WEBVTT - TechStuff Redux: How Google DeepMind Accidentally Started the AI Race 0:00:00.120 --> 0:00:02.440 Hey, it's us here. I'm traveling this week and next 0:00:02.440 --> 0:00:04.680 week is the fourth of July, so we won't be 0:00:04.680 --> 0:00:07.960 publishing a Week in Tech episode for the next two Fridays, 0:00:08.119 --> 0:00:10.680 but you'll hear from all of our favorite tech contributors 0:00:10.760 --> 0:00:14.040 once again starting July tenth. In the meantime, I wanted 0:00:14.040 --> 0:00:17.080 to share an episode that demystified the man behind one 0:00:17.120 --> 0:00:20.480 of the most influential AI labs in the world, Google 0:00:20.520 --> 0:00:21.040 Deep Mind. 0:00:21.480 --> 0:00:47.000 Hope you enjoy. 0:00:38.159 --> 0:00:40.920 Welcome to Tech Stuff. I'm os Voloscan, and today we 0:00:41.000 --> 0:00:44.280 get the opportunity to go behind the curtain at Google's 0:00:44.360 --> 0:00:47.959 Deep Mind. For almost three years, in the upstairs room 0:00:48.000 --> 0:00:51.880 of a pub in North London, journalist Sebastian Malabi met 0:00:51.880 --> 0:00:55.959 regularly with the company's CEO and co founder, Demis Hasabis. 0:00:56.680 --> 0:01:02.480 They spoke about artificial intelligence, philosophy, neuroscience, motivation and consequence, 0:01:03.320 --> 0:01:06.559 all against the backdrop of an increasingly intense three way 0:01:06.680 --> 0:01:10.400 race between Open AI Andthropic and Google to win the 0:01:10.480 --> 0:01:14.520 race towards Agi so Ba see. Congratulations on your new book, 0:01:14.840 --> 0:01:17.240 The Infinity Machine and welcome to tex Stuff. 0:01:17.440 --> 0:01:18.600 Thank you ask great to be here. 0:01:19.120 --> 0:01:21.240 You begin the book with a quote from one of 0:01:21.280 --> 0:01:24.399 the scientists who work on the Manhattan Project who said, 0:01:24.880 --> 0:01:27.520 what we are creating now is a monster whose influence 0:01:27.600 --> 0:01:30.280 is going to change history. Yet it would be impossible 0:01:30.480 --> 0:01:33.080 not to see it through. The energy source which is 0:01:33.120 --> 0:01:36.959 now being made available will make scientists the most hated 0:01:37.319 --> 0:01:39.959 and the most wanted citizens of any country. 0:01:40.400 --> 0:01:43.919 We're reliving that now with AI. Agree. I mean, I 0:01:43.959 --> 0:01:47.880 began this project wanting to capture the tingling sensation of 0:01:48.000 --> 0:01:51.600 human beings like demisis ABIs creating the new version of 0:01:51.880 --> 0:01:56.320 atomic weapons. Right, This incredibly powerful AI technology that has 0:01:56.440 --> 0:02:00.480 enormous upsides could also be very, very dangerous. The prize 0:02:00.560 --> 0:02:02.040 was I didn't have to bring it up to them. 0:02:02.080 --> 0:02:03.480 They brought it up to me. I mean, it's so 0:02:03.560 --> 0:02:05.880 much on their minds. And so that's why I put 0:02:05.920 --> 0:02:08.679 this quotation about the Manhattan Project at the start of 0:02:08.720 --> 0:02:11.400 the book, because it's kind of the It sums up 0:02:11.800 --> 0:02:14.600 one of the main threads, which is this, you know, 0:02:14.800 --> 0:02:20.400 scientists can't resist inventing something which is exciting technically, and 0:02:20.440 --> 0:02:22.520 then they're going to be the most hated and most 0:02:22.520 --> 0:02:24.760 wanted people in the country. 0:02:25.160 --> 0:02:29.240 You described the question of motivation when it comes to 0:02:29.280 --> 0:02:33.320 Demis hanging in the air like the mushroom cloud over 0:02:33.360 --> 0:02:39.040 Los Alamos, very arresting visual image. Why that image and 0:02:39.440 --> 0:02:40.840 what did you want to stand in the end about 0:02:40.880 --> 0:02:41.440 his motivation? 0:02:42.000 --> 0:02:43.760 I think when you look at that picture of the 0:02:43.880 --> 0:02:48.760 mushroom cloud over Los Alamos, you're kind of thinking both wow, 0:02:49.200 --> 0:02:52.840 but also why why did human beings do this? You 0:02:52.880 --> 0:02:55.960 know it's so destructive? Why did you do it? And 0:02:56.120 --> 0:02:59.240 I guess you know. Part of the thread in my 0:02:59.280 --> 0:03:03.040 book is he had a series of ideas about how 0:03:03.080 --> 0:03:06.520 he could build AIE and make it safe for humanity 0:03:06.560 --> 0:03:10.200 and beneficial for humanity, and one by one these ideas 0:03:10.280 --> 0:03:13.560 become unraveled as they collide with reality. So the story 0:03:13.600 --> 0:03:16.799 of demisisabass in some ways, you know a story. I 0:03:16.840 --> 0:03:19.000 think there's two categories of screw up in the world. Right. 0:03:19.919 --> 0:03:24.440 Sometimes you get something where basically idiots are in charge, 0:03:24.800 --> 0:03:27.040 they don't understand what they're doing, and they make a 0:03:27.080 --> 0:03:30.920 humongous mistake Iran War for example. Right Then you have 0:03:30.960 --> 0:03:34.720 another much more interesting category of screw up, and that 0:03:34.880 --> 0:03:38.480 is where intelligent people know from the beginning exactly what 0:03:38.520 --> 0:03:40.680 they're doing. They can see the risks, they think they 0:03:40.680 --> 0:03:43.560 can manage them, but then forces which are larger than them. 0:03:44.120 --> 0:03:46.560 In this case with AI, it's a race dynamic between 0:03:46.600 --> 0:03:50.080 multiple labs and multiple countries take over and they can 0:03:50.120 --> 0:03:55.280 no longer control the technology that they've invented. And I 0:03:55.320 --> 0:03:58.320 think those episodes where you couldn't just switch out the 0:03:58.320 --> 0:04:02.920 individuals and have a better act come where the individual 0:04:03.000 --> 0:04:07.080 is good, sincerely, good, intelligent, thoughtful, has foresight, and yet 0:04:07.120 --> 0:04:08.840 you still end up in a bad place. That's what's 0:04:08.880 --> 0:04:12.720 really fascinating. When Demis sold his company deep Mind to 0:04:12.840 --> 0:04:16.160 Google in twenty fourteen, there was a condition which was 0:04:16.320 --> 0:04:18.880 this will never be used for weapons. Well, you know 0:04:19.000 --> 0:04:22.440 now it's twenty twenty six it is being used for weapons. 0:04:22.560 --> 0:04:25.600 And you know, so in time after time he tried 0:04:25.640 --> 0:04:29.080 to draw lines in the sand, and they've all been erased. 0:04:29.240 --> 0:04:32.680 It's harder and harder to say what these AI companies 0:04:32.760 --> 0:04:36.479 are today. I mean, for a moment last year, OpenAI 0:04:36.680 --> 0:04:39.880 was the most popular social video app company in the world, 0:04:39.920 --> 0:04:42.560 and now it doesn't do Sora anymore. But like, what 0:04:42.720 --> 0:04:44.880 is your working definition of what deep Mind is. 0:04:45.440 --> 0:04:48.280 I mean, it's a laboratory for the invention of machine intelligence, 0:04:48.440 --> 0:04:52.080 and machine intelligence is a very compacious thing. You know, 0:04:52.120 --> 0:04:57.240 it goes from text to video to images to a 0:04:57.320 --> 0:05:01.479 system like Alpha fold, which divined all the shapes of 0:05:01.520 --> 0:05:04.960 proteins in nature, and one demis the Nobel Price. So 0:05:05.080 --> 0:05:08.440 it's a huge field, and indeed the creation of it 0:05:08.480 --> 0:05:11.760 is a huge thing because you bring in experts in neuroscience, 0:05:11.880 --> 0:05:15.920 experts in chemistry, experts in physics, experts, computer science, ethical 0:05:16.240 --> 0:05:19.520 experts who can philosophize about the personality that an AI 0:05:19.560 --> 0:05:22.920 system should have. I mean, it's a very multidisciplinary thing, 0:05:22.960 --> 0:05:24.839 which is part of what makes the story fascinating. 0:05:24.960 --> 0:05:26.400 When did you first meet Demis Now? 0:05:26.480 --> 0:05:30.040 I first met him when I'd being interested in, you know, 0:05:30.080 --> 0:05:32.800 technology generally. My last book was about Silicon Valley and 0:05:32.880 --> 0:05:36.839 venture capital, and in the process of writing that book, 0:05:37.360 --> 0:05:40.040 I would go to tech conferences in Europe and there 0:05:40.080 --> 0:05:43.240 would be this sort of diminutive figure with a big, 0:05:43.279 --> 0:05:46.200 big smile and sort of a kind of boyish charm, 0:05:46.360 --> 0:05:49.600 really unassuming kind of guy with you know, a sort 0:05:49.640 --> 0:05:53.200 of round neck sweater and his hair falling forward in 0:05:53.240 --> 0:05:56.240 a fringe, and he would get up on the stage 0:05:56.560 --> 0:05:59.000 with a big grin and sort of just almost as 0:05:59.040 --> 0:06:01.360 if he was talking about how he was about to 0:06:01.600 --> 0:06:03.920 wash the dishes after lunch. You know, he would in 0:06:03.960 --> 0:06:07.400 a very plain spoken way, talk about well, when I 0:06:07.440 --> 0:06:09.360 was a child, I had two ambitions. One was to 0:06:09.400 --> 0:06:11.799 understand all of science and the other was to understand 0:06:11.800 --> 0:06:15.440 all of philosophy. So I resolved this dilemma by deciding 0:06:15.440 --> 0:06:17.559 to build AI, which would help me to understand both. 0:06:17.960 --> 0:06:22.320 And so he had this mind blowing mental reach combined 0:06:22.360 --> 0:06:26.880 with this totally approachable nextdoor friend kind of attitude. 0:06:27.080 --> 0:06:29.720 And when did you have the idea to pitch him 0:06:29.760 --> 0:06:32.159 on being a biographical subject for you? 0:06:32.440 --> 0:06:34.520 So after finishing my last book, The Power Law about 0:06:34.560 --> 0:06:36.520 Venture Capital, I was thinking, in you know, this is 0:06:36.520 --> 0:06:39.599 now mid twenty twenty two, what would be a good 0:06:39.680 --> 0:06:42.679 next subject. And because I had met Demis several times 0:06:42.680 --> 0:06:45.159 and I kind of followed what Deep Mind was up to. 0:06:45.240 --> 0:06:48.479 I knew about the protein folding system, I knew about 0:06:48.560 --> 0:06:51.599 Alpha Go, the go playing system before that, and so forth, 0:06:52.240 --> 0:06:54.719 and I had a sense that it would probably go 0:06:54.800 --> 0:06:57.760 from the fringe to the mainstream at some point in 0:06:57.760 --> 0:06:59.719 the next year or so. And then it took me 0:06:59.760 --> 0:07:03.320 if few months after that conversation inside my head to 0:07:03.560 --> 0:07:05.640 get my act together and listen to every podcast that 0:07:05.680 --> 0:07:07.920 Demis had ever done, read all his lectures, really think 0:07:07.960 --> 0:07:10.680 my way into his brain, and then go and see 0:07:10.720 --> 0:07:13.400 him to pitch him on giving me a ton of 0:07:13.440 --> 0:07:15.440 time because I need a lot of time with people 0:07:15.720 --> 0:07:17.920 if I'm going to write a book with him. And 0:07:19.280 --> 0:07:20.560 I said, look, you know, I want to write this 0:07:20.560 --> 0:07:23.120 book about you. And it seems to me Demists that 0:07:23.120 --> 0:07:26.520 you may not want a book about you, but you've 0:07:26.560 --> 0:07:29.080 said repeatedly in all of your lectures that AI will 0:07:29.120 --> 0:07:32.920 be the most important invention in all of human history. Demists. 0:07:33.160 --> 0:07:36.880 So that means if you're the creator of this AI, 0:07:37.320 --> 0:07:40.080 you must be one of the most important people in 0:07:40.160 --> 0:07:42.440 human history. And if that's the case, you don't have 0:07:42.440 --> 0:07:45.600 a choice. Somebody's going to write a book, right And furthermore, 0:07:45.600 --> 0:07:48.640 you should welcome this because if you're going to invent 0:07:48.680 --> 0:07:52.920 a technology that is going to disrupt people's lives so thoroughly, 0:07:53.040 --> 0:07:55.680 you know your job will be different, how you raise 0:07:55.720 --> 0:07:58.000 your children will be different, how you think yourself as 0:07:58.040 --> 0:08:00.840 a human will be different because you now have this 0:08:00.920 --> 0:08:04.720 different source of intelligence competing with you. You can't disrupt 0:08:04.720 --> 0:08:07.480 people from head to toe and then not explain them 0:08:07.680 --> 0:08:10.680 why you did it. You need to explain your motives, right, 0:08:11.360 --> 0:08:14.840 and that's that's the project I'm proposing to you. And 0:08:15.000 --> 0:08:17.120 he thought about it, and he seemed well disposed to this, 0:08:17.440 --> 0:08:21.280 and then one week later Chatchipt came out. 0:08:21.440 --> 0:08:22.360 Oh my goodness, and. 0:08:22.360 --> 0:08:24.720 My expectation of the technology going from the fringe to 0:08:24.760 --> 0:08:27.000 the mainstream happened a whole lot quicker than I expected. 0:08:27.920 --> 0:08:30.520 And I mean, obviously a lot of the book of 0:08:30.800 --> 0:08:33.559 your interest is in the technology, how it might change 0:08:33.559 --> 0:08:37.640 the world, the kind of financing and deal making shenanigans 0:08:37.640 --> 0:08:40.480 that made deep mind in many ways what it is today. 0:08:40.520 --> 0:08:45.400 But what other personal side of an extraordinary biographical portrait 0:08:45.559 --> 0:08:51.280 to very specific parents, a prodigious talent for chess, which 0:08:51.280 --> 0:08:52.960 he then gave up because he thought it wasn't in 0:08:52.960 --> 0:08:55.439 some sense consequential enough. I mean, did you know when 0:08:55.480 --> 0:08:57.319 you got it, you knew that AI was the next thing, 0:08:57.400 --> 0:08:59.400 and you know there he was. But did you know 0:08:59.400 --> 0:09:02.559 what an extra personal story he had before you really 0:09:02.600 --> 0:09:03.400 got into it with him. 0:09:03.520 --> 0:09:05.880 No? I didn't, And in fact, I remember very clearly 0:09:07.000 --> 0:09:11.080 too early experiences in the first discussions. You know, one 0:09:11.280 --> 0:09:14.400 was I was going to have this dinner I told 0:09:14.400 --> 0:09:17.000 you about and he told me to read a book 0:09:17.040 --> 0:09:19.800 before I came to the dinner and the book was 0:09:19.920 --> 0:09:22.320 Enders a game. Now, this is a science fiction story 0:09:22.960 --> 0:09:27.160 about a sort of diminutive boy genius hero who has 0:09:27.200 --> 0:09:31.720 to save planet from invading space aliens. And he's at 0:09:31.760 --> 0:09:33.840 the end of the book he saves all of humanity 0:09:33.840 --> 0:09:36.800 from space aliens. And Demis said to me, well, I 0:09:36.800 --> 0:09:38.880 wanted you to read this book because I really identify 0:09:39.000 --> 0:09:42.680 with that character Ender. And I'm thinking, wait, so you're 0:09:42.720 --> 0:09:44.920 telling me you're the savior of humanity. I mean, even 0:09:44.960 --> 0:09:47.600 if you think that Demis, maybe you shouldn't be announcing 0:09:47.600 --> 0:09:49.040 it to the person who's about to write a book 0:09:49.040 --> 0:09:51.400 about you. I mean, surely that's too messianic, to over 0:09:51.400 --> 0:09:54.640 the top, too ridiculous. But he's right out there with it. 0:09:54.679 --> 0:09:56.400 You mean, that is how he thinks, and he's not 0:09:56.440 --> 0:09:59.640 ashamed to tell you. And so that was pretty extraordinary. 0:09:59.679 --> 0:10:02.320 And then and the second thing was I went to 0:10:02.320 --> 0:10:05.319 see Shane leg his Scientific co founder, and he told 0:10:05.360 --> 0:10:07.360 me the story about how. I said, you know, what 0:10:07.400 --> 0:10:09.880 was it light to work with Demis? And he said, well, 0:10:10.120 --> 0:10:13.439 you know, Demis has crazy determination. I said, well, what 0:10:13.520 --> 0:10:15.960 do you mean He said, well, you know, one day, 0:10:17.120 --> 0:10:19.439 according to Demis, his dad said to him during the 0:10:19.559 --> 0:10:22.559 chess period of his life. Listen, you're going to go 0:10:22.600 --> 0:10:24.880 play chess today. You know you just have to try 0:10:24.920 --> 0:10:27.400 your best. Now, when I say that to my son, 0:10:27.480 --> 0:10:28.920 I mean, you know, it's fine to lose so long 0:10:28.920 --> 0:10:31.200 as you try your best. The way that Demis apparently 0:10:31.240 --> 0:10:34.560 interpreted it, according to Shane, was you have to try 0:10:34.600 --> 0:10:37.599 your absolute, absolute, absolute best. And it's like running a 0:10:37.679 --> 0:10:39.920 race and at the end of the marathon you fall 0:10:39.960 --> 0:10:43.120 over the tape and you're on the ground and you 0:10:43.120 --> 0:10:45.440 have to be taken to hospital because you're almost dead. 0:10:45.640 --> 0:10:48.160 And if you haven't been taken to hospital, it means 0:10:48.200 --> 0:10:51.439 you didn't try hard enough. That is what try your 0:10:51.440 --> 0:10:55.920 best meant to Demis aged about ten or twelve, And 0:10:56.000 --> 0:10:58.240 I went to see Demis the next time and I 0:10:58.320 --> 0:11:01.120 replayed this back to him and said, is that really true? 0:11:01.160 --> 0:11:03.199 Is that how you interpreted? He said, oh, yeah, absolutely. 0:11:03.720 --> 0:11:05.960 You know you have to give it every single drop 0:11:06.280 --> 0:11:06.880 all the time. 0:11:07.200 --> 0:11:10.520 And there's an amazing moment where Demis describes talking about 0:11:10.640 --> 0:11:15.560 hearing nature or science screaming at him and him struggling 0:11:15.640 --> 0:11:16.800 to hear and to understand. 0:11:17.200 --> 0:11:20.600 Yeah, that was the most extreme expression of his desire 0:11:20.679 --> 0:11:23.720 to invent AI. So one day I was with him 0:11:23.720 --> 0:11:26.720 in Hampstead Heath, which is a park in North London. 0:11:26.840 --> 0:11:28.520 Not in the pub, Not in the pub. This time 0:11:29.200 --> 0:11:30.560 it was a nice day, so we went for this 0:11:30.640 --> 0:11:33.280 cafe instead, and you know, there he was. There was 0:11:33.360 --> 0:11:35.400 kind of a classic English scene. There was somebody in 0:11:35.400 --> 0:11:37.600 front of me who was on his cell phone. You're 0:11:37.600 --> 0:11:39.679 doing some sort of sales job, and two women behind 0:11:39.720 --> 0:11:43.200 me talking about their friend who had a medical incident 0:11:43.240 --> 0:11:44.840 and had to get to a hospital. So all these 0:11:45.080 --> 0:11:47.760 quididian noises in the background, and there is demis a 0:11:47.800 --> 0:11:50.000 sabists looking at me, talking about the creation of this 0:11:50.080 --> 0:11:52.920 godlike machine and saying that when he's up at two 0:11:52.960 --> 0:11:55.880 in the morning at his desk at home thinking about this, 0:11:56.280 --> 0:12:00.280 he can sort of feel reality summoning him, streaming at him, 0:12:00.400 --> 0:12:03.439 understand me, understand me, And you know he would then 0:12:03.480 --> 0:12:07.280 slam the table and say, look, Sebastian, this table it's 0:12:07.320 --> 0:12:10.200 made of atoms, buzzing around with electrons. Why should it 0:12:10.200 --> 0:12:12.600 be solid? Why should that laptop you've got there, why 0:12:12.640 --> 0:12:15.040 should it you know, pieces of sand and metal? How 0:12:15.040 --> 0:12:17.439 could that turn into something which can think? I mean, 0:12:17.480 --> 0:12:20.160 what's going on? Here. There must be some intelligent force 0:12:20.280 --> 0:12:23.440 designing all these things. And so he kind of basically 0:12:23.480 --> 0:12:27.560 told me that inventing AI and understanding the universe is 0:12:27.600 --> 0:12:29.920 like getting closer to what he thinks of as God. 0:12:30.480 --> 0:12:31.560 So he's a religious man. 0:12:31.679 --> 0:12:33.640 I don't know if he would agree with religious because 0:12:33.640 --> 0:12:37.160 he doesn't go to organized religious services, but he's spiritual. 0:12:37.240 --> 0:12:38.640 I would say, interesting. 0:12:39.440 --> 0:12:41.439 And I mean that scene you described could be a 0:12:41.440 --> 0:12:45.360 scene from Oppenheimer, right, I mean, it's so cinematic. Did 0:12:45.400 --> 0:12:47.400 you ever think is he doing this for me? Or 0:12:47.480 --> 0:12:51.319 is he crazy? Or is it just absolutely captivating? The 0:12:51.400 --> 0:12:53.640 energy in the sense of purpose that he brings to this. 0:12:53.840 --> 0:12:57.280 He just exudes both energy and intelligence but also storytelling 0:12:57.360 --> 0:13:00.080 natural talent. It's just amazing. I mean, you know, one 0:13:00.160 --> 0:13:03.080 time I asked him about his first office in London 0:13:03.120 --> 0:13:05.959 and Russell Square, which is a sort of storied square, 0:13:06.000 --> 0:13:08.400 you know, near the British Museum and so forth. And 0:13:08.559 --> 0:13:10.679 you know, normally, as a writer, you are somebody to 0:13:10.720 --> 0:13:15.480 recapture the emotion of opening their first office fifteen years ago. 0:13:16.120 --> 0:13:17.959 It's fifteen years ago. They're going to say, oh, yeah, 0:13:17.960 --> 0:13:19.600 it was cool. You know, That's all you'll get out 0:13:19.640 --> 0:13:22.920 of them. But demis just flows with stories. So he said, well, 0:13:22.960 --> 0:13:25.120 you know, I was in the attic that's where the 0:13:25.120 --> 0:13:26.640 office was, and of course you had to come down 0:13:26.679 --> 0:13:28.440 the stairs. They were all Rickety's. So I came down 0:13:28.600 --> 0:13:30.679 ding ding ding ding ding, bang bang bang, And then 0:13:30.720 --> 0:13:32.200 I come out on the square and there's these beautiful 0:13:32.200 --> 0:13:34.040 trees in front of me. Blond to the right. If 0:13:34.080 --> 0:13:36.040 you just go three doors down, Sebastian, that's where you 0:13:36.080 --> 0:13:41.080 see the London Mathematical Society where Turing invented the origins 0:13:41.120 --> 0:13:44.040 of computer science, which we are now completing. And then 0:13:44.320 --> 0:13:47.160 if you go beyond that to the level crossing black white, 0:13:47.200 --> 0:13:50.360 black white crossing the street, the pedestrian crossing, that is 0:13:50.440 --> 0:13:54.800 where the Hungarian nuclear scientist Zilad had the idea for 0:13:54.840 --> 0:13:57.560 a nuclear chain reaction back in the nineteen thirties, which 0:13:57.640 --> 0:14:00.320 led to the atom bomb. And of course we are 0:14:00.360 --> 0:14:02.640 now creating the equivalent of the atom bomb with Ai. 0:14:03.120 --> 0:14:06.960 What a subject, Yeah, I mean, he is such a storyteller. 0:14:06.960 --> 0:14:08.480 And I heard that he has a sense of humor 0:14:08.559 --> 0:14:10.280 or perhaps a sense of humour about himself in some 0:14:10.320 --> 0:14:12.760 ways as well. Didn't he say when he lost the 0:14:13.600 --> 0:14:16.480 table football competition? The office that his soul was on fire. 0:14:16.720 --> 0:14:19.080 He did say that, yes, you know, one can mock 0:14:19.160 --> 0:14:23.000 him for being too competitive and taking trivial things like 0:14:23.040 --> 0:14:25.280 table football seriously, but he actually really feels it. 0:14:25.520 --> 0:14:28.760 We talked about the sort of mushroom cloud of motivation. 0:14:29.440 --> 0:14:30.840 One of the things that doesn't seem to be so 0:14:31.040 --> 0:14:34.040 motivating to him is money. I mean, there's a story 0:14:34.080 --> 0:14:37.400 about the offer as an eighteen year old of a 0:14:37.400 --> 0:14:41.880 half million pounds to join game development studio right, which 0:14:41.880 --> 0:14:44.400 he turned out right, despite coming from me. I know 0:14:44.440 --> 0:14:47.080 his mother had gone through homelessness her in her youth. 0:14:47.120 --> 0:14:49.520 I mean, was that a hard decision for him? Why 0:14:49.520 --> 0:14:50.120 did he make it? 0:14:50.360 --> 0:14:52.720 He said it was completely easy in today's money. It 0:14:52.840 --> 0:14:55.040 was well over a million dollars that he was being offered. 0:14:55.440 --> 0:14:57.680 He was eighteen. As you say, his parents were not rich. 0:14:58.200 --> 0:15:01.360 I mean, you know, any self respecting sort of Stanford 0:15:01.440 --> 0:15:04.440 character would at this point of fall, you know, taking 0:15:04.440 --> 0:15:07.280 the money dropped out of Stanford and you know, written 0:15:07.320 --> 0:15:10.240 off into the sunset with a loot. No, Demis is different. 0:15:10.280 --> 0:15:13.600 Demis wanted to understand science. That was his primary motivation. 0:15:13.640 --> 0:15:15.400 That's what he's up when he's up at two o'clock 0:15:15.440 --> 0:15:17.840 in the morning. He's thinking, how do I understand nature? 0:15:18.240 --> 0:15:20.760 And so he turned down the cash to go and 0:15:20.800 --> 0:15:22.240 study computer science instead. 0:15:22.480 --> 0:15:26.360 Fast forward a few years and he meets Peter Teel, 0:15:27.480 --> 0:15:31.480 who gives him a A plus for science fiction and 0:15:31.520 --> 0:15:34.680 an F for business model, but nonetheless a size skiins 0:15:34.680 --> 0:15:35.080 some money. 0:15:36.320 --> 0:15:38.560 Well, actually there's a fraudulent slip there. You said an 0:15:38.560 --> 0:15:40.520 A plus for science fiction. I think you met an 0:15:40.520 --> 0:15:41.840 A plus for science. 0:15:42.160 --> 0:15:45.160 Science fiction, but science fiction. 0:15:45.360 --> 0:15:47.720 Maybe that would have been better, because in fact Demis 0:15:47.800 --> 0:15:50.960 was spinning this vision and this is twenty ten, right. 0:15:51.000 --> 0:15:54.080 He was saying, I'm going to invent very powerful AI. 0:15:54.480 --> 0:15:58.280 This is at a time when AI literally couldn't recognize 0:15:58.280 --> 0:16:01.960 the photograph of a cat. Nothing was working, and you 0:16:01.960 --> 0:16:03.440 have this character coming and say, oh, I'm going to 0:16:03.440 --> 0:16:06.680 create artificial general intelligence. It was nuts. So it kind 0:16:06.720 --> 0:16:07.640 of was science fiction. 0:16:08.320 --> 0:16:11.720 What was his entree to the world of technology investors? 0:16:11.720 --> 0:16:13.920 And when did deep Mind actually start as a company. 0:16:14.040 --> 0:16:16.400 Deep Mind started in twenty ten, having raised the money 0:16:16.400 --> 0:16:20.000 from Peter Teel. The entree is very interesting because in 0:16:20.000 --> 0:16:23.000 fact what happened was, you know, Demis had done a 0:16:23.040 --> 0:16:26.080 small games company before, and he made some money. Wasn't 0:16:26.080 --> 0:16:29.440 a terrific success, but nonetheless it wasn't total failure. And 0:16:29.480 --> 0:16:31.400 he went back to the same investors. They all said, 0:16:31.400 --> 0:16:34.240 you must be joking. There's no product if you're doing 0:16:34.280 --> 0:16:36.680 AI and not putting money into that. So then he 0:16:36.720 --> 0:16:39.800 had to think again, and his entree into the world 0:16:39.800 --> 0:16:43.400 of Peter Teel and what's called the Singularity summits, where 0:16:43.440 --> 0:16:46.960 all these very early believers in AI would gather, people 0:16:47.000 --> 0:16:50.320 like Ray Kurtzweil, and they would dream about a future 0:16:50.640 --> 0:16:54.120 of an AI that totally did not exist. And when 0:16:54.160 --> 0:16:56.120 they got up on the stage, actually they did often 0:16:56.320 --> 0:17:01.400 draw more on science fiction novels than on science when 0:17:01.440 --> 0:17:04.159 they were kind of imagining a future with AI. And 0:17:04.320 --> 0:17:08.480 so in this strange cauldron of mythology and reality with 0:17:08.600 --> 0:17:12.760 all kinds of weirdos trotting about, demis a Sabots, who 0:17:12.760 --> 0:17:15.000 by this point has a computer science degree and a 0:17:15.000 --> 0:17:18.280 PhD in neuroscience as a proper scientist, shows up and 0:17:18.320 --> 0:17:20.679 he's asked by a journalist what do you think of 0:17:21.040 --> 0:17:24.720 the Singularity conference? Are you a Singularitarian? And he says 0:17:25.400 --> 0:17:28.040 it's a bit Californian for me, and he's, oh, you 0:17:28.040 --> 0:17:30.720 could sort of feel the kind of anxiety of being 0:17:30.760 --> 0:17:32.960 seen in this crowd. But that's why you had to 0:17:33.000 --> 0:17:35.360 go to meet Peter tele And then when he met 0:17:35.400 --> 0:17:37.879 Peter Til, he had this clever trick. Peter Teel is 0:17:37.880 --> 0:17:40.600 a chess player. Demis is a chess player. So rather 0:17:40.640 --> 0:17:43.840 than pitch Peter tele On some idea about a company, 0:17:43.840 --> 0:17:45.640 and he said, well, I think the interesting thing about 0:17:45.760 --> 0:17:50.080 chess is that the knight and the Bishop are supremely 0:17:50.080 --> 0:17:52.719 well balanced, and it's in that tension between those two 0:17:52.760 --> 0:17:55.160 pieces that much of the joy of the game resides. 0:17:55.720 --> 0:17:58.080 So p t is like, WHOA, that's a conversation I 0:17:58.119 --> 0:18:00.760 want to pursue. And so that got him and got 0:18:00.760 --> 0:18:03.640 Demssus Habits an invitation to Peter Teel's house the next day, 0:18:04.200 --> 0:18:06.080 and then that's when he pitched him on Deep Mind 0:18:06.119 --> 0:18:08.600 and got the money he needed to start the company. 0:18:08.920 --> 0:18:14.840 And then Fast forty twenty thirteen and which the excerpt 0:18:14.840 --> 0:18:16.600 in the Wall Street Journal of your book tells the 0:18:16.640 --> 0:18:20.399 story of a birthday party for Elon Musk, replete with 0:18:20.440 --> 0:18:24.040 all kinds of costumes and strange things and fake battlements. 0:18:24.080 --> 0:18:28.920 But this is, perhaps apart from the founding, the most 0:18:28.960 --> 0:18:31.879 crucial moment in Deep Minds Genesis as a company. 0:18:31.960 --> 0:18:34.640 Right, yeah, that's right. So you know, by this point, 0:18:35.240 --> 0:18:39.440 Demis had raised three rounds of venture capital, including from 0:18:39.440 --> 0:18:42.000 Elon Musk, and you know, there are various people could 0:18:42.040 --> 0:18:44.480 come in, but it was a total pain in the neck. 0:18:44.520 --> 0:18:47.879 He hated it. You know, he would sometimes have this expression, 0:18:48.119 --> 0:18:50.320 I don't want this part of my brain to expand. 0:18:50.840 --> 0:18:53.840 He wanted to be doing science, and so what he 0:18:53.920 --> 0:18:57.439 wanted was to be liberated from this hamster wheel of fundraising. 0:18:58.280 --> 0:19:01.399 And along at this party, this birthday party that Enol 0:19:01.480 --> 0:19:05.960 Musk had, along comes Larry Page from Google who's also there, 0:19:06.680 --> 0:19:08.840 and says, let's go for a walk, and they walk 0:19:08.880 --> 0:19:12.679 around the castle grounds, and in this bizarre setting, Larry 0:19:12.680 --> 0:19:15.200 Page says to him, well, you know, you could spend 0:19:15.200 --> 0:19:18.480 your career building another company like Google. That's fine, but 0:19:18.560 --> 0:19:21.480 if you really want to do science, just join Google 0:19:21.760 --> 0:19:25.479 and we'll give you the resources, use our platform, and 0:19:25.520 --> 0:19:28.080 you'll be able to do what you really love. And 0:19:28.359 --> 0:19:31.880 Demis not only agreed with that pitch in the sense 0:19:31.920 --> 0:19:34.399 that yes, he preferred to do science than to be 0:19:34.440 --> 0:19:37.840 a billionaire, but he felt that Larry Page himself would 0:19:37.840 --> 0:19:41.560 have accepted that pitch, that Larry Page cared about science. 0:19:41.560 --> 0:19:43.840 He could have been a standard professor of computer science. 0:19:44.160 --> 0:19:47.840 So Demis really identified with Larry Page, and that was 0:19:47.840 --> 0:19:48.840 why he sold to Google. 0:19:48.920 --> 0:19:52.399 And Paige had his eye on Demis or this was impulsive. 0:19:52.680 --> 0:19:55.479 Had he planned out his chess game for this party 0:19:55.600 --> 0:19:57.040 like Demis had three years before. 0:19:57.080 --> 0:19:59.320 He had totally planned the chess game. He'd been thinking 0:19:59.359 --> 0:20:03.720 for a while about buying up nascent AI companies, and 0:20:03.760 --> 0:20:07.359 he'd bought the boutique founded by the Toronto professor Jeffrey 0:20:07.400 --> 0:20:11.360 Hinton together with Iliasaskeva and one other person. Uh, and 0:20:11.680 --> 0:20:14.200 so he was in a buying mode. 0:20:14.880 --> 0:20:16.719 It came into that was twenty twelve, right the image 0:20:16.720 --> 0:20:17.880 net team exactly. 0:20:17.880 --> 0:20:19.960 He bought the image net team, and then the next 0:20:20.000 --> 0:20:23.800