[0:00:05] Not so long ago, the AI of today seemed unimaginable. Now it can write code, generate video, help diagnose diseases, and reason through complex problems in seconds. And as the race to dominate artificial intelligence accelerates, Google is increasingly poised to take the top spot, changing the way we search, how we create, even how we monitor our bodies. And we spoke exclusively with the architect behind it all.
[0:00:34] Google DeepMind co-founder and CEO, Sir Demis Hassabis. He says this is just the beginning. When we look back at this time, I think we will realize that we were standing in the foothills of the singularity. We are so happy to be here having this conversation with you. So thank you for inviting us here. It's great to do it. And a good time for you to do it, I would imagine, as Google and DeepMind.
[0:01:01] Really seems to be pulling out in front of all of the AI competition. Is that what you expected?
[0:01:06] Well, I think that we're very happy with our progress that we've made last few years. I think we've taken giant leaps on the modeling and the research side, but also integrating it, especially over the last year, into all of our product surfaces. So I think it's an amazing thing that only at Google we can really do because we have the full stack all the way from the chips to the frontier models to the AI products. We are seeing right now of...
[0:01:31] A growing AI backlash. Parents sounding the alarm about the dangers of AI chat bots.
[0:01:37] The Department of Homeland Security says AI offers opportunities for bad actors to undermine our democratic system. Data centers that power the AI boom. They're hulking size, dividing communities, draining energy, and spiking neighbors' electric bills. As the former CEO and chairman of Google. Former Google CEO Eric Schmidt was booed during a commencement speech at the University of Arizona. Time magazine selected its person of the year for 2025.
[0:02:06] And this time, it was the architects of artificial intelligence. Interesting. People see it increasing the cost of their energy. They see it claiming jobs. They see it claiming their future. And they hear all of the incredible things that it might do, but they also feel very uncertain about what it is doing right now. How do you answer to that?
[0:02:33] Well, I think it's right to have some worry and some concern about the future because a lot of it is very uncertain. But I think part of it is also the way some of the industry is communicating about what things are going to happen. In my view, all of this is still to be determined.
[0:02:56] Especially the youth of today, they are going to be the ones to write the next chapter and make use of all of these incredible technologies to come up with incredible new opportunities that would not exist without it. And I think that's been the history of technology advances in the past 20, 30 years. They're always worried about at the time, and then we figure out with human ingenuity, you know, how to overcome those challenges. But we really need to do this very urgently and ideally internationally too.
[0:03:25] Internationally, could we make rules here, implement them?
[0:03:29] And say whatever China does is up to China? Ideally, it would be a kind of international, let's say, standards around AI. But I think the U.S. and the West is, of course, a big enough market and easily powerful enough to kind of make some of those standards themselves. And I think we are going to need them as these systems become more and more powerful because they're dual purpose technologies. They can be used for amazing good, which is obviously what myself and others who've been working this for a long time are hoping for. But they can also be...
[0:03:58] You were a chess prodigy as a child. You designed video games as a teen, studied the human brain, won the Nobel Prize. And now you're at the forefront of this technology that is revolutionary. What is it about this moment that you want the world, people, humanity to wrestle with?
[0:04:26] Well, I think that AI, it's actually going to be almost a new era for humanity. I think it's that big and profound a change. Sometimes I compare it to, you know, it's going to be 10 times the impact of the Industrial Revolution and maybe 10 times faster, you know, occurring over a decade instead of a century. So we're talking about really profound transformation. And what I would like to see AI being used for is to advance science and medicine. That's what I set out, you know, over 30 years ago now to do with AI.
[0:04:56] My whole career working on AI is I see it as the ultimate tool to push forward scientific discovery. When you say this is going to be 10 times bigger than the industrial revolution and 10 times faster, and then you talk about curing disease, that is incredibly exciting. But there's the flip side of that. It's jobs. It's are we ready as humanity for this technology? Yeah. How do you balance those two things?
[0:05:22] Well, I think there's a lot of challenges, too. It's a double-edged sword, AI. On the one hand, we want to solve disease and help with things like energy and environmental issues. And I think AI can help with all of that on the plus side. But it will also disrupt jobs. And I think there'll be huge new opportunities as well, actually, is what I believe will happen with AI and these AI tools sort of superpowering the youth of today to be able to do incredible things that were, you know,
[0:05:52] I think the question that people ask is, okay, we understand the new opportunity, but there's lag time there. The industrial revolution happened over a century plus of time. If this is moving so fast, and it's clearly going to displace some workers,
[0:06:21] What becomes of them in the meantime? Well, I think that this is where broader society, civil society, academia, social sciences need to come together to try and chart out how do we as humanity want the next stage to go. The future is not written. I think it's for us collectively to decide how we want economies to change, what do we imagine things like jobs are going to become, but also make sure that we distribute all of the incredible productivity gains that are going to happen with AI.
[0:06:50] You know, to as many people as possible so that everyone around the world benefits. And I think those are all choices to be made, actually. And they're not really to do with the technology itself. They're more wider societal issues that I think a broad group of people needs to come together to discuss and debate. And I think also to get humanity ready for these big changes, it's going to kind of need almost new philosophy, new great philosophy.
[0:07:16] Again, it shouldn't just be about technologists and the people building this technology. It's going to affect everyone. So everyone should be involved in that conversation. How optimistic are you that that conversation is going to happen to solve this?
[0:07:30] So given where things are right now,
[0:07:58] Would you be in favor of hitting the pause button?
[0:08:01] If this goes...
[0:08:30] I don't think it should be.
[0:08:52] I think there needs to be an international standards approach to this. It shouldn't just be the tech companies. It should be broader society. Everyone's got to take this really seriously, like the technologists do. And I think maybe that's starting to happen now. But I think there's real urgency needed around that. You see it happening? I think they are starting to, yes. What do you think they need to know or understand that they're missing at this point?
[0:09:18] Well, I think it's just hard for the human mind to comprehend something that's a kind of exponential. We're just not really built for that with the way our brains have evolved. You know, we think in more linear terms, but this is an exponential technology. And a lot of things are going to happen very fast that are going to be very impactful. So I think that's probably the hardest thing to get your head around. And I think everyone's actually able to, of course, interact with...
[0:09:42] Pretty much the latest AI technologies in chatbots every day, you know, things like Gemini. So everyone now has a feel for what is this technology? What can it do? What is it still weaker doing? And that's great, I think, because it allows them to be understanding of the enormity of what's coming and hopefully help to shape that. One of the other big issues people have with AI right now is the energy cost. Data centers,
[0:10:11] Should the average ordinary person bear that cost? I don't think. I think if you look at most of the energy costs, first of all, they're overstated with the data centers, but also...
[0:10:24] Most of the big companies and the hyperscalers, they're building out their own energy infrastructures to make sure that the brunt of those costs and the energy that they're using is not passed on to the consumer. And Google's definitely doing its part there. The second thing I would say on that is that I believe AI is actually going to be one of the best technologies for environmental concerns. I think AI in the next 10 years is going to save far more energy than it uses in the long run. 10 years from now. Sure.
[0:10:54] It'll save the energy. I think it will less than that. It will save more energy than it's using. And in the meantime? And in the meantime, we'll need to build up that infrastructure and the big companies and the organizations that have those resources will need to bear the brunt of those costs. Coming up, what's the question we're not asking right now about AI that we should be? If you could correct anything.
[0:11:30] About AI that you believe exists in the public's mind right now, that you think is misinformed, what would that be? Well, I think first of all, in terms of the approach to it, I would say that those of us who are building it need to be approaching it with sort of some humility around the amount we understand about these systems and what we think about the certainty of certain paths from here. I think...
[0:11:55] The future is still to be written. I think there's going to be some incredible things that are going to happen, amazing things that are going to be incredibly beneficial for the world. But we each step.
[0:12:04] What evidence have you seen?
[0:12:30] of dramatic change, things that are so much better today because of AI.
[0:12:35] Well, if you take AlphaFold, our big breakthrough in understanding the 3D structure of proteins, over 3 million scientists in the world make use of that now in all of their really important scientific research and medical research. So I think that there won't be a drug designed in the future that won't have made some use of AlphaFold looking into the future. So I think that's already an unbelievable benefit to the world. And that's just one example. What's the single biggest thing that people are not taking advantage of?
[0:13:05] I always recommend to students and when I give talks to universities that they should lean into the tools and become kind of incredibly native with them. So I think there's so much value in what these tools can do already today. They'll kind of be like superpowered over the next 10 years in terms of whatever.
[0:13:29] What do you tell students right now who are so nervous about where the world is? They hear the stories that jobs aren't going to be there for them because of AI, internships don't exist. They have no idea what to study in college.
[0:13:56] What would you tell them to study? I would ignore all the noise around that and just be confident about your capabilities. So the next generation will own how this goes. And I would first of all say, still study the basic STEM subjects, get really good at math and computer science and physics and all the normal subjects because that will make you better at using whatever these tools become and understanding that better, just like my generation did with computers and the internet back in the day. It's another one of those.
[0:14:26] When that comes, there's always a lot of uncertainty, and I get the worries about that. But the best of the next generation will create incredible new opportunities, take advantage of the opportunities that are going to come with any type of disruption like this. What kind of human skills are most important? First of all, getting really natively good with the tools and understanding what they can do for you and how to incorporate that into your workflow. And I think that's going to become more important with the agentic era.
[0:14:55] And then I think it's going to be things like creativity, having taste, bringing sort of what I would call the soul to a product or a design or a feature. I think those are the kinds of things that are going to differentiate what people output or what other people want beyond just the underlying technology. And where does that soul come from in your view?
[0:15:18] It comes from human ingenuity and our experience and intuition. So I often say in science, for example, what separates the great scientists from the good scientists is not their technical ability. You know, if you're a professional scientist, you're already technically really able. It's actually creativity and having a kind of...
[0:15:35] Great taste, sometimes we call it, for what's the right question, what's the right hypothesis, what's the real sort of meaningful experiment that needs to be done. And for now, that's all going to be needed to be driven by the human component of all of these technologies. What's the question we're not asking right now about AI that we should be?
[0:15:57] I think we should be asking ourselves what is this kind of future society and how the world works, including the economy. How should that work? I think all of the top people in the various fields need to pay, you know, really take that seriously. When I talk to economists, I'm not sure they're really taking this seriously. What change is coming down the road to social scientists, to philosophers about.
[0:16:22] How should we adapt society to these enormous, impactful transformations that are coming down the road? How do you answer that question? When you look out five years from now, what does it look like?
[0:16:36] Well, what I'd like to see is the benefits of this spread across the world. I'd like to see the latest systems deployed after serious amounts of testing and that they're secure and they're safe. And we understand what the guardrails around and that those guardrails are robust around the technologies, especially as they get more powerful. And then I want to see incredible new breakthroughs in science and medicine and energy and all these types of fields where
[0:17:04] I think AI is going along with, you know, as this amazing tool is going to allow experts to make probably 50 years worth of advances in five. 50 years worth of advances in five. And why? How does that work? Well, it's basically accelerating the pace of discoveries by finding patterns in data, finding correlations between connections between different subjects and just allowing the human scientists and experts to
[0:17:34] Get on top of so much more information so they can ask the right questions and set the right problems to crack. I talked earlier about the fact that you were a chess prodigy. How did you learn chess? I think I mean, I was four years old. I don't really remember. But apparently I saw my dad and my uncle play and they don't play very well. But apparently I asked to play. And after a couple of weeks, I was beating them. So they took me to the chess clubs and I started winning there. And it sort of took off from there. Do you see.
[0:18:03] A four-year-old like you in the future learning from AI to do what you did with your dad and your uncle? I think not necessarily learning from AI, but I think a kid like me who grew up with computers and programming, I can't wait to see what an equivalent kid like that would do today, growing up with all these amazing AI tools and expanding what they're capable of doing. I think the next generation is going to come up with amazing things. So how do we make sure that
[0:18:31] In everybody's hands, people don't do the wrong things with it, especially when that's becoming more and more easy to do. Yeah, so that's one of the worries is to make sure, of course, you want all the...
[0:18:43] All the people are going to do good things with it to be able to access it widely, things like open source and so on. We open sourced AlphaFold. Everyone in the scientific community can make use of the structures that we predicted. But as they think systems get more powerful, we do have to work out how can we build robust enough guardrails. They can't be removed by a bad actor once you deploy it. And that's a very difficult challenge, a cybersecurity challenge that all the leading labs are working on.
[0:19:11] If I came to you and said, Sir Demis, what do I study in college, given everything you know about AI and the world you are building, what do you tell me? I would say still study the basic core STEM subjects of maths and physics and computer science and biology. I think all of those things, those foundations will still be critical in the AI era. And what's more...
[0:19:37] I think a lot of the big advances will come by applying AI and combining it with one of the other subjects, whether that's creative arts to biology. I think that's where in the next 10 years all of the most amazing advances are going to come from. So you're saying don't give up on the humanities either, if that's your interest. I think actually maybe the humanities are going to be more important than ever because of charting out what should happen in the next era for humanity. Is there anything at this point that keeps you up at night?
[0:20:06] I don't see very much given the amount of work I have to do. But yeah, you know, all these things worry me both. Of course, I'm on the one hand, unbelievably excited with the potential of where we are now with things and trying to cure diseases with AI through isomorphic labs. We're on the cusp of all of those amazing things. But also, I worry a lot about the downsides and making sure, you know, can we mitigate that for society? And when you wrestle with those topics, are you talking to a human or AI?
[0:20:36] I'm mostly thinking about it myself in quiet moments of reflection and talking with my friends. So you're not asking Jeb and I yet? Not yet, but maybe in the future.