The AI Breakthrough That Will Change Everything (Google DeepMind CEO Interview)

[00:00:00] Speaker 1: This is Sir Demis Hassabis. He's the co-founder and CEO of Google DeepMind and Isomorphic Labs. For over two decades, Demis has been all-in on artificial intelligence. He believes that within the next decade, AI could help us cure every disease on Earth. Cancer, Alzheimer's, Parkinson's, and even diseases we haven't discovered yet. So when his team at Google DeepMind invited me to Mountain View the day after Demis closed the I.O. keynote, I had exactly one question I needed to ask him. What has this mission cost you that success can't repay? It's a great question. I think... Before we get to that question, let me tell you what I was actually doing in that room. Because I didn't fly out to Mountain View just to ask Demis what AGI will look like in the next few years. I flew out because for the last 12 months, Demis has been telling anybody who will listen that the most important thing AI is going to do isn't making movies, replacing developers, or running a search engine. He believes that the single most important thing AI will do for humanity is to end disease for good. I wanted to know how he thinks that could actually happen. So that's where I started our conversation. You said something which I agree with. You said the number one application of AI is to improve human health. Yes. And so I'd love to dig into Gemini for Science. As I noticed yesterday, the most vigorous applause was for the scientific research. [00:01:16] Demis Hassabis: What exactly does Gemini for Science unlock for scientists? It's one of the things I'm most excited about. It sort of pulls together a few different projects we've been working on, probably most notably co-scientists, which you can sort of think of as a fine-tuned version of Gemini that has extra additional tools and harnesses, citations, looking up literature, reading graphs, the kinds of things that scientists need to do. It's almost like it's this great research assistant that you have. [00:01:42] Speaker 1: I've had the chance to sit from a lot of tech executives over the last few years, and almost none of them get visibly excited about a product that isn't going to show up in the next quarter's earnings call. But Demis is different. This is the version of AI he cares about most. The version that turns scientific breakthroughs, once thought to be impossible, and the things that suddenly aren't. To understand why he believes what he believes, you need a bit more context on his origin story. In 2020, his team released a system called AlphaFold. It does one core thing. It takes the amino acid sequence of a protein and predicts what shape that protein will fold into. Now, that might not sound groundbreaking, but trust me, it is. Every single thing happening in your body right now is mediated by protein. Your muscles contracting, your neurons firing, your immune system recognizing a virus and fighting it off. All of it. And until AlphaFold, scientists had only mapped the structures of about 1% of all known proteins. It took them 60 years to get there. AlphaFold mapped the other 99%. 200 million structures in about a year. By every standard that existed before it, that was supposed to be impossible. Then, Demis and his DeepMind colleagues gave the entire database away for free to the public. That's the work that Demis was awarded the Nobel Prize for in October of 2024 alongside John Jumper and David Baker. But the AlphaFold story isn't a one-off. Demis didn't get lucky once. He's been at the forefront of AI for over two decades. Back in 2010, he founded DeepMind in London with two friends and a whiteboard. At the time, most of the world had never even heard the phrase artificial general intelligence outside of science fiction. Google bought DeepMind four years later in 2014 for around $650 million. It was the biggest European acquisition Google had ever made. And Demis insisted on one non-negotiable condition. DeepMind could continue doing fundamental science research to push humanity forward. Which brings me to the people Demis is still working alongside on Gemini today. People you've probably heard of. And I've heard that Larry and Sergey are actually back and like reinvigorated over the mission. [00:03:39] Demis Hassabis: Yes, of course. Yeah, Sergey's actually there coding away in the weeds of Gemini. And Larry's always around at the board meetings and strategically, you know, he's very brilliant with sort of far future planning. So it's super fun talking to them all the time. [00:03:52] Speaker 1: Larry Page is in his early 50s, a billionaire many times over, and he's still in the code base. The kind of person who doesn't actually retire, whose hands need to be on the thing pushing society forward. Which is reassuring, because there's a problem with the cure-all-disease mission that nobody has fully addressed yet. If an AI eventually discovers a cure for something humans don't fully understand, if the math is too complex for a human brain to follow, are we supposed to just blindly follow it? That's the kind of question I wasn't sure I was smart enough to put into words, but I asked anyway. A question that I actually wrote down, I was like, you know what, I'm not smart enough to understand this. Let me ask Demis, a guy who is. And it's, say AI discovers like a really complex cure. Yes. And it works. You know, the math and the logic behind it, it just, a human brain cannot actually understand. Yeah. Are we then at that point, is it just, hey, we just have to blindly sort of trust the AI? [00:04:39] Demis Hassabis: Actually, this is, this is a great question. The good thing about the practical sciences is we could test the answer empirically. So with a drug, for example, obviously you wouldn't just trust what the model says. You would need to test it in clinical trials and test it in the laboratory. The thing that takes all the time and a lot of the cost is the searching to find the needle in the haystack, but you still need to validate it at the final step. That would be kind of empirical proof that it's safe. Like you don't really have to understand how it works. Although maybe if you pair an alpha fold with a Gemini, future version of that could explain what alpha folds do it in a way that human scientists can understand. But even if not, when it comes to solving diseases, we really just care about the efficacy of the final outcome. What they do need to know, though, is any uncertainty around which bits does the model think is very confident about. And alpha fold does that, so it outputs its confidence on the different parts of the structure. [00:05:27] Speaker 1: That answer reassured me because empirical validation actually works for drugs. You give them to test subjects, you measure what happens, and you find out if they worked. That's been the standard model for 100 years. But the part Demis kind of glossed over is that empirical validation takes time. I mean, we're talking over a decade and more than a billion dollars to bring a single drug to market today. Demis has said publicly that AI can compress the discovery part, the needle in a haystack part, from years down to weeks. And I believe him. I mean, he's already done it once with alpha fold. But if AI cuts the search to weeks and validation still takes 10 years, that's still 10 years to a cure. Which made me wonder, if the prize is curing every disease, why are most of the GPUs at Google DeepMind going somewhere else entirely right now? And I figured the only person who could actually answer that was the man sitting across from me. You've been really vocal, you know, that you think it's a possibility that we can cure all the disease in the next 10 years. Yeah. But you've also been vocal that you believe sort of the commercial success of generative AI is actually like a hamper in that mission. Do you still believe that? [00:06:24] Demis Hassabis: No, I don't think it's a hamper. I would like to see more work going on in the scientific fields and the medical fields. A lot of the technologies that we're developing for things like chatbots and assistance and so on, not only do they increase productivity and create the commercial flywheel, which then allows us to fund our science work and in some cases give it away for free like AlphaFold. But it also develops some of the underlying technologies which can be used for the scientific [00:06:46] Speaker ?: domain. [00:06:46] Demis Hassabis: I would like to see more work being done in those kinds of medical fields. I read the book, yeah, phenomenal. [00:06:53] Speaker 1: You are a humanitarian scientist. You also have to play the role of CEO of a massively commercial entity. Tell me about that tension when it comes to things like allocating computational resources to open sourcing things versus proprietary things. [00:07:04] Demis Hassabis: Yeah, it is a tricky balance to get right. And I think Google has always been, and obviously DeepMind as well, they've always been sort of, I would say, pretty research and engineering-led company with a huge deep respect for research. In fact, the original Google search was Larry and Sergey's PhD project, right? People on the board, you know, there's Nobel Prize winners, Turing Award winners and so on. Of course, you have to pay the bills. You have to be commercially successful in order to be able to pay for the research. If you do that in the right way, that's fuel to be pulled back onto the commercial side. But you can also explore a few scientific branches, like we did with AlphaFold, kind of for its own sake. They have both sides to me. I'm a scientist and an engineer. So on the science side, I have very interesting big questions going after blue sky research. But I'm also very practical. At the end of the day, I don't want to just be philosophizing about it. We actually need to build things. That the acid test is, you know, is someone actually willing to pay you for something in order to use it, right? And then, you know, it's actually truly useful. [00:08:01] Speaker 1: That's the CEO version of Demis. The man who has to keep billions of Gemini users happy while still pushing DeepMind's medical research forward without the next earnings call spooking the stock. But the real question about which GPUs go toward chatbots versus medicine all hinges on one number. How far away from this medical breakthrough are we actually? If we're 40 years out, this is a fascinating experiment. If we're only four years out, DeepMind's work is the most important thing happening on Earth right now. Out of all the experts that I could have asked this question to, Demis is probably the one whose answer matters most. Have there been any significant or I guess like specific breakthroughs in the last year [00:08:38] Demis Hassabis: that have been sort of tightening that confidence in that timeline prediction? I think we're very close to AGI now. You know, maybe around 20, 30 plus or minus a year. It's not sort of unexpected breakthroughs, but things going as expected. If you refer to interviews I did two, three, four years ago, I would have been saying a five to 10 year time scale, we're still on track for that, the lower end of that. My confidence interval around when it's going to happen is a bit more tight now because of the things we're seeing agents and coding systems that are really helpful to top engineers, mathematics breakthroughs, things like Omni, banana, banana, all of those things in aggregate, as we get closer, the uncertainty around that comes in. All the focus is on the line that you ended IO with, right? [00:09:16] Speaker 1: And I want to read it again. You know, the kids would call this line a bar. It's a great line. When we look back at this time, I think we'll realize that we were standing in the foothills of the singularity. Making the choice to end IO on the human mission, sort of the philosophical angle, I'd imagine that was pretty deliberate. [00:09:31] Demis Hassabis: Yes, it was. I'm still a bit surprised how much pickup it's got. The singularity, for me at least, that term means the era that will begin when AGI has arrived. Given how transformative the technology will be, perhaps the most important technology ever invented, you know, it's going to be a kind of new era for humanity. I think we can feel the beginnings of that. [00:09:48] Speaker 1: Demis believes AGI might show up around 2030. From there, he thinks we could be only a decade away from curing every disease on Earth. He's got the Nobel Prize, the platform, and the leadership on one of the most resourced research labs in human history. If Demis is right, he's about to spend the next decade of his life trying to do the most important thing anybody's ever attempted. But the AGI timeline isn't really what kept me thinking about this interview after I left the room. The person behind the timeline did. There's a piece of Demis' story that often gets buried under the CEO title. He never planned to be a CEO. This is the same kid that taught himself to program on a ZX Spectrum at age 8, reached chess master at 13, and got accepted into Cambridge at 16. The same kid who was the second ranked under 14 chess player in the world before he stopped playing because he believed he was wasting his intelligence and needed to focus it on something else. Demis has been working toward building AGI for decades. The CEO title is just his most recent endeavor. You know, you've dedicated your life to this mission, and solving intelligence is sort of singular North Star. Yeah. The 20-year-old Demis, would you be proud of the progress today? [00:10:50] Demis Hassabis: Well, I was a pretty hard-to-impress kid, I have to say. So I think 20-year-old Demis would have been reasonably satisfied, let's put it that way. [00:10:57] Speaker ?: Yeah. [00:10:59] Demis Hassabis: I would have probably been like, I'm thrilled, this is way ahead of schedule. You should have met me when I was a kid, you would have seen, I had, you know, a lot of things that were needed to be done. But it's sort of, yeah, roughly what I hoped, I guess. [00:11:10] Speaker 1: That kid, who was so hard to impress, is still, somehow, the man sitting in front of me. And that kid will always push what he truly believes in forward to no end. Now, there was one question I'd been waiting all day to ask Demis, a question you'd normally never ask a CEO. There's a science fiction book a lot of people read as teenagers, it's called Ender's Game. A kid is bred from birth to win an interstellar war. Every adult around him knows it, they isolate him, they push him past what's reasonable, and he ultimately wins. But winning leaves him hollow and unsatisfied. When I was a teenager, I read the book, and I thought of it as just a fun sci-fi story. Demis read Ender's Game for the first time when he was 30, and he read it from a very different perspective than I did. [00:11:50] Demis Hassabis: Ender's Game, you read it at 30? Yeah, I'm glad I didn't read it when I was a kid, because I think it might have messed me up. When you're an adult already, you can look back with a bit of reflection and then go, oh, wow, I didn't realize there was this really compelling sci-fi book that somewhat explained some of the things I was feeling as a kid. I wanted to ask, Ender wins, but the victory sort of hollows him out. So the question is, what has this mission cost you that success can't repay? I've always had this sense of, you know, destiny or mission, AI from the very beginning. Since early teenage years, I also knew how important the mission would be and how round it would be for humanity. And it's taken a lot. I mean, I barely sleep. I can't remember the last time I took out a holiday. But it's just, the mission's so important, I think, for the world. And using it for things like human health and advancing science, what could be more exciting and also more meaningful? I think it's just the best thing I can spend all of my energy and life force on and to try and help humanity. So that's maybe a good path forward. Yeah. [00:12:44] Speaker 1: There's something Demis said in that interview that I haven't been able to stop thinking about. When I asked him what this mission had cost him, he answered without directly putting a name to it. He just said that he barely sleeps and can't remember the last time he took a holiday. Then he made a joke about taking a long sabbatical when this is all over. Much like Ender at the end of the book, and I think he meant it, but I also know he's nowhere near taking that sabbatical. Demis has been actively working on this mission since he was eight years old. He was awarded a Nobel Prize for one of the most important scientific contributions of the century, and he went back to work the next day. He believes he can help end disease for good, and he thinks it can be done in the next 10 years. Now, whether or not he's right is yet to be seen, but I can tell you he's not going to stop trying until that mission is complete. It was like the forefront, you know, because like the media for clicks, it's always like the negativity, but yeah, I wish this was the foremost central conversation.