# The Infinity Machine  ## Metadata - Author: [[Sebastian Mallaby]] - Full Title: The Infinity Machine - Category: #books ## Highlights - Growing up in North London, he had decided that two fields of inquiry stood out: physics and neuroscience. Physics explains the external world, from the behavior of particles to the functioning of the universe. Neuroscience explains the internal world—the neurons and synapses and electrical pulses that constitute intelligence. Later, at some point in his twenties, Hassabis had concluded that neuroscience was the more important of the two: The internal trumped the external. Intelligence is fundamental; it is the root of all else. It is the mechanism through which humans perceive reality. ([Location 94](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=94)) - To Descartes’s dictum, “I think, therefore I am,” perhaps we should add: I imagine, therefore I am; I hypothesize, therefore I am; I invent, therefore I am. The urge to invent lies deep within us. ([Location 136](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=136)) - “I am in the camp that is hopeless,” Hinton informed Bostrom. “In that you think it will not be a cause for good?” Bostrom inquired. “I think political systems will use it to terrorize people,” Hinton answered. “Then why are you doing the research?” Bostrom asked. “I could give you the usual arguments,” Hinton replied. “But the truth is that the prospect of discovery is too sweet.”[4] ([Location 148](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=148)) - During this process, Hassabis revealed himself as an extraordinary consumer and teller of stories—his outlook is shaped by novels and movies, and his gifts as a leader are bound up with his genius for narrating his experiences. ([Location 177](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=177)) - The true reason to build artificial intelligence, Hassabis was now saying, went beyond Kant and Feynman. The goal was to draw closer to what might be called God—to the intelligence that may presumably have designed everything around us. “I am first and foremost a scientist,” Hassabis began. “My goal is to understand nature. “But doing science is, sort of, like reading the mind of God. Understanding the deep mystery of the universe is my religion, kind of. “We humans, we have these faculties. The world is understandable. But why should it be that way? I think there is a reason. “Computers are just bits of sand and copper,” Hassabis continued, now sounding more urgent. “Why should these combine to do anything? I mean, it’s absurd! The electrons move around and then that creates an AI system that can defeat a Go master? Why should that be possible? “This table, Sebastian!” Hassabis rapped his palm on it for emphasis. “Why should it be solid? “This is beyond evolutionary coincidence. We can build electron microscopes and interrogate reality down to the most minute level. We can build systems that detect black holes colliding more than a billion years ago. I mean, what is this? What the hell is going on here?” There was a pause, but Hassabis was not yet finished. “I sit at my desk at two a.m., and I feel like reality is staring at me, screaming at me. “Literally, screaming at me. Trying to tell me something if I could just listen hard enough. “That’s how I feel every day. So, you can see why I’m trying to build AI. I’ve felt that since I was very young: that there’s a deep, deep mystery about what’s going on here. “You can frame it how you want. You can call this God’s design, or you can say it’s just nature. I’m open-minded about the description, and I don’t know what the answers will turn out to be. But at the moment we don’t really know what time is, or gravity is, or any of these things. So there is a mystery waiting to be solved, and it encompasses just about everything. “I would like to understand before I croak. I would like to understand, and then I’m perfectly fine to shuffle off my mortal coil.” ([Location 186](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=186)) - Demis Hassabis discovered a work of science fiction that made sense of who he really was. The book, called Ender’s Game, by Orson Scott Card, tells the story of a diminutive boy genius who is taken from his family and sent off to a space station. There, at an intergalactic battle school, Ender is manipulated by adults, bullied by classmates, and put through extreme mental testing, all to discover whether he can shoulder responsibility for the survival of the human race. By dint of grit and talent, Ender rises to the challenge. At the climax of the novel, he outwits an army of alien invaders, destroying their armada and saving planet Earth, though the question of whether he committed genocide in the process hangs over the outcome. Hassabis was around thirty years old when he discovered Ender, and he was so taken with the story that he asked his wife to read it. She told him she felt sorry for the central character—a boy deprived of childhood and harnessed to a mission chosen for him by adults. But Hassabis identified powerfully with Ender. He, too, had been a diminutive boy genius, socially isolated by his own prodigious talent. He, too, had undergone extreme mental testing, and was consumed by a desire to make his mark on the universe; one of his ambitions was to surpass his scientific heroes, Newton and Einstein, and “understand the fabric of reality itself.” The fable of Ender—a gifted, bullied boy who saves all humanity—tapped into Hassabis’s deepest preoccupations, even if (especially if!) the savior had been required to pay an immense personal price. ([Location 233](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=233)) - But then he continued, weighing his words deliberately. “Demis has an extraordinary level of determination. Unlike pretty much anybody. Astonishing, incredible determination. That’s his most defining characteristic. Just unbelievable determination.” “What do you mean?” “He works, sleeps, eats, breathes the mission, twenty-four hours a day. To a degree that I just haven’t seen with other people.” “No hobbies?” “Football. Big fan of Liverpool. But other than that, it’s the mission.” “And that was evident even back when you met him, more than a decade ago?” “Always,” Legg answered. ([Location 255](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=255)) - “Demis tells a story about his father saying that whether you win or lose, the really important thing is that you try your best. And Demis says he took that very literally. As in, absolutely try the absolute, absolute, absolute best you can possibly do, pretty much to the point of breaking yourself. “That’s how he is, twenty-four seven.” I nodded, kept eye contact, and hoped Legg would continue. “I don’t think his father meant his comment in quite that literal sense,” Legg reflected. “Like, ‘try your best’ wasn’t supposed to mean ‘try literally to the point of destroying yourself, go absolutely, completely 100 percent.’ But that’s how Demis understood it. “There is no 50 percent mode in Demis. There is not even a 99 percent mode in Demis. There is only 100 percent.”[1] ([Location 262](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=262)) - In contrast with his family circumstances, the young Demis’s talents were not at all modest. When he was four, he climbed up on a chair to watch his father play chess against his uncle. Within a few weeks, he had mastered the game well enough to defeat adults. At five, he began competing in tournaments, sitting on a telephone book on top of two stacked chairs so that he could get his head above the table, and frequently beating older kids. ([Location 277](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=277)) - “Your son is the best six-year-old I’ve ever seen,” Barden said. “What are you going to do when someone tells you that?” Hassabis reflected. “My parents were fairly normal people living fairly normal lives. And a renowned expert is telling you this.” Demis’s father responded to Barden’s message as though instruction had been handed down from God. For the next half dozen years, weekend after weekend, he bundled his young prodigy into the family’s red VW van and drove him off to tournaments in shabby, far-flung church halls, leaving his wife with the two younger children and her various jobs. Sometimes the father-son duo spent the night in sleeping bags laid out over the engine at the back of the VW; other times they found a cheap hostel and shared a bunk bed. If Demis won the tournament, the prize money would cover the hostel fees, but his mother still fretted about the cost of the travel. “She grew up in absolute poverty,” Hassabis said later. “I imagine my parents had a lot of arguments about money, because we didn’t have much.” ([Location 286](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=286)) - Demis’s father was taking progressively more time away from his work and his music to keep driving him to tournaments, which only redoubled the pressure on his son. When the boy had a bad game, the father would erupt. “There was one time, I was a rook up and then lost horribly, and my dad went mental,” Hassabis remembered.[6] “He was screaming, ‘How could you have done this? This is unbelievable. How could you have just thrown this away?’ “It was just awful. We were out in some hostel, and he was going on about this, screaming. And this used to be a fairly regular occurrence with my dad. “So I said to him, ‘This is ridiculous. I obviously tried my best. I’m not intentionally losing.’ And that was that. I wasn’t going to take it anymore. That was the last time I remember him screaming at me, whereas he used to all the time before. ([Location 301](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=301)) - “the slightly warped way I took that was, how do you know you’ve done your best? “The only way I could know is basically if I pushed myself to the point just before death. Because that is literally when you have done your best. If you die—by die, I mean burn out or something—then you’ve slightly overdone it. “It’s like running a marathon. You have to basically fall over the line. And then ideally you should be hospitalized, but not dead. That’s when you can say you’ve done your best. But if you’ve got any energy left, you’re still standing, maybe you could have tried harder? “That’s how I took it. I must have been about nine or ten.” ([Location 317](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=317)) - For the rest of the day, Hassabis felt sick to his stomach. But the next morning he experienced an epiphany. That tournament hall near Liechtenstein had been packed with brilliant brains, dueling over black and white squares until stamina was drained to nothing. Surely that immense collective mental effort should have been harnessed to some higher cause—science, say, or medicine? “I thought we were wasting our minds,” Hassabis said later.[8] ([Location 338](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=338)) - to the surprise of many in the chess fraternity, Hassabis also began to compete in other mind games: bridge and backgammon, Diplomacy and draughts, as well as the Japanese chess variant shogi. His versatility was even more striking than his prowess at chess: After university, he went on to win a record string of gold medals at the five-game international Mind Sports Olympiad. “What Demis did with shogi was really impressive,” a chess veteran reflected. “Another complicated game and he got to the top of the English rankings,” said another.[10] But Hassabis could scarcely imagine why he wouldn’t sample other challenges. The world offered so many enthralling ways to test his mental acumen. ([Location 349](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=349)) - as Hassabis matured in his teen years, he began to flourish socially. The switch flipped when he was fifteen and his school arranged a special class for a handful of elite math pupils. For the first time, Hassabis found himself surrounded by similarly curious, driven kids, and he befriended all of them. The students came from a variety of backgrounds: a Nigerian whose father was a diplomat; a boy of Indian descent; two Jewish kids; and one Cypriot-Singaporean-Bohemian chess prodigy. All were united in their enthusiasm for the class. Not only were they learning math, they were discovering that joyful, abstract mental play could be a way of bonding with your peers, not a route to isolation. ([Location 361](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=361)) - At twelve, he bought a much more powerful device, a Commodore Amiga 500. His dad took him to Foyles, a labyrinthine bookshop in the heart of London that boasted thirty miles of weathered shelves, and Hassabis discovered a slim volume called The Chess Computer Handbook, by the Scottish international master David Levy. The marriage of computing and chess united Hassabis’s two worlds. He bought Levy’s handbook and read it in one sitting. Levy introduced Hassabis to the themes that would animate his lifelong quest to build artificial intelligence. To show how chess programming worked, Levy invoked the information theorist Claude Shannon, who would become another of Hassabis’s intellectual heroes. In 1950, Shannon had published a paper, “Programming a Computer for Playing Chess,” arguing that, while chess programs were of no practical importance, mastery of complex games might “act as a wedge in attacking other problems of a similar nature and of greater significance.” These similar but more significant problems might be as various as translation, military strategy, and the generation of music. Chess programming was merely the first step toward what Shannon termed “a modern general-purpose computer.”[13] When Shannon wrote those words, no such general-purpose computer was in sight. “I started writing a program for a machine that did not yet exist, using a set of computer instructions that I dreamed up,” he confessed cheerfully.[14] But, blessed with a rare gift for theorizing the future, Shannon proceeded to describe the difference between a “numerical computer” and a “general” one. A numerical computer was basically a calculator: It followed rigid programs and tackled questions that had clear right-or-wrong answers. In contrast, a general computer could make sense of subjects that demanded more than mere logic: It could assess a chess position, or grasp linguistic nuance. To grapple with this sort of material, the program would have to apply “general principles, something of the nature of judgment, and considerable trial and error, rather than a strict, unalterable computing process.” A general computer would not be merely deductive. It would consider examples, try things out, and make sense of the world around it. Levy’s book, written a third of a century later, surveyed the relationship between computers and chess as it stood in the early 1980s. By this point, chess programs were starting to exhibit the features that Shannon had imagined, and their strengths and weaknesses shed light on the nature of intelligence. Silicon transmits electric signals much faster than the human brain, so computers could rapidly calculate several moves ahead; alacrity enabled them to defeat top humans at speed chess.[15] However, computers had yet to develop anything resembling intuition, the flashes of brilliance that decided longer matches. Levy predicted that chess programs would never overcome this lack of flair, but that steadily expanding processing power would… ([Location 370](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=370)) - The Fire and Water program proved intelligent enough to beat Demis’s little brother, George. Demis was delighted. Admittedly, George was all of five years old at the time, but Hassabis still chalked this up as a famous achievement. Thanks not least to Demis’s tutoring, George was pretty good at games. “It was amazing that I’d made something that could beat him,” Hassabis remarked, with satisfaction. ([Location 421](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=421)) - Hassabis loved Molyneux’s god games, and he resolved to compete in Bullfrog’s contest. Fittingly, his variant on Space Invaders involved chess: The player’s avatar was positioned in the middle of a chess-like grid, and chess-piece enemies advanced on it from either side in chess-like formation. A few months later, another announcement in the Amiga magazine listed Chess Invaders as a runner-up.[22] It was not quite enough to win Hassabis a job. But he called up the company and landed an invitation to visit. ([Location 440](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=440)) - Hassabis was soon invited to stick around at the studio for a week, and was assigned a desk next to Molyneux. His quick intellect brought out the boss’s good side: The savant handed down instruction and the pupil soaked it up; the two got along famously. Hassabis was also fascinated by the other Bullfrog employees: technically talented, self-made young men, many of whom had dropped out of high school, being too idiosyncratically gifted or plain wild to sit meekly in a classroom. ([Location 454](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=454)) - The following winter, Hassabis won admission to the University of Cambridge, where he would study computer science. The college authorities ruled that although he was academically ready, at sixteen he was too young to enroll, so he should find something else to occupy himself for the next year or so. This suited Hassabis just fine: He would graduate from high school a year early and go back to Bullfrog until Cambridge was ready for him. ([Location 460](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=460)) - But Hassabis was happy to embrace Molyneux’s vision. Together with a handful of other young employees, he moved into Molyneux’s higgledy-piggledy country house, a mysterious old rectory with hidden doors and secret passages and plenty of gear for gaming.[26] There, the coding commune worked both day and night, gathering periodically in the kitchen for impromptu spaghetti and discussion. The line between working and philosophizing blurred, Hassabis recalled. “We were brainstorming these big ideas. There was this thrill of unbridled creation.” ([Location 468](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=468)) - Sometime in this period, Molyneux gave Hassabis a copy of Gödel, Escher, Bach, a fire hose of a book that has inspired a remarkable number of future AI scientists.[28] This tome, which won the Pulitzer Prize, delivers a torrent of ideas on “fugues and canons, logic and truth, geometry, recursion, syntactic structures, the nature of meaning, colonies, concepts and mental representations, translation, computers and their languages, DNA, proteins, the genetic code, artificial intelligence, creativity, consciousness and free will,” as the author, Douglas R. Hofstadter, proclaimed, without evident modesty. Seventeen years old and voraciously curious, Hassabis was deeply fascinated by all of the above. But the passages that influenced him most were the ones on intelligence and consciousness. ([Location 484](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=484)) - Readers needed to reckon with the reality that, notwithstanding such feelings, human intelligence and machine intelligence resembled one another closely. “Only if one keeps on bashing up against this disturbing fact can one slowly begin to develop a feel for the way out of the mystery of consciousness: that the key is not the stuff out of which brains are made, but the patterns that can come to exist inside the stuff of a brain,” Hofstadter wrote.[29] For a youth who was already fascinated by programming intelligence, this line of argument was thrilling. If the patterns were what mattered—those crackles of electricity, ultimately governed by genetic code—then similar patterns could be encoded into artificial brains. What the mind could do, computers should be able to do—one day. ([Location 499](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=499)) - Even as he plowed his way through Hofstadter’s dense work, Hassabis was inhaling science fiction. When he was younger, he had read Isaac Asimov’s Foundation series, and his imagination was fired by the main character, Hari Seldon, who prophesies the collapse of the Galactic Empire and plots to mitigate the fallout. “The only way for us to have that capability of predicting disasters, and then averting them, would be to have AI,” Hassabis said later. At Bullfrog, Hassabis ripped through the first books in Iain Banks’s Culture series, which described a post-scarcity, interstellar society dominated by intelligent artificial beings. In this world of Banks’s imagining, AI systems would generate economic abundance, and citizens would lack for nothing. Space travel would be as simple as hopping on a London bus, and people could choose among hundreds of planets to live on. What’s more, the intelligent machines that Banks envisaged would exist peacefully alongside humans; they would be too preoccupied with their own intrigues to pick a fight with mortals. It followed that artificial intelligence was not to be feared. To the contrary, it would enrich human experience. ([Location 511](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=511)) - Taken together, Hassabis’s experiences at Bullfrog answered his big post-Liechtenstein question: His mission and purpose would be to build artificial intelligence. Molyneux and Gödel, Escher, Bach had planted the idea: Computers would soon do whatever the brain could do. Iain Banks had supplied a utopian vision of what AI’s realization could mean: boundless human flourishing. And the Carnegie Mellon professor had inadvertently established that Hassabis possessed the requisite talent: If he could impress an eminent scientist before even attending university, there was no limit to what he might accomplish in the future. “I decided then that I was going to dedicate my career to working on AI,” Hassabis recalls. “I already had the kernel of the idea for what eventually became DeepMind.”[32] ([Location 530](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=530)) - Hassabis’s determination to attend Cambridge owed much to a film, Life Story. The movie celebrates the scientists James Watson and Francis Crick. They meet at the university, enjoy sunny strolls along the River Cam, and hurry along rain-drenched cobbled streets to the shelter of the ill-lit Eagle pub, where they speculate exuberantly about DNA and conspire to become famous. Ultimately, with the help of an X-ray image created by their rival, Rosalind Franklin, Watson and Crick discover the double-helix structure of DNA, winning the Nobel Prize for their achievement. The importance of this film to Hassabis, like the influence of Iain Banks’s Culture series, says much about how he came by his worldview. Dropping out of school for periods, and operating beyond the understanding of his parents, Hassabis’s ambitions were shaped by a magpie collection of encounters. “I’m quite indiscriminate about knowledge,” he said. “I’ll have any knowledge. Chess game, book, philosophy, I’ll drink it all in.” But the Life Story movie points to something else as well. Multiple forces had driven Hassabis down the path toward AI. But perhaps the strongest driver was the thrill of science: the prospect of discovering the truth behind the other truths. “What’s fun?” Watson asks Crick, early on in Life Story. “Oh, the big questions,” Crick responds. “What is man? What is life? How did we come to be the way that we are?” ([Location 545](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=545)) - Thrilling though this prospect seemed, Hassabis was also practical. When he signed up for a game, he liked to feel that he could win, and physics seemed like a long shot. The way he saw things, all physicists since Einstein had ultimately come up short. They had failed to hit on a theory that explained all of reality. “Even Richard Feynman couldn’t do it,” Hassabis said, matter-of-factly. “He died without understanding everything. I realized that however good I was going to be, I was unlikely to surpass him.” Following this line of thought, Hassabis hit upon a strategy. He resolved to go after the infinite mysteries of physics with the help of artificial intelligence. Science had always advanced courtesy of new tools: Telescopes had allowed humans to peer into space; X-ray machines had made it possible to see into humans without invasive surgery. AI would be the ultimate lever: an extension not merely of vision but of the capacity for understanding. ([Location 564](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=564)) - for those who expect geniuses to be one-track obsessives, he was something of a disappointment. In his first year as an undergraduate, he developed a taste for electronic music: After nights of raving with friends, he would flop down on his bed at dawn and put on Music for the Jilted Generation by the electronica band The Prodigy. In his second year at Cambridge, he took buddies for joyrides in his new car—a Porsche 911 Turbo. In a flourish of nineteen-year-old chutzpah, Hassabis had persuaded Molyneux to lend him the Porsche, saying that he needed it for his commute to the Bullfrog studio, where he had agreed to contribute as a consultant.[2] Sometime in this period, Hassabis fell in love with an Italian undergraduate who would become an academic bioscientist. They would later marry. ([Location 573](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=573)) - “But I also think it’s just my personality. I want to help people and I feel very strongly that it’s just really bad to manipulate or control people.” ([Location 591](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=591)) - In this sense, his gentle affability and his ferocious competitiveness were two sides of the same coin. On the one hand, Hassabis loved to yank and spin the table-football figurines in the college bar: It was an everyman hobby, telegraphing his approachability. On the other, he became so fixated on the game that he organized a college team and vanquished adversaries all over campus. ([Location 597](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=597)) - As he made the most of Cambridge life, Hassabis remained a magpie. Whatever his fellow students were excited by, he was eager to wrap his mind around, too—and his enthusiasm was contagious. One time, between bouts of table football, a biologist acquaintance told him about the mysteries of protein shapes that, if only they could be accurately plotted, would unlock extraordinary medical breakthroughs. It was a passing conversation, so fleeting that it failed to register in the memory of the biologist friend, but Hassabis filed it away in the back of his extraordinary mind as a potential double-helix-type challenge.[5] A quarter of a century later, DeepMind unraveled the mystery of proteins, winning a Nobel Prize for revolutionizing the field of structural biology.[6] ([Location 607](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=607)) - When the two eventually met, mutual respect and overlapping interests made for a quick melding of minds. “We shared the same passions,” Silver recalled. “The big debates about AI, the deep philosophical questions about computer science. It was just really fun talking to him.”[8] ([Location 619](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=619)) - Silver, like Hassabis, had read Gödel, Escher, Bach: The first name in the book’s title belonged to the mathematician Kurt Gödel, who had proved that, contrary to the Dartmouth pioneers’ presumption, no system of logical deduction could encompass all possible true statements. To Hassabis and Silver, Gödel’s “incompleteness theorem” merely confirmed what was intuitively obvious. After all, humans engage in deductive logic only a small fraction of the time. Mostly, they take in jumbled images, words, smells, and sensations; then they extract meaning from the noise—a process that logicians call induction and lay people might call pattern recognition. “The idea of using first order logic to understand language—it was obvious to me this was nonsense,” Hassabis remarked later. “We don’t speak in first order logic and yet we can understand each other.” If intelligent humans could comprehend one another’s messy sentences, it followed that intelligent machines should be able to make sense of imprecise, unstructured data. They should digest examples and derive general truths. They should be inductive as well as deductive. ([Location 654](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=654)) - In their third year at Cambridge, as they imagined this futuristic infinity machine, Hassabis and Silver persuaded an unusual professor, John Daugman, to have them over to his office for a series of tutorials. This sort of small-group teaching was what made Cambridge special, and Daugman’s interests ranged far beyond symbolic programming. He taught courses on information theory and computer vision, becoming famous for inventing an algorithm for iris recognition. Hassabis recalls the tutorials with Daugman as “nirvana sessions,” and Daugman took an instant liking to the friendly pair. “You could actually talk to them,” the professor remarked. “I’m sorry to say this, but in general that’s not true about computer scientists.”[12] The sessions with Daugman led Hassabis to his next epiphany. He realized that a superhuman computer would be more than just a means to a scientific end, the end being progress in scientific understanding. Rather, the computer might itself be the end, because information, marshaled by computer science, was the basic unit of reality. The traditional contenders for the status of fundamental building block—energy, matter—were less compelling by far; only information provided the basis for explaining all facets of experience. The behavior of particles, the flow of energy, and even human consciousness could be seen as examples of information processing. Of course, Hassabis had already absorbed the germ of this idea from Hofstadter. Biology, Hofstadter had argued, was an information processing system; what defined life was not muscle or tissue but the signals that animated them.[13] But with Daugman, Hassabis went deeper, studying Claude Shannon’s theories on what information is: how it can be quantified, stored, and transmitted over time and space; how it is defined by a simple but profound insight—as the opposite of uncertainty. Seen in this light, any reduction in uncertainty would depend on information, intelligently processed. A theory of everything—that is, a theory that reduced uncertainty to something near zero—would in all probability take the form of a computer program. “That’s the way I still view the whole universe,” Hassabis said later. “I think information is the fundamental unit.” ([Location 682](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=682)) - Hassabis sometimes explained the need for an infinity machine by contrasting physics with biology. “A deductive system like mathematics may be the perfect description language for physics,” he said; Newton had managed to capture the nature of motion in a series of equations. “But AI may be the right description language for biology, because biology is so messy, emergent, dynamic, and complex.” It was impossible to imagine something as elegant as Newton’s laws to describe a cell. But if you fed an infinity of data about cells into an inductive computer, the machine might figure out a way of describing what was going on—it would see the unseen patterns, the hidden laws, that explained cellular behavior. “AI—the kind of information system we’re building—will probably be the right tool for this,” Hassabis suggested.[14] ([Location 708](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=708)) - Over the ensuing years, Hassabis’s two-part epiphany stuck with him. First, information was the fundamental unit of reality. Second, a machine that learned for itself how to induce nature’s patterns was the most powerful imaginable tool with which to apprehend reality. And while artificial intelligence could push the frontiers of science, it could also do much else besides: discover medicines, extending the lifespan of humans; solve the obstacles to nuclear fusion, rendering energy clean and abundant. As Hassabis once put it to the Guardian, “What we’re working on is potentially a meta-solution to any problem.”[15] A machine that could navigate an infinity of data would be infinite in its reach. ([Location 715](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=715)) - Hassabis was drawn to Black & White by the opportunity to experiment with AI programming. Whereas the characters in Theme Park had been complex finite-state machines, obeying fixed rules governing their preferences, Black & White would be the first game in which the avatars’ internal rules changed based on feedback. If a digital creature hurled rocks at villagers and the player responded with a slap, the creature would learn not to repeat this transgression. If a creature ate excessively and the player reassured it with a pat, the creature would adjust its algorithmic preferences in favor of continued bingeing. This basic “reinforcement learning” system was a small step in the direction of AI: a program that adjusted its program—that was capable of learning. On a more practical level, the creature’s adaptability made every experience with the game feel fresh. When Black & White eventually appeared, it was another Molyneux blockbuster. Hassabis contributed to the early brainstorming for the game, but he did not stick around to implement the vision. He had matured since his first stint with Molyneux, and his reaction this time was different. Before going to Cambridge, Hassabis had been so excited to be at Bullfrog that he ignored Molyneux’s volatile side; now he noticed it. He could see that, despite his undisputed creativity, Molyneux was a fabulist, a teller of tall tales, often promising journalists that his next project would include some fantastical technical advance, never mind that his own coding team had assured him that it was impossible. In his conversations with Hassabis—his protégé but now, potentially, his rival—Molyneux would claim to have discovered a secret new path forward to AI, but he would never quite produce the evidence: He was by turns emphatic, vague, elusive, and menacing, yo-yoing between warmth and iciness, bravado and tears, stoking the anxiety of everyone around him. Years later, Hassabis compared Molyneux to the mysterious character in The Magus, a novel by John Fowles. The magus is manipulative, mendacious, a master of illusions and mind games. “That was pretty much how Peter approached me,” Hassabis said.[19] I thought back to what Hassabis had told me about his mother’s religion—about how bad he felt it was to dominate people. Perhaps, if life were a god game, Molyneux would be the black god, exercising power through manipulation and menace. Hassabis would be the white god, exercising power by dint of contagious enthusiasm and lucidity. ([Location 732](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=732)) - Toward the end of his time at Cambridge, he had confided to his friends that, to pursue his dream of building AI, he planned to found a company.[20] It was a shocking idea. Entrepreneurship was a foreign concept on the Cambridge campus; Britain had no equivalent to Silicon Valley. “If you had looked at the students and asked, ‘Who’s going to set up a company?’ the answer would’ve been nobody,” one of Hassabis’s contemporaries recalled. “It was like, who are you going to work for, or what PhD are you interested in? Of course you don’t set up a company!”[21] Perhaps thanks to his exposure to Molyneux, perhaps also to the influence of his free-spirited father, Hassabis was an exception. He saw no reason not to start a company—and so he did. ([Location 753](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=753)) - Hassabis’s first call the next day was to his friend David Silver. When they were at Cambridge, Hassabis had talked about his plan to found a company, and Silver had been intrigued without ever quite believing him. After all, the celebrated Peter Molyneux pretty much worshipped the ground on which Hassabis walked. Why take the risk of starting a competitor? “That games company we talked about, do you want to do it?” Hassabis asked. The previous evening’s doubts had been erased. His conviction was infectious. Silver had taken a job at a cool software boutique, building special effects for movies. He didn’t miss a beat. “Absolutely. Let’s do it.”[1] Silver went over to the Hassabis home, and the two of them thrashed out a business plan. ([Location 767](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=767)) - Hassabis quickly talked McDonagh and Clarke into joining. His powers of persuasion were uncanny: “Demis had what we called his Jedi mind trick,” Silver said later. “He would kind of be like, ‘You will believe the things I’m going to say,’ and then people did believe them.”[3] The stint with Molyneux had no doubt helped Hassabis to develop this presentational flair. Entrepreneurship flourishes in clusters such as Silicon Valley, where technologists learn how to project confidence by apprenticing to one another. Britain was, to a first approximation, a start-up desert. Hassabis was lucky to have spent time working for a master storyteller. ([Location 786](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=786)) - When Hassabis had talked of starting a company during his last months at Cambridge, his ambition had been to build powerful AI, not just to design video games.[6] In founding Elixir, he was balancing his ambition against his practical side. A games studio would allow him to at least experiment with AI, and it would give him entrepreneurial experience. It might also make him rich—rich enough, perhaps, to launch his ultimate dream: a Manhattan Project for artificial intelligence. ([Location 818](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=818)) - Hassabis installed table football and announced to all and sundry that nobody would beat him, ever. His lieutenants trained maniacally to prove him wrong. At length, the evil day arrived. The champion’s reign ended.[9] Hassabis retreated from the table and sat in his chair mutely. “There was this dark cloud over his chair. He had this somber, cloudy face. And at a certain point he just couldn’t contain it anymore. “He stood up and said, ‘It’s like my soul is on fire!’ ” David Silver remembered. ([Location 850](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=850)) - To stir the creative juices, he led viewings and discussions of film noir classics: Fritz Lang’s Metropolis, Batman, and his own favorite, Blade Runner. ([Location 859](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=859)) - When Silver eventually located Hassabis, it transpired that no apology was needed. Hassabis had proceeded with his presentation despite the hardware disaster, and everyone had loved it. The demo had been clunky—even when he got the PC working, it could only display the frames in slow motion—but the accompanying patter had been genius. Hassabis had waxed effusive about Republic’s imaginative scope. Players could order beatings of noted community members, plot the martyrdom of a revolutionary student, and experience the dynamics of rioting throngs—here Hassabis dazzled his audience by invoking the book Crowds and Power by the Nobel Prize winner Elias Canetti. Such was the curiosity and anticipation that Hassabis generated, Republic: The Revolution won an award at the expo for the best upcoming game, and a reviewer declared it the most exciting strategy concept since Civilization.[15] Standing on the precipice, teetering at the edge, Hassabis had pulled off his greatest Jedi mind trick ever. ([Location 904](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=904)) - Despite this improbable escape, Elixir was in trouble. The hardware crash reflected a hard truth: The computers of the time couldn’t handle Hassabis’s ambitions. Round-the-clock coding sessions were taking their toll: Silver’s sudden exit from the presentation room had been a warning of incipient burnout. Several other members of the team were in a similar condition. Hassabis’s mind tricks had helped him to recruit talent, raise money, and create spectacular buzz. But they had raised expectations to the point where delivery became almost impossible. Over the next couple of years, Republic’s release date was pushed back repeatedly.[16] To get the product out the door, the team diluted the game’s most innovative ideas, and morale inevitably suffered. As the keeper of the vision, Hassabis fought a rearguard action against compromise, and it took time for him to recognize the trap that his own charisma created. “Who would’ve thought that you can actually inspire people too much?” he reflected, years later. “Well, you can, because you can get to the point where you are deluding your team, and then they are deluding you also. “It’s like, I’m making the judgment this is possible because the engineers are telling me it’s possible; but they’re only telling me it’s possible because I’ve over-inspired them,” Hassabis said. “So, in fact, none of us is getting real feedback.” ([Location 914](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=914)) - Over the previous year or so, Elixir’s coders and designers had taken to pulling Silver aside: You tell Demis that this feature will not work, they’d plead. You’re the only one he’ll listen to. But the job of communicating reality to Hassabis was grueling. “You had to push the conversation to the point where he got more and more intense and defended his positions more and more strongly,” Silver said later. “The stronger he got, the closer you were. Then eventually he might go quiet.… ([Location 929](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=929)) - Apologizing to his girlfriend, who preferred the climate in the south of France, Silver wrote to Sutton and proposed that he should do a PhD in snowy Canada. ([Location 979](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=979)) - As a precocious undergraduate, Hassabis had rejected symbolic programming because it failed a basic test: It was not how human intelligence operated. The way Hassabis saw things, this test was crucial, because powerful machine intelligence, when it eventually arrived, would emulate the human variety. After all, the human brain provided the only grounds for believing that a general, flexible intelligence was even possible: The brain was, as Hassabis said, the “existence proof” that underpinned AI endeavors. It followed that, to build artificial intelligence, one should understand human intelligence first. Following this logic to its daunting conclusion, Hassabis resolved to do a PhD in neuroscience.[24] ([Location 987](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=987)) - Hassabis lacked the normal training for this. He was neither a doctor nor a biologist nor even a psychologist; he knew little of the chemistry of the brain and the nervous system. Undismayed by this considerable gap in his résumé, he brandished his credentials as a computer scientist, talked up his exotic entrepreneurial record, and won acceptance to the doctoral program at University College London. ([Location 993](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=993)) - On his first day as a PhD student, Hassabis arrived bubbling with research hypotheses. Not all of them were sensible. “He’d have these crazy ideas,” his friend and fellow PhD student Dharshan Kumaran recalled. “But he was very creative.”[25] ([Location 1015](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1015)) - “We always used to have our discussions in the tiny kitchen in the neuroscience building,” Kumaran remembered. “Demis would explain some idea and I’d say, ‘No, that doesn’t seem right to me.’ Then he’d go back and think about it in a different way. And then we’d be back in that kitchen.”[26] After batting ideas back and forth, Hassabis and Kumaran eventually came up with a workable experimental strategy. They would test patients who had suffered damage to the hippocampus, a pocket of tissue located deep within the brain where the day’s memories are recorded. If Hassabis’s hypothesis was correct—if memory and imagination were linked—patients with memory loss would also struggle to imagine things. With Maguire’s help, the two friends identified a handful of rare patients who had damage to the hippocampus but were otherwise healthy. Sure enough, when Hassabis and Kumaran carried out their tests, they got the result that Hassabis had anticipated. Damage to the hippocampus harmed patients’ ability to visualize a three-dimensional scene, like a day at the beach or a shopping trip. A part of the brain associated with memory was indeed crucial to imagination. ([Location 1022](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1022)) - “I like Kant’s idea that the world out there is basically a mental construct,” Hassabis reiterated to me now. The stuff that physicists studied—matter, energy, time—was ultimately less real than the bits of information pulsing between neurons. “I’ve always been fascinated by the Brain in a Jar thought experiment,” Hassabis continued, referring to a philosophical device for exploring the relationship between reality and consciousness. The experiment posits that a brain is removed from a body and kept alive in a jar of nutrients. It is hooked up to a computer that simulates sensory inputs: a view of the sea, the sound of the waves, the smell of salt and seaweed. The computer creates a virtual reality so convincing that the brain believes it is still living a normal life, taking in the outside world through eyes, ears, and nostrils. Of course the brain is really just experiencing a simulation—a series of ones and zeroes generated by a computer. I could see why Hassabis was taken with this thought experiment. It was the premise for The Matrix, another classic sci-fi film that he was fond of. The humans in the movie believe they are living real lives, but their minds are connected to a master computer and their bodies are harvested for energy. If the brain in the jar could be misled into believing that it was experiencing reality, how can the rest of us be sure that we are not also living in a simulation? If the brain in the jar can see and hear and smell things that have no presence in the real world, perhaps our consciousness may also be a product of ones and zeroes in the brain, utterly divorced from whatever physical reality exists around us? “Reality may be constructed by the mind,” Hassabis repeated. “That’s what I think Kant was getting at.” I marveled at how Hassabis’s experiences and ideas appeared to slot together. His curiosity about physics had spurred him to work on AI, the ultimate tool to unlock science. His curiosity about AI had led him to investigate the human brain, the existence proof for intelligence. His work on simulations in video games echoed his research on simulations in the mind. And influences as various as Immanuel Kant, Gödel, Escher, Bach, John Daugman’s tutorials, and neuroscience had pushed Hassabis toward the same bottom line: that information was the fundamental unit of reality. ([Location 1044](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1044)) - Hassabis toyed with start-up ideas that, like Elixir, would combine an opportunity to work on AI with a commercial payoff. He imagined an AI-powered recommendation algorithm, focused on matching consumers with the right TV shows. He considered a variation: a recommendation system for the bag of tricks in Apple’s new app store. But these stepping-stone projects failed to quicken Hassabis’s pulse. Now thirty-two, he was conscious of his biological clock. It was time to pursue his life’s ambition more directly. ([Location 1073](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1073)) - Still determined to identify soul mates, Hassabis figured that perhaps his fellow postdocs might be more on his wavelength, so he scanned the bios on the Gatsby website. But out of the entire cohort of Gatsby fellows, only one single researcher confessed to an interest in building powerful AI. Hassabis made a mental note of the photo accompanying the bio, and tried to remember the researcher’s name: Shane Legg—it had a cowboy ring to it. Perhaps he would bump into this cowboy guy in the Gatsby cafeteria. ([Location 1086](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1086)) - Later, Poggio would say that, of the many Nobel Prize winners he had encountered, the majority were both brilliant and lucky—lucky in the sense that they had chosen a research problem that turned out to be both consequential and soluble. But a handful of Nobel laureates, Poggio said, were so exceptionally gifted that they were going to win the prize no matter what. In this category, Poggio placed the physicist Richard Feynman, the biologist Francis Crick, and his postdoctoral student Demis Hassabis.[3] ([Location 1093](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1093)) - Hassabis put a brave face on this rejection. “Well, this is kind of fantastic,” he remembers telling himself. If an establishment figure had embraced his vision, it would have signaled that it lacked originality—and that rival entrepreneurs with similar ideas would soon be circling. The fact that Winston had dismissed his arguments, while saying nothing that Hassabis hadn’t heard before, confirmed that an AI company would at least be contrarian. ([Location 1116](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1116)) - At the start of October 2009, Hassabis boarded an elevator at the Gatsby Unit. A curly-haired man with the lithe figure of a dancer maneuvered himself and his suitcase into the small space beside him. Hassabis immediately recognized the stranger as the elusive Shane Legg, the sole postdoc in the building who shared his AI ambitions. Glancing at the suitcase, and realizing that Legg was about to disappear out of town, Hassabis struck up a conversation. “Where are you going?” he asked brightly. Legg answered in a pronounced New Zealand accent. He was on his way to New York to attend something called the Singularity Summit. Hassabis had never heard of this event. “What happens there?” he asked. Legg explained that the Singularity Summit was an annual gathering of AI believers. The “singularity” was the moment when machine intelligence would surpass human intelligence. At that point, machines would know how to improve themselves without human assistance, and their capability would explode upward. The elevator reached the ground floor. The doors opened; the singular New Zealander marched off; the conversation was over. Hassabis was left to ponder what had just happened. After years of preaching the AI gospel to others, he had just met somebody who might be more wired into that world than he was. ([Location 1127](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1127)) - In late 1999, at the height of the bubble in internet stocks, Legg moved on to his next gig: a start-up called Intelligenesis. This shift, as random as the proverbial flapping of a butterfly’s wings, would affect the path of AI history. Intelligenesis turned out to be no ordinary company, and its founder, a Brazilian American prodigy named Ben Goertzel, was not remotely normal, either. Perpetually unkempt, his face framed by waves of curly hair that fell below his shoulders, Goertzel had earned a PhD in mathematics by the age of twenty-two; by thirty, he had published four textbooks and had held university appointments in math, psychology, and computer science. In the mid-1990s, Goertzel had decided that the rapidly expanding internet was not merely a revolutionary communications platform or an opportunity to rethink commerce. It was a new form of intelligence, a sort of worldwide mind: Each computer on the network resembled a neuron, each hyperlink a synapse, each human user a sense organ.[8] In 1997, Goertzel founded Intelligenesis to build upon his vision. Goertzel set his team to work coding a digital brain. In some ways, this was a precursor to Hassabis’s Project Orion, but coming a decade earlier, and with computing still much less advanced, it was even more of a long shot. The plan for how this brain would function was a lot wackier, too. Goertzel planned to build a “Baby WebMind,” then release it on the internet and watch it mature into adulthood. Pretty soon, Goertzel insisted, a worldwide, self-organizing machine consciousness would spring forth, ending humanity’s monopoly on intelligence.[9] Goertzel’s baby-brain project was a bubble-era flight of hubris. But Legg enjoyed this trippy stuff so much that, after working remotely from New Zealand, he picked up and moved to New York, where Intelligenesis had an office. The way Legg saw things, the WebMind’s viability was almost beside the point. What captivated him was the question of what it meant to build intelligence, which further led him to consider what intelligence was, or what it might be. To someone whose intelligence had been misclassified as a child, no subject could be more riveting.[10] ([Location 1163](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1163)) - Around the time he coined that term, Legg read The Age of Spiritual Machines by the inventor and futurist Ray Kurzweil. The book’s central message was that the power of computers was about to surge dramatically. Since the 1960s, technologists had referred knowingly to Moore’s Law, and the very familiarity of this concept had caused people to stop thinking about it. But the prophecy that the power of semiconductors would double every two years was not something to be filed and forgotten in the back of the collective mind. To the contrary, the more time went on, the more mind-boggling its implications. A doubling every two years implied exponential progress—the curve would start off relatively flat and then later explode upward. As of the year 2000, human brains were fully one million times more powerful than the most advanced machines, a huge gap in capability. But at some point in the 2020s, Kurzweil calculated, computers would draw even. And once that happened, they would accelerate past their creators. The singularity was approaching. ([Location 1193](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1193)) - Kurzweil’s exposition of Moore’s Law affected Legg deeply. Reading his book, Legg realized that not only was the power of computers set to explode, the amount of data that could be fed into the machines would also explode, thanks to the spread of the internet. With better hardware and more data, the third component of AI—algorithmic advance—would become insanely valuable. “I was thinking AI is going to be real. I buy the basic argument of Kurzweil. So, if that’s the case, I should go get a PhD,” Legg recalled later. “So, I thought to myself, ‘OK, what’s the biggest issue in AI as I see it?’ “And I thought, ‘Well, the biggest problem is there isn’t a measure for intelligence.’ “It’s very hard to build an intelligent machine if you can’t even measure it.” Having discussed these matters thoroughly with himself, Legg found his way to Marcus Hutter, a researcher with similar ideas at IDSIA, an AI institute in southern Switzerland. Under Hutter’s supervision, he embarked on a PhD, assembling multiple possible conceptions of intelligence, and applying his mathematical training to their measurement. ([Location 1206](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1206)) - the true mark of intelligence was generality. Together with Hutter, Legg landed on a summarizing phrase: “Intelligence measures an agent’s ability to achieve goals in a wide range of environments.” ([Location 1220](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1220)) - Hassabis was less fixated on the surreal threat than on what he had heard beforehand. His impression from that elevator encounter had been confirmed. Legg’s imagination was wide open to the possibility of superhuman AI, and he had the technical tools to help build it. Hassabis was not going to miss the opportunity to bond with the New Zealander this time. He went over and introduced himself. ([Location 1257](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1257)) - “I’d never read any of Kurzweil,” Hassabis said. “For me, it was Gödel, Escher, Bach, Asimov, Iain Banks, my own practical work with Molyneux, Blade Runner. Those were my influences. I don’t think I even knew who Kurzweil was because I was sort of in my own parochial backwater, just dreaming dreams. ([Location 1268](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1268)) - Hassabis countered that academia was too slow: To build powerful AI, they would need a team, a sense of urgency, and freedom from academic bureaucracy. A mission-driven start-up—a Manhattan Project, as Hassabis liked to say—could surely be funded by the right sort of investor: a billionaire, or possibly a multibillionaire, with the stomach for the long horizon. ([Location 1285](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1285)) - Since that fleeting barbecue meeting a few years before, Suleyman had struck up a relationship with the older Hassabis. Demis had exited Elixir with a few million pounds, and in 2007 Suleyman had proposed a business partnership. Hassabis would provide capital; Suleyman would purchase apartments; these would be rented out, and the two would split the proceeds.[22] The partners discussed their plans over a few lunches near University College London, where Hassabis was studying for his PhD, and Suleyman talked about what he was reading. In contrast to Hassabis, who loved books about science, Suleyman devoured volumes on politics, sociology, and complexity theory, logging every title that he went through. ([Location 1370](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1370)) - Two days later, Suleyman followed up with an email. He congratulated Hassabis on his poker winnings and pointed him toward a Wired magazine profile of Sergey Brin, the cofounder of Google. According to the article, Brin was pouring part of his $15 billion fortune into computational medicine, hoping to drive a revolution in the pace of drug discovery. Perhaps Brin was the sort of billionaire who might fund an AGI company?[27] Hassabis emailed back. “Very cool article,” he said approvingly. And then he proposed a collaboration. He was gearing up to pitch a different billionaire at the next Singularity Summit, in August; he had the outline of a business plan, but it needed fleshing out, and the summit was approaching. Might Moose have time to help draft the document? ([Location 1421](https://readwise.io/to_kindle?action=open&asin=B0FQGPMD79&location=1421))