← Intelligence Squared
Intelligence Squared
Demis Hassabis and Sebastian Mallaby on The Quest for Artificial General Intelligence (Part One)
00:37:52
Loading summary
Transcript40 lines
- [00:00]
Quince Brand Narrator
Springtime is my catalyst to switch out the major players in my closet and take stock of what I have and haven't been wearing over the last year. It's a great time to get a bit more intentional about what you're wearing day to day and if I'm getting rid of anything, I want to make sure that I'm replacing it with quality pieces and I've been turning to Quince for that so often recently. Their clothes are made really well and price even better, so it makes shopping for and wearing their pieces simple. Quince uses premium materials like organic cotton and ultra soft denim and their lightweight linen pants, dresses and tops start at just $30. They've also got incredible accessories. I just picked up a cognac Italian leather sling bag which is a huge upgrade from the crossbody that I have been using. The leather itself is really beautiful and I also love the gold hardware it comes with. I think it's just such a sleek bag. Everything at Quint's is priced 50 to 80% less than similar brands because they work directly with ethical factories and out the middlemen. So you're paying for the quality and craftsmanship of the products, but not a brand markup. Refresh your everyday with luxury you'll actually use. Head to quints.com intelligence for free shipping on your order and 365 day returns. Now available in Canada too. That's Q-U-I-N-C-E.com intelligence for free shipping and 365 day returns. Quince.com intelligence
- [01:26]
Progressive Insurance Advertiser
insurance isn't one size fits all, and shopping for it shouldn't feel like squeezing into something that just doesn't fit. That's why drivers have enjoyed Progressive's Name your price tool for years. With the name your price tool, you tell them what you want to pay and they show you options that fit your budget enough. Hunting for discounts, trying to calculate rates and tinkering with coverages. Maybe you're picking out your very first policy, or maybe you're just looking for something that works better for you and your family. Either way, they make it simple to see your options. No guesswork, no surprises. Ready to see how easy and fun shopping for car insurance can be? Visit progressive.com and give the name your price tool a try. Take the stress out of shopping and find coverage that fits your life on your terms. Progressive Casualty Insurance Company and Affiliates Price and coverage match limited by state law
- [02:19]
Mia Sorrenti
welcome to Intelligence Squared, where great minds meet. I'm producer Mia Sorrenti. Today's episode is part One of our recent live event with CEO and co founder of Google DeepMind and Nobel laureate Demis Hassabis and journalist and author of the Infinity Machine, Sebastian Mallaby. Hassabis and Mallaby joined us at the Friends House in London to discuss the rise of artificial intelligence and the pursuit of AGI. Drawing on Mallaby's unprecedented access to DeepMind, the conversation explores Hassabis journey from chess prodigy to AI pioneer and the ambitions driving one of the world's most advanced research labs. Together they examine what artificial general intelligence might unlock, what it might cost and what this story reveals about humanity's enduring drive to build and to understand. Let's join our host, Tom McKenzie live at Friends House.
- [03:11]
Tom McKenzie
Thank you very much indeed. Let's get to the introductions that really matter. So Demis Hassabis, obviously sitting to my Left is the CEO and co founder of Google DeepMind and Isomorphic Labs. In 2024, Demis was jointly awarded the Nobel Prize in Chemistry for AI research contributions for protein structure prediction via an AI model named AlphaFold. DEMIS is a former child chess prodigy and is a five time winner of the World Games Champion. He's a fellow of the Royal Society and the Royal Academy of engineering. In 2017 and 2025 he featured in the Time 100 list of most influential people and in 2018 he was awarded a CBE. So a very average CV, I think you'll agree. Sebastian Malaby. Sebastian is the author of several books including the Power Law, More Money Than God, the Man who Knew, which won the Financial Times Ann McKinsey Business Book of the Year. A former Financial Times contributing editor and two time Pulitzer Prize finalist, Mallaby is the Paul Volcker Senior Fellow for International Economics at the Council on Foreign Relations. And before we get started, the Infinity Machine, the book about which we are going to be speaking this evening, has become an instant best seller. Congratulations, Sebastian and Davis on both sides of the Atlantic. So make sure to grab your copy after this conversation when Sebastian will be signing. Thank you both very much for taking the time. I'm going to start with you, Sebastian.
- [04:57]
Sebastian Mallaby
Okay.
- [04:57]
Tom McKenzie
Why write a book about Demis Hassabis and why now?
- [05:01]
Sebastian Mallaby
So the short answer is AI is the most interesting transformation in the world and Demis Hasabis is the most interesting figure in AI. If you want me to elaborate, I can. So this is. We're talking mid-2022 and I'd finished my previous book about Silicon Valley and technology investing and I was looking around for the next subject and it seemed to me that I'd been to tech conferences. This intriguing, interesting character called Demis had showed up. You know, he seemed like the guy next door whose manner of speaking was kind of, you know what, after lunch maybe we just do the dishes quickly and then we go and play football. You know, it was that kind of, you know, very approachable vibe. But then what he was actually saying was not about the dishes. It was some fusion of neuroscience, computer science, the history of movies, philosophy, Kant. I mean it just was a mind blowing diversity of thought, agility of mind combined with this approachable Personas. I was always intrigued and it seemed to me that AI, although it was a bit niche in mid 2022 was clearly breaking out. Right? You had two systems created by Demis company DeepMind, which had effectively solved an infinitely large or almost infinitely large search space. So with AlphaGo, this was a system which looks at a game where there's a 19 by 19 board. So the very first move there's 361 options and then the next move there's 360 options. You multiply that out, you get to a massive number. So finding the right strateg gene go is like conquering an almost infinitely large search space. And then the AlphaFold system to which Tom referred is another thing where you can bend the strand of amino acid into a protein shape in an almost infinite number of ways. If you can predict how it's going to fold up and what shape it's going to have, you've solved another almost infinitely large problem. And once you have an infinity machine, it seemed this thing would break out. So I went and pitched Demis on the idea of a book about him. We can maybe get to that conversation later. And this was November 2022 and I thought, great person, slightly niche subject. And then a week later ChatGPT came out. It went from the fringe to the mainstream faster than I could have imagined.
- [07:23]
Tom McKenzie
And the book, which is fantastic, charts not just your life and career, the accolades, the wins, the successes, but also Demis, the ups and downs, the missteps, the miscalculations, the moments of real tension, the battles that you had to fight through your career. And again, it starts when you're four or five years old and talks about your childhood and all the way through to where we are now. Any regrets?
- [07:48]
Demis Hassabis
Well, no, I mean overall it's been, you know, I look back on it, it's been an almost unbelievable journey really. Even though I had planned a lot of this, the type of things that have happened that sort of, I guess I dreamed about it and I planned for it, maybe using all my chest training. I think that's sort of the way I think about life. Is this considered way planning back from your goal, breaking that down into sub goals? I think it's generally applicable to life, or at least that's what I've tried to do. And it's been pretty effective. But even still, it all coming to fruition and then now seeing AI is the biggest topic in the world, which I kind of always felt was going to be the case 20, 30 years ago. But even still, it's been surprising to actually see it being realized. And it's been amazing. It's been very hard journey and lots of twists and turns that SRI Sebastian describes very really well in the book. But in the end, my North Star of what I wanted to do has basically not changed since I was a teenager. So I always felt I was going to. I was committed to this route no matter what happened. Even if we were sitting here, you know, in one of the universities around the corner, and still nobody was interested in AI Today, I would still be working on AI because I believed it was the most important thing and also the most interesting thing, frankly, that I could think of working on.
- [09:15]
Tom McKenzie
There's a lot of key lessons that cut through or key takeaways that cut through from the book. One of them is that that maybe unlike some others in the field, this really has been a guiding light for you from a very, very early age. And it's been so instructive as you followed that goal. What was it like? It seems like you both spent quite a lot of time in the pub, by the way, as part of this book. Demis, what was it like? You had to relive parts of your life, but also relitigate some parts of your life with Sebastian. Was this a cathartic process for you?
- [09:43]
Demis Hassabis
Well, I approached it a little bit with trepidation to begin with because I'm pretty private person and now a lot of my life is out there. So that was a strange feeling. And it's something you have to. When I met Sebastian for the first time, I kind of trusted him with this story and I'd read some of his other books and I was really impressed with both the expanse of the books, but the thoroughness of them and the accuracy of them. I think that's a hallmark of Sebastian's work. And I thought it was really important here for this story to be told in a kind of accurate way. And that's why I also wanted to put time into it myself to make sure that the stories that were foundational to what ended up happening were correct. And then I think as we got started, I don't think we agreed on a number of hours, but it ended up being a lot because I think I just enjoyed Sebastian's company and these discussions we were having. We were only having coffee in the pub, just to be clear. We weren't drinking pints and pints in there at lunchtime, but it's just. It was near to my house, and so it was easy for me to get to in the middle of all my meetings. And then I actually found it quite helpful for myself. Maybe not cathartic, but at least in terms of going back to those. All the things that happened to me and just trying to actually piece it together for myself. And then it made more sense to me, actually, at the end of that process. And I think we also enjoyed. Sebastian has many interests, like I do, and many eclectic interests. I think we just enjoy discussing all these different things, all of it. Which came back to AI in the end. Because in the end, I mean, that's the whole point of AGI. It's applicable to almost anything. But I think there's this richness around it, from philosophy to the technology to the arts. And we're both, I think, interested in all of that. So I think it was. It was really fun conversations.
- [11:40]
Tom McKenzie
And Sebastian, clearly one of the key questions for you that threads its way through this book is a fairly fundamental one, which is, can we trust this person, Demis, who is leading the introduction of this technology, at the forefront of leading what could be, and it's likely to be the most impactful technology that humankind has ever had to wrestle with and ever benefited from. And the downsides to that as well, can we trust Demis to shepherd this technology into the world? At the end, just between us and the audience, at the end of the writing process and spending time with Demis, what is your conclusion? Can we trust Demis?
- [12:24]
Sebastian Mallaby
Yes. Again, there's a longer version of the answer, which is that you can't necessarily trust everybody who's out there in the field. And I think by now quite a lot is known about the different AI leaders. If you look at Sam Altman, he's somebody who thought about running for governor of California. He's apparently contemplated a run for president. He just wants to be a very powerful person, and AI is kind of his way of getting there. And because of that motivation set, whenever he has a choice between sort of being careful with the technology and going faster, he embodies that Silicon Valley move fast and break things ethos, and he goes faster. He doesn't pause. And this is true right the way through the story. And the New Yorker had a long piece which was published last week, basically calling him a repetitive liar. And to be honest, everybody who knows Silicon Valley is in that conversational loop, which I've been since my last book. It wasn't a surprising article. So that's one kind of person. Then you've got somebody like Mark Zuckerberg at Facebook who wants to make AI to basically make Facebook more compelling, one could say more addictive. And he's about commercial advantage. And then you've got Demis, who I think I can say with pretty good authority, having not only talked with more than 30 hours in the pub, but also I did talk to 100 other people around Demis, including kind of bumping into strange old friends of mine who had kids in the same school as Denis. I really do know the person next to me extremely well. And so I'm very confident in saying that his values are good. He wants this, first of all, the AI to be good for society. And secondly, his primary motivation is scientific discovery. And that's what he's dreamt of from the beginning. And this is who he is.
- [14:31]
Tom McKenzie
Scientific discovery and some words of advice from your father at a very young age, when you were a chess champion competing stacked up on books on top of chairs and stools against the age of 5, 6, 7, against grown adults in the room. And one of the pieces of advice from your father was, do your best, do your best. And you took that very literally, Dennis.
- [14:52]
Demis Hassabis
Yes, I did. I mean, look, I don't think it's my father's fault that he meant it in a. I'm sure in a very comforting way. You know, it was more like, it's all fine if you don't win as long as you do your best. But I'm a bit of an extreme person, as sort of maybe comes out in the book, and I took it in a way, but it's extreme and logical. Maybe I was sort of thinking, I must have been about six or seven. Like, what. What does your best mean then? Exactly. So if what. How do I know I've done my best? Well, it must mean to the point of complete exhaustion just prior to near death. Then you've done your best, haven't you? Isn't that logical? I thought. I thought it. It's definitely not what my dad meant by it, but I was just thinking about this as I was playing my games and that's how I took it. And I agree, it's good, it's all okay as long as you do your best. But that means near the absolute limit of exhaustion. So the only thing you've got to make sure you can keep going. But just one step off of that.
- [15:53]
Tom McKenzie
I've told my children this, so there's new expectations in my household. Sebastian, how has that drive, very fundamental drive for Demis, how has that threaded its way through his life and career?
- [16:05]
Sebastian Mallaby
Look, I mean, all the time is the answer. So to begin with, and this is another contrast with some of the other players in the AI space, the amazing thing is that Demis conceived this ambition to build powerful AI before he went to university. That's crazy. I mean, this is like 1993. AI couldn't recognize the picture of a cat until 2012. So we're talking 19 years before that. This man wants, you know, he's determined to build powerful AI. You've got to be off the charts determined to conceive that kind of ambition. Then he goes to Cambridge. Then, you know, I know what, I'll found my own company. I talked to Demis friends from Cambridge. Like, nobody talked about leaving Cambridge and starting their own company. There was no Silicon Valley in sight. There was no tech ecosystem in sight. This was pretty radical. And then he says, oh, my next move, I think I'll do a PhD in neuroscience. This is a man who didn't do neuroscience. He didn't do medicine before, he didn't do psychology before. Like he was a computer scientist. Oh, I'll just do neuroscience. Well, it's not that simple. You have to kind of talk your way into a top program. So he manages to do that and is told by the professor, okay, you can come, but you do have to read a few papers in neuroscience before you show up. So you get married, you go to the beach, you read a whole stack of papers on memory and imagination, and by the end of your honeymoon, you have an idea which turns into one of the most cited papers in neuroscience that decade. This is, this is off the charts extreme in terms of determination. So all the way through. And we'll get to this maybe later. But I remember very clearly when we first began talking. I remember I said at the beginning, I pitched the idea of this book right before ChatGPT came out. And I think maybe the second or third meeting showing up to see Demis. I said, ChatGPT has gone viral. It's the most quickly expanding piece of software in terms of its reach in the history of Software, this is your rival. You were supposed to be the leader. What happened? How does it feel? And Demis said, this is war. They have parked the tanks on my lawn. So you do have to be competitive to change the world.
- [18:32]
Tom McKenzie
Yes, and we will get to that, to that part of the story as well, in a little bit more detail. Hopefully that determination also played out when you were courting one of your first investors, Peter Thiel, and the Founders Fund, and famously they said, well, obviously you're going to move to Silicon Valley. The idea of investing in the UK seemed completely off the charts for them. They had no concept about doing that. You determined to stay in London, and the consequences of that are all around us. And the founders and CEOs that I speak to are startups, almost to a name will say DeepMind. And the impact of DeepMind has been such a catalyst. Has there ever been a moment, Dennis, when you thought London wasn't the right choice?
- [19:14]
Demis Hassabis
No, I was determined to do it in London. And as you say with Peter Thiel at the time, he's obviously very famous investor, very famous contrarian investor. They had never invested outside of Silicon Valley, let alone somewhere like the uk, and really it was just considered a backwater, especially for deep tech. So things that are going to require a lot of research and many take many years before you get any kind of commercial traction, which Obviously something like AI back in 2010 was in that position. But I was determined to do it. Partly, obviously, I had personal reasons to stay. There was a bit of an underdog in me of like, I want to show that London and the uk, I'm very passionate about the uk, could do this. But the main thing was I knew there was the talent here, the people I'd studied with at Cambridge, and phenomenally smart people. I'd even did a post. I did a. After my PhD, I did a postdoc and I specifically went to MIT and Harvard. I did a joint postdoc@the2 to sample both at the same time. And I did it mostly, but I did it mostly because I wanted to check if it was true that there was another level, because my American friends say Cambridge, Oxford's pretty good, but there's MIT and Harvard and Stanford. So I needed to just go and sample that for myself. And they were very good places, but they were not better than Cambridge and Oxford and UCL and Imperial. So they were the same level. So that part was a myth in terms of like, no, the talent is here. Then the question was, I think what was missing was the ambition to do something really world changing. And then it was also, I thought it could be a competitive edge because if you're a Cambridge PhD in theoretical physics, but you didn't want to stay in academia and you didn't want to go and work in the city in a hedge fund, but you wanted to stay in the UK and utilize your skills and your brain to the fullest level. What could you. What other options were there? So I wanted to create that option with DeepMind in the early days and in fact be this talent magnet for not just the uk, but actually the best talent all around Europe. And there's a ton of talent in places like Switzerland and Eastern Europe and so on, mathematical talent. So we would. And we had the field to ourselves for about four or five years, the formative years of DeepMind where we got this incredible talent that was being overlooked in the US and for various reasons they wouldn't want to move to the us, maybe their family was in Europe or things like this. And we had that to ourselves and for obviously a lot lower salaries and other things. That's all changed now when I'm pleased that it's changed that. I think with the success of DeepMind and a few other companies here, it's shown that deep tech can be viable outside of Silicon Valley. And of course the US Giants have realized this, both the VCs and the big tech companies, and they've invested now, many of them have their European sort of head offices now here in the UK and in London. And then the final part was actually to do with AGI, which is that we planned for success even back in 2010, which seems crazy because nothing was working. In fact, the thing that everyone understood was that AI didn't work. It was a dead end. We know this doesn't work. This was the prevailing view in MIT as well, which was kind of the heart of traditional AI. We tried it in the 90s, people like Minsky and Chomsky, all these greats, and we know it doesn't work and. But we wanted to give it another go here. And I just think AGI was going to be so important, which is how it's turned out, AI, that it shouldn't just be like the people that are making it shouldn't just be from 20 square miles or kilometers of the US. It's going to affect the entire globe. So I think a global perspective on AI, what it should be used for, how it should be deployed, the ethics of it, the technology itself, and I think we contribute partly to that by having a big base here and thousands of researchers and we just opened a beautiful new office just around the corner. Platform 37 as part of that and making that conversation global even within Google and then hopefully for the whole field.
- [23:37]
Tom McKenzie
And you've built that multidisciplinary team and that comes through through the book as well.
- [23:45]
Progressive Insurance Advertiser
Starting or growing your own business can be intimidating and lonely at times. Your to do list may feel endless with new tasks and lists can easily begin to overrun your life. So finding the right tool that not only helps you out but simplifies everything as a built in business partner can be a game changer for millions of businesses. That tool is Shopify. Shopify is the commerce platform behind millions of businesses around the world and 10% of all e commerce in the US from household names like Gymshark, Rare Beauty and Heinz to brands just getting started, Shopify has hundreds of ready to use templates that can help you build a beautiful online store that matches your brand style and you can tackle all the important tasks in one place, from inventory to payments to analytics and more. No need to save multiple websites or try to figure out what platform is hosting the tool that you need. And if people haven't heard about your brand, you can get the word out like you have a marketing team behind you with easy to run email and social media campaigns to reach customers wherever they're scrolling or strolling. Start your business today with the industry's best business partner, Shopify, and start hearing. Sign up for your $1 per month trial today at shopify.com realm. Go to shopify.com realm that's shopify.com realm
- [25:11]
Various Advertisers
it's time to bring on the blooms at the Home Depot with Spring Garden deals. Find savings on hanging baskets and flower flowers to brighten your backyard or any space that needs instant color. Then get everything you need to plant and protect them with low prices guaranteed on soil and mulch. Dig into Spring garden deals for four days at the Home Depot now through May 10th exclusion supply. See homedepot.com pricematch for details. Study and play come together on a Windows 11 PC and for a limited time, college students get the best of both worlds. Get the Unreal College deal everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 Premium and a year of Xbox Game Pass ultimate with a custom color Xbox wireless controller. Learn more@windows.com studentoffer while supplies last ends June 30th terms at aka Ms. CollegePC
- [26:12]
Progressive Insurance Advertiser
ready to soundtrack your summer with Red Bull Summer All Day Play. You choose a playlist that fits your summer vibe the best. Are you a festival fanatic, a deep end dj, a road dog, or a trail mixer?
- [26:23]
Demis Hassabis
Just add a song to your chosen
- [26:25]
Progressive Insurance Advertiser
playlist and put your summer on track. Red Bull Summer All Day Play Red Bull gives you wings. Visit red bull.com brightsummerahead to learn more. See you this summer. When you need to build up your
- [26:39]
Various Advertisers
team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost
- [26:45]
Progressive Insurance Advertiser
it needs to be seen and helps reach people with the right skills, certifications and more.
- [26:50]
Various Advertisers
Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a
- [26:56]
Progressive Insurance Advertiser
$75 sponsored job credit@ Indeed.com podcast. That's Indeed.com podcast.
- [27:02]
Various Advertisers
Terms and conditions apply. Need a hiring hero? This is a job for Indeed Sponsored Jobs.
- [27:13]
Tom McKenzie
Do you think, Sebastian, do you think, given your time looking at the VC world, the hedge fund world, and now of course, Demis and DeepMind, has the UK done enough to embrace the kind of strengths that Demis and the team at DeepMind have lent into?
- [27:28]
Sebastian Mallaby
Well, I think there's a common perception, which I hear from technology people and investors in London, which is, oh, it's a terrible thing, that DeepMind was sold to Google in 2014 and now it's basically a subsidiary of an American company. And I think that's wrong for a couple of reasons. One is that in 2014, and we're talking early 2014, there literally wasn't a choice, right? If you wanted a lot of money to build ambitious AI, you couldn't get it from venture capital investors. They didn't have enough cash. They certainly weren't going to deploy enough cash in Britain at that time. You could choose between different big American companies to sell to. And indeed, Demis did have a choice. Elon Musk wanted to buy DeepMind. Mark Zuckerberg at Meta wanted to buy DeepMind. There are some funny stories about why Demis chose Google, but that turned out to be a very smart choice because then Google a plowed in nearly a billion dollars a year of investment for the next 10 years or whatever before there was really a commercial application and secondly allowed Demis and DeepMind and the team to remain in London with autonomy being good for the British ecosystem because there are lots of scientists who come and work at DeepMind and some of them stay and some of them leave. And if they leave, they're still in London and they're doing other deep tech things. And so I view this not as a tragic sellout to the American parent. I view it as a brilliant wheeze to get the Americans to pour loads of R and D money into London. Great. But I think there is another aspect in which London should beat itself up more than we do. And this is that there was an opportunity, and I describe this in some detail in my book, where DeepMind really wanted to help. The National Health Service really wanted to do AI for medicine, really wanted to solve some of the kind of critical bottlenecks inside a public health system that we all believe in, but we're kind of disappointed it doesn't work better. And so it's like almost a dream. Okay, now you've got these brilliant technologists. They're British, they're based in Britain, they love the nhs, they want to help. They did it pro bono at first. This is just amazing. And instead of embracing that, there was an absurd kind of backlash against some allegations of data privacy problems, which I really looked into deeply, and I can tell you they were all bogus. And so that's where we should be angry about the British record on tech. We need to embrace it, absorb it, use it, and not be.
- [30:14]
Tom McKenzie
And that remains a very live debate today, doesn't it? I want to go back to November of 2022, as you brought that up, and that moment, the launch of ChatGPT, something of a code red moment for Google. I'm quoting from the book now. At many points in his career, he had exhibited exquisite taste, seeing scientific trends ahead of his rivals. And yet when it came to language modeling, Hassabas taste buds went awry. And I'm paraphrasing now, and OpenAI took the lead, at least temporarily, with that ChatGPT model. It was launched in November 2022. What was that moment like for you personally, Dennis?
- [30:51]
Demis Hassabis
Well, look, actually the story is pretty complicated in the sense that the leading labs at the time, so us included, and actually both DeepMind and the Google Lab, Google Brain, had large language models. In fact, the Google Brain side of it had a product, a very early one, it was called Lambda in those days. And there'd been a conscious decision not to release that. And it was obviously, in retrospect, probably the wrong decision, but you can sort of see, and I wasn't involved in that because we were at the time just supposed to be working on research. Right. Because DeepMind, prior to that, prior to Google, DeepMind being formed, was we were just sort of running independently, mostly as a research organization. So. But I think that's an interesting Thing of where you can be sometimes too close to your own technology. So at building that technology, those early language models, they're very interesting, but they were very flawed. Even today's ones, very impressive as they are, have things that we all know when we use the chatbots they can't do very well, they make mistakes, they hallucinate. And those early ones were doing that a lot. So you can imagine, and I can understand that I wasn't involved in decision making but you can sort of think about Google's known for accurate information. That's what Google is for. Google search is for organizing the world's information. That's its bread and butter. That's what all of us come to it Google search for. So I think it's sort of doubly tricky to think about putting out a bot that 20% of the time gives you the wrong information. And so I think there's a clear issue there for a company like Google. Whereas obviously a startup, especially someone with someone like Sam leading it, it's like go for it, you only live once, yolo it and it's fine. Right. And the thing is to be fair to us as well, from a research perspective, no one knew how big that was going to get. ChatGPT including OpenAI because, and it's obvious, I mean Sam's admitted that, but it's also obvious from the name. ChatGPT is a programmer's name. You can you just like, you know they would have named it something catchier if they knew how big it was going to get. Right. It's not a catchy name. Although you know, it's obviously synonymous with AI now. So you can see and I think they've talked about they just thought it was going to be a research experiment and obviously they could take more risks with given that they were a startup with reputation and other things if it makes mistakes, so what? They don't have any existing products that rely on reliability. So the thing that caught everyone by surprise, I would say including them, was actually people finding use for even these early language models that even though they had all of these flaws, people still found obviously very creatively once millions of people use it. Oh, you can summarize things, can help with homework. So there were these sort of low hanging fruit that it didn't matter do some brainstorming. It didn't matter if it made some errors that weren't to do with information seeking. These were new use cases that hadn't existed before and obviously there was this turned out there was this huge demand for Even those kind of early use cases. And that's what caught everyone by surprise in terms of the technology being mature enough to come out of the lab and be put into the general public's hands.
- [34:08]
Tom McKenzie
And one of the upshots was that for Google, Google Brain and DeepMind merged. You now run essentially all AI at Google and obviously you've not just closed the gap, but on some metrics are leading in terms of some of these models that you are putting out with Gemini. That in itself, I mean we probably haven't had time to go into the intricacies of that merger, but that in itself, Sebastian, you were saying that merger of Google Brain and DeepMind stands out to you just in terms of a corporate merger as something of a standout moment.
- [34:42]
Sebastian Mallaby
Yeah, I mean, the conventional wisdom, if you ask a business school professor about mergers is that they're very difficult because there's always cultural differences, it's painful to smush two different organizations together and so forth. And in the case of Google Brain, which was doing AI research in Mountain View, and then DeepMind, which was doing AI research in London, you were putting together two teams which had eight hours of time difference between them, which doesn't make it easier. And they had a little bit of a record, which I hope you don't mind me saying this, there was a bit of a record of competition between the two because to some extent, if DeepMind got compute resources to run some tests and build a model, the people in Mountain View might see that as at the expense of their experiment they were trying to do so. There was, at least on the level of engineering gossip, quite a lot of back and forth about this bad blood. So here you are, you've got to do a merger, you've got eight time zones of difference, there's a bit of a record of competition and you're in the middle of this knockdown. Drag out you capitalist fight. Right, with OpenAI racing ahead and Google's got to catch up, what are the odds of this succeeding? I think most business school professors would have said, nah, not going to work. It worked like two and a half years later, at the end of 2025, DeepMind has an AI model, Gemini 3.0, which beats the ChatGPT rival on the leaderboards on these independent metrics of power and accuracy and so forth. So in two and a half years it was a total turnaround. It's like, I think this will go down not merely as a story about a frontier technology, but as a sort of case study in business mergers.
- [36:25]
Tom McKenzie
And this is helping us. This merger and being under Demis and the ambitions around AGI is helping you to get to that long term ambition that's driven you since you were a student, which is AGI and Superintelligent Machines. So I want to get to the title of the book, belatedly, maybe the Infinity. What is Demis? What is the Infinity Machine as you understand it?
- [36:47]
Demis Hassabis
Well, I mean, I think it's a great title for the book and I think actually Sebastian mentioned it in the opening, the way we look at problems, and it started really with AlphaGo and then onto AlphaFold is I think we found, and this is what AGI should be, a kind of general solution that can be applied to many, many problems. And the way I look at both GO and protein folding, which ostensibly look really different type of problems, is you're effectively navigating there's a needle in a haystack solution, but the haystack is unfathomably large. So for example, in Go, there are more possible board positions than there are atoms in the universe. So that means you can't brute force and enumerate every option. There's just not enough time in the universe. And the same with proteins, even more complicated. An average size protein has 10 to the 300 possible confirmations, possible shapes. So you can't find the right shape by looking at every single shape. You're going to have to do something much smarter. What we generically have built with our AI systems, at least applied to science, is using deep learning. Build neural networks, give them the data that you have on that particular domain. So Go games or protein structures that are known, and it's got to create a model of how that domain works. Then once you have that model, you can apply a search process, but in a smart way instead of having to search everything. So in a Go game, instead of you have 100 possible moves in a position, you don't search all of them, you just search. The model tells you which ones are the most likely, the most fruitful ones to look at. So you can spend your compute budget on very useful alternatives and only look at the ones that are likely to be fruitful. And so that way you can actually cut down these enormous search bases into something that's tractable in potentially in a very fast amount of time. And so this is a very, I sort of realize this is a very generic solution to many problems. If you can couch it in this way, this massive combinatorial space, you're trying to find the needle in the haystack and the final thing you need is an objective function. You need to have a clear goal you can hill climb against that the search is going to do and the model is going to help you find and in games and why we used games at the beginning apart from that I love games is that it's very convenient to train your AI systems on. In fact before go we used Atari games as the earliest measures of how our systems are doing because they have clear objectives, win the game or maximize the score. But in science there are also clear objectives. Minimize the energy in the system, for example is a very clear one that often you find in biology and chemistry and physics. So as long as you can couch the problem in that way and most of the time in a surprisingly large amount of times you can then these types of methods that we are using can be very very effective.
- [39:51]
Mia Sorrenti
Thanks for listening to Intelligence Squared. This episode was produced by Margarita Volpatto and it was edited by Mark Roberts. For ad free episodes and full length recordings. You can become a member@intelligencesquared.com membership and to join us at future live events. You can find our full program and buy tickets@intelligencesquared.com attend. You've been listening to Intelligence Squared. Thanks for joining us.
- [40:33]
Various Advertisers
You can't reason with the sun. Trust us. We've tried. This summer, it's time to put that angry ball of fire on mute. Columbia Columbia's OmniShadow technology is engineered to protect you from the sun's harsh rays that can burn and damage your skin. The sun is relentless, but so is our gear. Level up your summer@columbia.com to spend more time outside and less time slathering on aloe lotion. You're welcome Columbia. Engineered for whatever the right window treatments change everything. Your sleep, your privacy, the way every room looks and feels. @blinds.com, we've spent 30 years making it surprisingly simple to get exactly what your home needs. We've covered over 25 million windows and have 50,000 five star reviews to prove we deliver. Whether you DIY it or want a pro to handle everything from measure to install, we have you covered. Real design professionals, free samples, zero pressure right now. Get up to 45% off site wide plus get a free professional measure@blinds.com rules and restrictions apply. There's a new way to Sweetgreen Meet Wraps Handheld. Hearty and made for life on the move. With bold, chef crafted flavors, fresh Ingredients and over 40 grams of protein, they're built to satisfy without slowing you down. Try wraps today in the app or@order.sweetgreen.com Available at all participating locations. IBM a name synonymous with innovation for over 100 years. From typewriters to barcodes ATMs to PCs, we helped reach the moon. And we were in the AI game before it was cool. IBM has continually reinvented itself to do what it does best, solving the biggest challenges of the day through technology. Right now, we're putting hybrid, cloud and AI at the core of everything we do. We are champions of openness, infusing the greatest innovations from the community. We are pioneering responsible AI, leading governance, and partnering with other institutions to advocate AI ethics. We're helping solve some of the biggest problems facing humanity, from climate change and sustainability to development and human rights in the future. Our global network of research scientists continues to be at the cutting edge of innovation, whether it's in semiconductors or quantum computers. We're shaping the future one breakthrough at a time.