'm thinking that in rate tracing research we do a lot of Monte Carlo integration and we pray to John for no one.
Yes. And essentially you have uh you get a noisy image from a module that we call a sampler and then you pass it over to a deninoiser. Mhm. So you get a cleaner image. Now that kind of works but the best systems do more than that. They have a connection between the deninoiser and the sampler. So the deninoiser actually tells the sampler that hey I want this image with more samples in the high frequency regions so that the denoiser can do a better job. So it's not two separate units anymore. It's it's fused into one unit.
Now what I'm thinking is that do you have something like this between the hypothesis generator and the verifier? Well I that would be kind of like closed loop automated discovery right if one could fuse that. Um most of the the frontier labs are working on trying to think about recursive self-improvement. I think it could work in some domains like coding and math because the verifier that you can verify very easily and very quickly whether the answer is correct or making progress. Uh and you can also generate synthetic data that way as well. So um I think it's a bit harder if it's a physical science or kind of natural science like physics, chemistry or biology where the verifier
will probably also require an automated lab or something in the world of atoms and that will obviously make the loop a lot longer. Um we are thinking about things like automated labs uh in isomorphic um and I think at some point we will do that. I'm I'm waiting a little bit to see what type of data do we really need that we can't get from a CRO or something like that. And I'm also waiting for robotics to get a little bit more advanced too, which we're doing at Deep Mind. So that's going well. So maybe in 18 months, 24 months, I could imagine setting up some kind of lab like that. We're actually doing one for material science. We're putting a automated lab together. We're we're building one out in London because we
have uh material designs. Actually, we're sitting on 200,000 designs of new materials, which we don't know. There could be an amazing material in there, but we haven't got a way of testing that fast enough. There's some superconductors in there. All sorts of really interesting things. Sometimes hard. Yeah. So, you need So, I think for those types of domains, most of the science domains, the verify step is going to require some physical verification. And then the question is like what's the bottleneck in the discovery loop, right? is it the hypothesis generation or is it the validation step? So it may not be so easy to do that kind of recursive self-improvement. The other issue is even in maths and coding uh all of the frontier labs are kind of
thinking about this is you got to think about the safety of that type of uh process because it's just sort of with you know if no human is in the loop. Now at the end this is going to be a quick lightning round. Okay. Quick and easy questions. Please try to answer in one sentence sometimes one word. Okay. Okay. Discrete or continuous mathematics? Hint, there's only one good answer. Really? Okay. Well, I prefer discreet, but I don't know. You know, is that the right answer?
Um, for you? Yes. For me? Yes. Exactly. I think all of this is discreet. Okay. Fineman or Einstein? Oh, wow. That's hard. Um, probably Fineman for me, but my personal favorite. Okay. Fineman or Newton. Oh, wow. That's even harder. Uh there. Maybe I have to go. Maybe I have to go with Newton for my Cambridge connections. All right. Favorite two-minute papers episode. Oh, wow. Okay. Oh, I love loads of your two I use it a lot to get up to speed with things. But I tell you what, I loved your Alpha Fold uh videos. I thought they were excellent. All of
them. Some of the best explanations out there of how AlphaFold works. That is very kind. Thank you. Best Azimov novel uh foundation series. Yes, that's what was really influential for me. Have you read the robot series? Caves of No, I haven't. Interestingly. Please. Yeah. I just read the Foundation series. So that's But I should probably read the robot ones, too. Please. And now, Sir Demi Sasabis, a knighthood is one thing. But for your contributions to humanity today, I shall bestow upon you the two minute papers badge of honor.
Oh, wow. You're the first scientist to ever receive it. Amazing. So, with that said, Oh, thank you very much. I will wear it with pride. Thank you when I'm watching all your videos. Awesome. Thank you, Dennis. Thank you so much. I learned a lot today. So much fun. Yeah. Thank you so much. And thanks for the badge. Brilliant. Now, this is me from back home with some behind thescenes footage literally because you can't see us properly. Okay, better now. So, this was a huge honor and a dream come true. I can't believe that I got to be there. But what
happened afterwards? Well, I put this pillow on the table and told Deis that I couldn't get a proper one. So, I asked, "Want to see the other side?" And this is what it looks like. And when I showed him, there was huge laughter, which unfortunately we couldn't get on camera. However, Lucy Curly was in the room and shot a photo of it. So, check it out. Thank you so much. If everything goes well, next interview is Jeff Dean. So, subscribe and hit the bell if you enjoy this and would like to see that too. Here you see me running the full DeepSeek AI model through Lambda GPU Cloud. 671 billion parameters running super fast and super reliably. This is insane. I love it and I use it on a regular basis.
Lambda provides you with powerful NVIDIA GPUs to run your own chatbots and experiments. Seriously, try it out now at lambda.ai/papers AI/papers or click the link in the description.
