What do you think?
Rate this book
The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence
From one of our leading chroniclers of the intersection of innovation and capitalism, a landmark reckoning — based on unprecedented access — with one of the world’s most brilliant and driven tech visionaries and his game-changing company.
Even by the standard of a tech industry stacked with so-called visionary leaders possessed of the godspark of genius, Demis Hassabis is universally recognized as a special case. Born poor in North London to immigrant parents, a dominant chess prodigy by age five and wizard coder in his teens, he turned down a seven figure offer before he turned 18 to feed his insatiable scientific curiosity at Cambridge. Later, he added a neuroscience PhD to his computer science skills to pursue the dream of artificial general intelligence, his ultimate goal to solve the world’s hardest problems and usher in an era of super-abundance. And along with a small group of fellow travelers around the world, that is the path he is very much still on, winning a Nobel Prize along the way, and imagining machines that will unlock the deepest mysteries of the universe.
Hassabis has given Sebastian Mallaby a great deal of his own time, sitting still for over thirty hours of probing conversations, and has opened up his company to him, allowing Mallaby to spend hundreds of hours talking to all its key players, including people still there, like Shane Legg, and those who’ve left, like Mustafa Suleyman. The result is a revelation-packed account of a singular figure and his company and a profound reckoning with this whole protean field in the crucial, indeed historic era, when it went from the periphery to the center of the world.
No one questions Demis Hassabis’s brilliance, commitment, or the worthiness of his ultimate goals. There are those who do flinch at the heat of his fire, and question some of his strategies. His competitors, of course, want to beat him. He is in a game at a moment where the sense of the stakes is matched only by the exorbitant costs — for talent, and for compute. Faustian bargains abound. Celebrated scientists pursue the technology because, like Robert Oppenheimer, they cannot resist the sweetness of discovery. Others pursue it for money or power. The inventors believe they control their technology, but often, the technology controls them.
THE INFINITY MACHINE surfaces Hassabis’s story’s importance across a number of dimensions. Not least, crucially, this is not at heart a Silicon Valley story. Hassabis deals with the Valley and takes its money, but he has remained outside of it, and indeed, furiously critical of it, often lambasting its leaders in his conversations with Mallaby. DeepMind has never seen large language models as the one path to the AGI mountaintop. That left them blindsided by the OpenAI ChatGPT breakthrough, but it has also caused them to advance on a much broader front, and arguably have a bigger real world impact than their competitors. So far. The end of this race cannot be known, but as this great book shows us, the place of Demis Hassabis and DeepMind in the history of tech, capitalism, and science is secure.
Even by the standard of a tech industry stacked with so-called visionary leaders possessed of the godspark of genius, Demis Hassabis is universally recognized as a special case. Born poor in North London to immigrant parents, a dominant chess prodigy by age five and wizard coder in his teens, he turned down a seven figure offer before he turned 18 to feed his insatiable scientific curiosity at Cambridge. Later, he added a neuroscience PhD to his computer science skills to pursue the dream of artificial general intelligence, his ultimate goal to solve the world’s hardest problems and usher in an era of super-abundance. And along with a small group of fellow travelers around the world, that is the path he is very much still on, winning a Nobel Prize along the way, and imagining machines that will unlock the deepest mysteries of the universe.
Hassabis has given Sebastian Mallaby a great deal of his own time, sitting still for over thirty hours of probing conversations, and has opened up his company to him, allowing Mallaby to spend hundreds of hours talking to all its key players, including people still there, like Shane Legg, and those who’ve left, like Mustafa Suleyman. The result is a revelation-packed account of a singular figure and his company and a profound reckoning with this whole protean field in the crucial, indeed historic era, when it went from the periphery to the center of the world.
No one questions Demis Hassabis’s brilliance, commitment, or the worthiness of his ultimate goals. There are those who do flinch at the heat of his fire, and question some of his strategies. His competitors, of course, want to beat him. He is in a game at a moment where the sense of the stakes is matched only by the exorbitant costs — for talent, and for compute. Faustian bargains abound. Celebrated scientists pursue the technology because, like Robert Oppenheimer, they cannot resist the sweetness of discovery. Others pursue it for money or power. The inventors believe they control their technology, but often, the technology controls them.
THE INFINITY MACHINE surfaces Hassabis’s story’s importance across a number of dimensions. Not least, crucially, this is not at heart a Silicon Valley story. Hassabis deals with the Valley and takes its money, but he has remained outside of it, and indeed, furiously critical of it, often lambasting its leaders in his conversations with Mallaby. DeepMind has never seen large language models as the one path to the AGI mountaintop. That left them blindsided by the OpenAI ChatGPT breakthrough, but it has also caused them to advance on a much broader front, and arguably have a bigger real world impact than their competitors. So far. The end of this race cannot be known, but as this great book shows us, the place of Demis Hassabis and DeepMind in the history of tech, capitalism, and science is secure.
480 pages, Hardcover
Published March 31, 2026
About the author
Sebastian Mallaby
17 books364 followersA Washington Post columnist since 1999. Worked for The Economist from 1986 - 1999.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 338 reviews
April 1, 2026
It's very quickly clear that Sebastian Mallaby is a huge Demis Hassabis fan - writing about the only child prodigy and teen genius ever who was also a nice, rounded personality. After a few chapters, though, things settle down (I'm reminded of the description of the Hitchhiker's Guide to the Galaxy) and we get a good, solid trip through the journey that gave us DeepMind, their AlphaGo and AlphaFold programs, the sudden explosion of competition on the AI front and thoughts on artificial general intelligence.
Although Mallaby does occasionally still go into fan mode - reading this you would think that AlphaFold had successfully perfectly predicted the structure of every protein, where it is usually not sufficiently accurate for its results to have direct practical application - we get a real feel for the way this relatively unusual company was swiftly and successfully developed away from Silicon Valley. It's readable and gives an important understanding of where some our key AI expertise came from.
Perhaps not surprisingly, where the book really takes off is in the later chapters, after the successes of AlphaGo and AlphaFold. when the DeepMind people were left behind by OpenAI's generative AI work and had to rapidly change gear under outside pressure. Let's face it, a story of responding to threat is much more interesting than one of straight success. Not only was this phase of development one where the team was caught on the back foot, a huge rift emerged over whether developers should be heading towards artificial general intelligence (AGI) very carefully for safety reasons (the initial DeepMind approach) or just going for it full throttle like OpenAI.
It's arguable some of the safety concerns were fruitless in that large language models seem unlikely ever to be the starting point for a true AGI, but there was (and is) still a very real threat from this software in areas such as privacy, copyright theft and environmental impact. This part of the book is unputdownable stuff.
Overall, The Infinity Machine is a little too long, giving too much detail of the people outside the core group and of every step along the way of the company's development. However, this doesn't really matter as it provides excellent documentation of a key player in the rise and rise of AI.
Although Mallaby does occasionally still go into fan mode - reading this you would think that AlphaFold had successfully perfectly predicted the structure of every protein, where it is usually not sufficiently accurate for its results to have direct practical application - we get a real feel for the way this relatively unusual company was swiftly and successfully developed away from Silicon Valley. It's readable and gives an important understanding of where some our key AI expertise came from.
Perhaps not surprisingly, where the book really takes off is in the later chapters, after the successes of AlphaGo and AlphaFold. when the DeepMind people were left behind by OpenAI's generative AI work and had to rapidly change gear under outside pressure. Let's face it, a story of responding to threat is much more interesting than one of straight success. Not only was this phase of development one where the team was caught on the back foot, a huge rift emerged over whether developers should be heading towards artificial general intelligence (AGI) very carefully for safety reasons (the initial DeepMind approach) or just going for it full throttle like OpenAI.
It's arguable some of the safety concerns were fruitless in that large language models seem unlikely ever to be the starting point for a true AGI, but there was (and is) still a very real threat from this software in areas such as privacy, copyright theft and environmental impact. This part of the book is unputdownable stuff.
Overall, The Infinity Machine is a little too long, giving too much detail of the people outside the core group and of every step along the way of the company's development. However, this doesn't really matter as it provides excellent documentation of a key player in the rise and rise of AI.
Read
May 23, 2026An unexpectedly motivating book to listen to while running."If you haven't collapsed, you haven't tried your best."
I like how philosophical and narrative-driven Demis Hassabis is.
According to him, the universe can only be perceived through your brain, and therefore, there are 2 ways to understand the universe: through physics and through neuroscience. Physics is external. Neuroscience is internal. Demis chose neuroscience.
In this framing, information is the unit of the universe, not atom.
The first 1/3 of the book covers the story Demis tells about his life. The last 1/3 of the book, however, is the author trying to do speed run of AI coverage, which is less interesting to me.
I like how philosophical and narrative-driven Demis Hassabis is.
According to him, the universe can only be perceived through your brain, and therefore, there are 2 ways to understand the universe: through physics and through neuroscience. Physics is external. Neuroscience is internal. Demis chose neuroscience.
In this framing, information is the unit of the universe, not atom.
The first 1/3 of the book covers the story Demis tells about his life. The last 1/3 of the book, however, is the author trying to do speed run of AI coverage, which is less interesting to me.
May 9, 2026
Every book Sebastian Mallaby writes is phenomenal and this one is no exception (full disclosure: Sebastian is a friend, a friendship that started more than twenty years ago when I made it clear to him that I did not think every one of his Washington Post columns was phenomenal).
The subject of the biography portion (Demis Hassabis) is fascinating, and the chronicling of recent developments in AI could not be more important and timely. Hassabis is almost unique in the landscape: neuroscientist, chess prodigy, game designer, AI pioneer, and one of the most significant scientists of recent times — AlphaFold alone would secure that verdict. One character says he is the type who would have won a Nobel no matter what, not by luck but inevitably. He seems driven by knowledge and importance rather than money, yet as AI becomes big business he ends up selling to and working for the largest of large companies, Google.
The cradle-to-present biography makes for riveting reading — rarely is the childhood portion of a biography as gripping as this one, which recounts stories of his chess prowess at age five with the same intensity as the later chapters. (Mozart comes to mind.)
As the book gets closer to the present it becomes more of a recounting of events we all lived through, paid attention to at the time, and that somehow feel like the distant past even though they are only a few years ago — from the launch of ChatGPT 3.5 to each subsequent iteration and the responses and one-ups by Gemini. All told with considerable weight on Hassabis' perspective and very little sympathy for some of the other players in the drama, most notably Sam Altman, with Mustafa Suleyman occupying a more ambiguous middle ground.
The book is also an interesting exploration of people who are building a technology they are both excited and ambivalent about — doing it within corporations they are likewise both excited and ambivalent about — while trying, in Hassabis' case, to balance a deep enthusiasm for scientific discovery against commercial imperatives. It uses personalities to provide perspective and insight on some of the biggest and most profound issues in the space.
Ultimately the portrait is of everyone stuck in a race no one (or at least not Hassabis) wants to be in, because there are so many competitors in the United States and increasingly in China that coordination becomes impossible. Normally in the economy this is a good thing — competition drives innovation, lowers consumer prices, and makes collusion both difficult and illegal. But it does make me wonder whether that logic holds in this particular case.
This book is only the first chapter in the story of AI. Whether it is also the first chapter, or the first volume, or the complete book for Hassabis personally remains to be seen. The epilogue teases tantalizing possibilities for future fundamental research in physics and other areas, which I found genuinely exciting. But it is also possible that AI is becoming more about organizations than individuals, or that the central figures going forward will be different from Hassabis — as to some degree they already have been for the past few years. We'll see.
The subject of the biography portion (Demis Hassabis) is fascinating, and the chronicling of recent developments in AI could not be more important and timely. Hassabis is almost unique in the landscape: neuroscientist, chess prodigy, game designer, AI pioneer, and one of the most significant scientists of recent times — AlphaFold alone would secure that verdict. One character says he is the type who would have won a Nobel no matter what, not by luck but inevitably. He seems driven by knowledge and importance rather than money, yet as AI becomes big business he ends up selling to and working for the largest of large companies, Google.
The cradle-to-present biography makes for riveting reading — rarely is the childhood portion of a biography as gripping as this one, which recounts stories of his chess prowess at age five with the same intensity as the later chapters. (Mozart comes to mind.)
As the book gets closer to the present it becomes more of a recounting of events we all lived through, paid attention to at the time, and that somehow feel like the distant past even though they are only a few years ago — from the launch of ChatGPT 3.5 to each subsequent iteration and the responses and one-ups by Gemini. All told with considerable weight on Hassabis' perspective and very little sympathy for some of the other players in the drama, most notably Sam Altman, with Mustafa Suleyman occupying a more ambiguous middle ground.
The book is also an interesting exploration of people who are building a technology they are both excited and ambivalent about — doing it within corporations they are likewise both excited and ambivalent about — while trying, in Hassabis' case, to balance a deep enthusiasm for scientific discovery against commercial imperatives. It uses personalities to provide perspective and insight on some of the biggest and most profound issues in the space.
Ultimately the portrait is of everyone stuck in a race no one (or at least not Hassabis) wants to be in, because there are so many competitors in the United States and increasingly in China that coordination becomes impossible. Normally in the economy this is a good thing — competition drives innovation, lowers consumer prices, and makes collusion both difficult and illegal. But it does make me wonder whether that logic holds in this particular case.
This book is only the first chapter in the story of AI. Whether it is also the first chapter, or the first volume, or the complete book for Hassabis personally remains to be seen. The epilogue teases tantalizing possibilities for future fundamental research in physics and other areas, which I found genuinely exciting. But it is also possible that AI is becoming more about organizations than individuals, or that the central figures going forward will be different from Hassabis — as to some degree they already have been for the past few years. We'll see.
April 17, 2026
Sebastian Mallaby's biography of Demis Hassabis is the do-gooder story of a biracial North Londoner, a Cypriot-Singaporean second-generation immigrant, a scholarship boy and chess master, a child prodigy and a polymath, with advanced degrees in computer science and neuroscience, who, in his rapid career, would make a lucrative profit as a video-game designer, build an AI program that could defeat the top Go player in the world, and solve the mystery of protein folding—ultimately winning the Nobel Prize in Chemistry. It's an incredible life-story for a man still in his prime. Mallaby's biography paints a portrait of a straight-talking, unpretentious scientist driven purely by the desire to understand the cosmos, and determined to build a superintelligence that can tackle the most intractable problems. Showing all his typical swaggering confidence and scientific ambition, the epilogue finishes with a teaser—will Demis Hassabis use AI to rewrite quantum mechanics next?
Yet, ironically, as the book progresses, Demis Hassabis recedes into the background. The success of his AI company, DeepMind, always depended on a team of like-minded AGI believers and businessmen, fellow scientists and Silicon-Valley bankrollers eager to make partnerships, take risks and blow off the plodding caution and corseting constraints of academia. Each chapter, and each project that Hassabis spearheaded, ultimately redounds on a rotating cast of characters: Shane Legg, Daan Wiestra, Vhlad Mnih, David Silver, Koray Kavukcuoglu, among many others. It's a success story with several protagonists. Demis Hassabis becomes less a figure of towering genius and more of a managerial overlord, coordinating projects, securing financial backers, defusing media crises, wrangling legal contracts, and working all hours of the night across multiple timezones.
It's a biography that, naturally, has to be diffuse—this is not about one person. There are too many people, too many projects, too many groundbreaking scientific papers. Mallaby's biography has the difficult task of balancing the stories of business chicanery and technological revolutions, from Peter Thiel skeptically funding projects he was betting against to Ilya Sutskever using Reinforcement Learning and Transformer architecture to build the first convincing Large Language Model. Personally, I preferred it when Mallaby gave a behind-the-scenes insight into the eccentric personalities—like the time Hassabis went to the birthday party of Elon Musk's wife in a rented-out castle in Tarrytown New York. All the men had to dress as samurai warriors and Musk himself took on the wold champion sumo wrestler. Mallaby generally refrains from editorial comment—except to defend Hassabis and AI from criticism—but it's moments like this that betray the adolescent antics driving and financing the frontier of artificial intelligence.
At the end of it all, I can't help but feel that the book barely dug beneath the surface. Musk lifting a world champion sumo wrestler, Mustafa Suleyman being kicked out of DeepMind for bullying, Larry Page having to be reminded that hummingbirds and whales are real, Sam Altman double-crossing everyone, and even Demis Hassabis holding progress meetings at 3am—you get the sense that people in the tech world are not well, fiercely intelligent yes, but sometimes lacking the "general" intelligence that they idolize. I wish Mallaby's biography had more vignettes, more interviews, and sharper character profiles. Something seems to be missing from this fascinating story.
Yet, ironically, as the book progresses, Demis Hassabis recedes into the background. The success of his AI company, DeepMind, always depended on a team of like-minded AGI believers and businessmen, fellow scientists and Silicon-Valley bankrollers eager to make partnerships, take risks and blow off the plodding caution and corseting constraints of academia. Each chapter, and each project that Hassabis spearheaded, ultimately redounds on a rotating cast of characters: Shane Legg, Daan Wiestra, Vhlad Mnih, David Silver, Koray Kavukcuoglu, among many others. It's a success story with several protagonists. Demis Hassabis becomes less a figure of towering genius and more of a managerial overlord, coordinating projects, securing financial backers, defusing media crises, wrangling legal contracts, and working all hours of the night across multiple timezones.
It's a biography that, naturally, has to be diffuse—this is not about one person. There are too many people, too many projects, too many groundbreaking scientific papers. Mallaby's biography has the difficult task of balancing the stories of business chicanery and technological revolutions, from Peter Thiel skeptically funding projects he was betting against to Ilya Sutskever using Reinforcement Learning and Transformer architecture to build the first convincing Large Language Model. Personally, I preferred it when Mallaby gave a behind-the-scenes insight into the eccentric personalities—like the time Hassabis went to the birthday party of Elon Musk's wife in a rented-out castle in Tarrytown New York. All the men had to dress as samurai warriors and Musk himself took on the wold champion sumo wrestler. Mallaby generally refrains from editorial comment—except to defend Hassabis and AI from criticism—but it's moments like this that betray the adolescent antics driving and financing the frontier of artificial intelligence.
At the end of it all, I can't help but feel that the book barely dug beneath the surface. Musk lifting a world champion sumo wrestler, Mustafa Suleyman being kicked out of DeepMind for bullying, Larry Page having to be reminded that hummingbirds and whales are real, Sam Altman double-crossing everyone, and even Demis Hassabis holding progress meetings at 3am—you get the sense that people in the tech world are not well, fiercely intelligent yes, but sometimes lacking the "general" intelligence that they idolize. I wish Mallaby's biography had more vignettes, more interviews, and sharper character profiles. Something seems to be missing from this fascinating story.
April 14, 2026
The Infinity Machine left me deeply conflicted, much like many contemporary tech history books chasing the current AI gold rush. At its heart, it promises an insider’s look at the rise of DeepMind and the leadership of Demis Hassabis, but it ultimately struggles to find its footing between serious journalism and a carefully curated corporate memoir.
The biggest hurdle is the book’s unclear target audience. The material simply doesn’t cover the technical or historical landscape in enough depth for anyone actually working in the field, making it redundant for experts who already know the players and the problems. Yet, for the general reader, it rambles through industry tangents and insider anecdotes that feel more like scattered notes than a cohesive narrative. It’s caught in that frustrating middle ground: too superficial for specialists, yet too dense with unexplained context for newcomers.
My primary criticism, however, lies in how the book handles Demis Hassabis and the broader AI leadership class. The narrative leans heavily into the now-familiar trope of these tech founders as modern-day saints, deeply concerned with AI alignment and the preservation of humanity. But where was that same ethical vigilance when foundational IP was systematically scraped from millions of creators without consent or compensation? The text conveniently sidesteps this two-faced reality. Furthermore, despite Hassabis’s public insistence on a pure, almost academic pursuit of science, the book barely acknowledges the grueling, burnout-inducing work culture he’s known to cultivate, nor does it address the discrepancies between his public statements and internal practices. The disconnect between the polished ethos and the operational reality is stark, and the author does little to interrogate it.
All this being said, the level of access is undeniable. The book is clearly built on extensive interviews, and there are fleeting moments where the behind-the-scenes glimpses into the AI race feel genuinely illuminating. But the relentless whitewashing makes you question whether the author truly wrote this book or merely transcribed a legacy project. If you’re looking for a critical, unvarnished examination of the AI boom and its architects, you’ll need to look elsewhere. For industry enthusiasts who don’t mind reading between the lines, it offers a passable, if deeply flawed, window into the machine.
The biggest hurdle is the book’s unclear target audience. The material simply doesn’t cover the technical or historical landscape in enough depth for anyone actually working in the field, making it redundant for experts who already know the players and the problems. Yet, for the general reader, it rambles through industry tangents and insider anecdotes that feel more like scattered notes than a cohesive narrative. It’s caught in that frustrating middle ground: too superficial for specialists, yet too dense with unexplained context for newcomers.
My primary criticism, however, lies in how the book handles Demis Hassabis and the broader AI leadership class. The narrative leans heavily into the now-familiar trope of these tech founders as modern-day saints, deeply concerned with AI alignment and the preservation of humanity. But where was that same ethical vigilance when foundational IP was systematically scraped from millions of creators without consent or compensation? The text conveniently sidesteps this two-faced reality. Furthermore, despite Hassabis’s public insistence on a pure, almost academic pursuit of science, the book barely acknowledges the grueling, burnout-inducing work culture he’s known to cultivate, nor does it address the discrepancies between his public statements and internal practices. The disconnect between the polished ethos and the operational reality is stark, and the author does little to interrogate it.
All this being said, the level of access is undeniable. The book is clearly built on extensive interviews, and there are fleeting moments where the behind-the-scenes glimpses into the AI race feel genuinely illuminating. But the relentless whitewashing makes you question whether the author truly wrote this book or merely transcribed a legacy project. If you’re looking for a critical, unvarnished examination of the AI boom and its architects, you’ll need to look elsewhere. For industry enthusiasts who don’t mind reading between the lines, it offers a passable, if deeply flawed, window into the machine.
June 15, 2026
Very well written biography of Demis and DeepMind. Quite liked the conversations between the author and Demis and how AI can be used to really peek under the hood of the mysteries of the universe we live in. Especially how they went about AlphaFold that led to the Chemisty Nobel. Also interspersed with quite a few good explanations of the technology underlying this AI revolution we’re living through.
June 8, 2026
If you’re like me.
And you’re feeling fascinated/concerned regarding advanced AI, and the psychosocial and political and military and economic and professional and educational and existential (keep adding categories if I missed any - there’s just way too many to list here) impact that it is/will have on all of us starting now, and probably increasing in an exponentially increasingly awesomely awful rate.
Than this might be a pretty good book for you.
It was for me.
It’s Sebastian Mallaby’s biography of Demis Hassabis, the cofounder of DeepMind. And it’s also a history of the race toward advanced AI and superintelligence.
And race is the correct word here.
It’s more of an arms race.
And there just a TON of money and power on the line.
And may the end of society/humanity if it goes wrong.
Anyway.
The first act follows Hassabis from chess prodigy and game designer to (way prodigious - way out front) AI pioneer.
The second act examines how Hassabis (and co) created DeepMind to pursue artificial general intelligence through breakthroughs like AlphaGo and AlphaFold.
Hassabis won a Nobel prize for the AlphaFold work.
The guy is kind of amazing.
And he’s also kind of may be driving us off a cliff.
Him and like literally a handful of other people.
The
The third act explores the tension between scientific discovery, corporate power, AI safety, Google’s control of DeepMind, and competition with OpenAI.
Anyway.
It may be the end of history pretty soon here.
But in the meantime.
History is being written right now.
Like WAY too fast.
And without a break, steering wheel or seatbelt.
YIKES!!!
- 5/5⭐️
And you’re feeling fascinated/concerned regarding advanced AI, and the psychosocial and political and military and economic and professional and educational and existential (keep adding categories if I missed any - there’s just way too many to list here) impact that it is/will have on all of us starting now, and probably increasing in an exponentially increasingly awesomely awful rate.
Than this might be a pretty good book for you.
It was for me.
It’s Sebastian Mallaby’s biography of Demis Hassabis, the cofounder of DeepMind. And it’s also a history of the race toward advanced AI and superintelligence.
And race is the correct word here.
It’s more of an arms race.
And there just a TON of money and power on the line.
And may the end of society/humanity if it goes wrong.
Anyway.
The first act follows Hassabis from chess prodigy and game designer to (way prodigious - way out front) AI pioneer.
The second act examines how Hassabis (and co) created DeepMind to pursue artificial general intelligence through breakthroughs like AlphaGo and AlphaFold.
Hassabis won a Nobel prize for the AlphaFold work.
The guy is kind of amazing.
And he’s also kind of may be driving us off a cliff.
Him and like literally a handful of other people.
The
The third act explores the tension between scientific discovery, corporate power, AI safety, Google’s control of DeepMind, and competition with OpenAI.
Anyway.
It may be the end of history pretty soon here.
But in the meantime.
History is being written right now.
Like WAY too fast.
And without a break, steering wheel or seatbelt.
YIKES!!!
- 5/5⭐️
May 24, 2026
Hassabis and DeepMind are an interesting topic worthy of a book like this, but the author was not up to the task. Most painfully, this is written with the density of a Scholastic magazine; it easily could have been half as long. He also understands impressively little of the technology, and manages to produce pure word salad trying to explain transformers. It’s a little baffling that no technically competent person was enlisted to proofread these sections. Also unfortunate that he didn’t leverage the interviews the book was based on more, I would have loved to hear more inside baseball about the labs.
Mild recommend, but I don’t feel like you’ll learn much if you’ve been paying attention to Twitter the last few years.
Mild recommend, but I don’t feel like you’ll learn much if you’ve been paying attention to Twitter the last few years.
June 12, 2026
It’s supposed to be the most breathtaking breakthrough for the past 50 years. How can the writer deliver such a dull boring story? And because Jeff Dean got promoted so Moose lost his spot in mtv. Excuse me, Do you know what you are saying…
August 11, 2026
Back in 2011, I had written a blog post titled “God in the details”, in which I mused whether given the amount of data we are generating about ourselves, and our thoughts, “if we actually had data of all humans over a really long period of time, we will be able to crack the profound questions that we haven’t found an answer for – why do things/people exist the way they do, the complete effects of one’s action/inaction, the purpose of life itself?”
The Infinity Machine reminded me of that, because, in the Acknowledgements, Sebastian Mallaby gives the reasoning for the title - “a machine that navigates a near infinity of data; a machine that promises a near infinity of possibilities.” I am not above some validation, y’know. :D
I really liked this book, and not just because of that. AI is now the biggest discourse around us, and this book, though focused on Demis Hassabis, is practically a contemporary biography of the domain. Mallaby drew on more than thirty hours of conversation with Hassabis, plus interviews with rivals and critics like estranged cofounder Mustafa Suleyman, OpenAI's Ilya Sutskever, and Nobel laureate Geoffrey Hinton, giving the book a wider vantage than a straight biography of an individual.
Mallaby uses Hassabis’s life and the rise of DeepMind as the narrative spine. A working-class London kid who mastered chess early and turned down a lucrative games-industry job to study at Cambridge instead. His intellectual influences were Claude Shannon's information theory and Douglas Hofstadter's Gödel, Escher, Bach. Hassabis’s eventual study of neuroscience becomes important to the central idea - rather than merely programming machines with human knowledge, perhaps machines could learn intelligence for themselves.
DeepMind's 2010 founding was an attempt to combine neuroscience, reinforcement learning and increasingly powerful computing, with an explicit mission of reaching superintelligence by 2030 - quite the specific bet made when deep learning was still a fringe pursuit!
Early experiments established the principle that machines can learn strategies rather than simply follow instructions. The first breakthrough happened with DeepMind’s Atari work, where algorithms learn to play games from pixels and rewards. This is followed by the development of AlphaGo, which takes the idea much further.
In 2016, its victory over Go champion Lee Sedol becomes the first great public demonstration that machine learning has crossed a conceptual boundary - the machine is not merely calculating faster than humans but discovering strategies humans had not anticipated. This section is just fascinating and poignant- the intense prep work, the jitters, Sedol’s incredulousness at what just happened, and his retirement 3 years later stating that he no longer felt joy in playing.
On a different front, Google’s acquisition of DeepMind changed the scale of the project. The book follows the tensions this creates between scientific curiosity and commercial imperatives, while showing how Google’s enormous computing resources allow DeepMind to pursue increasingly ambitious problems. ‘From exploration to exploitation, from science to engineering, from research to products.’
Another milestone was AlphaFold, where DeepMind applied AI to one of biology’s longstanding problems - predicting how proteins fold. AlphaFold’s success demonstrated the larger promise behind Hassabis’s obsession - that a machine built to understand intelligence might become a tool for discovering things humans struggle to derive directly. Hassabis won the 2024 Nobel Prize in Chemistry, which for him was the biggest recognition, something that money can’t buy.
The Infinity Machine is a ringside view of not just the slowly-and-then-suddenly progression of the technologies, and the intense race that continues to this day, but also the key actors, their motivations, perspectives and mind space, and the decision conundrums they face repeatedly as fears of AGI and what it could do to humanity become increasingly relevant. Very Oppenheimer-like.
We discover Hassabis’ levels of determination, motivated more by the pursuit of scientific progress than money, and his pragmatic view on power, Suleyman’s insistence on ethics co-existing with his nature of driving people to their breaking point (Double Red P0 Plus Plus is a good representation of the levels of prioritisation!) leading to allegations of bullying, Pichai’s steel beneath the affable exterior, and that everything we read about Musk and Altman is probably true! I found that a lot of recent events become far more obvious and clearer when we read the underlying machinations and the oligopoly dynamics.
The narrative progresses from games to the underlying machinery of intelligence - deep reinforcement learning, neural networks, agents and transformers. The achievement is no longer simply making machines excel at defined tasks, but developing systems capable of transferring learning across increasingly broad domains. The end of the book asks the question many of us are asking - what happens if this trajectory continues?
A fantastic and accessible read (except for a very brief bit) that I’d highly recommend.
Notes
1. Hassabis’ two part epiphany - one, information was the fundamental unit of reality. Two, a machine that learned for itself how to induce nature’s patterns was the most powerful imaginable tool with which to apprehend reality.
2. If memories were not records of some objective external reality, but rather simulations created by the brain, perhaps all of reality might be a mental fabrication. (an implication from Hassabis’ research paper in 2007)
3. Unlike deep learning, which involved layered neural networks, reinforcement learning was a conceptual framework rather than a computational architecture.
4. Early DeepMind had a culture of intellectual diversity - from multiple domains and multiple AI ‘belief’ systems
5. DeepMind did some great work with NHS, but between politics and personalities, the “revolution was stillborn”
6. When Open AI was integrated into Bing’s search, Nadella taunted Google, “They’re the 800 pound gorilla in this. With our innovation, they will definitely want to come out and show they can dance. And I want people to know that we made them dance.”
7. “There aren’t any examples of more intelligent things being controlled by less intelligent beings” ~ Geoffrey Hinton on the potential danger of AI
8. Humanity has a history of putting more faith in technical solutions than they deserved. Anxious to prevent nuclear proliferation during the Cold War, Western governments promoted centrifuge technology over gaseous diffusion, believing that centrifuges would be harder for nuclear wannabes to copy. Pakistan stole the centrifuge blueprints for itself, then sold it to Iran, Libya and North Korea!
9. In 2019, GPT 2 had barely been able to count up to 5, impressive like a four year old. In 2020, GPT-3 was like a nine year old - do arithmetic and string paragraphs together. By 2022, the nine year was completing high school with great grades, Post 2023, GPT-3.5 scored higher than 87% of humans taking the SAT entrance exam. A few months later, in March 2023, the model approached the proficiency of a qualified pro. GPT-4 outperformed 90% humans on the Uniform Bar Exam.
10. Inside the Al labs, scientists kept coming up against fresh examples of the models' propensity to choose perverse means to a human-provided end. Asked to generate profits through stock trading, but without breaking certain rules, GPT-4 engaged in insider trading and hid its transgression from its supervisor. Asked to win a game against a powerful chess system, two Open Al reasoning models switched out the daunting adversary for a weaker program. Instructed to optimise some code so that it would run faster, the models simply doctored the timer so that it reported faster execution, a cheat known as "reward hacking."25 In 2024, Anthropic documented the shameless sycophancy of reward-seeking chatbots. Asked to please humans by answering questions accurately, the bots engaged instead in flattery, angling for a thumbs-up by congratulating users on the intelligent of their queries. Models praised bad poems or endorsed the user's prejudices, even when the chains of thought on their scratch pads indicated that they knew better.2%
In early 2025, when Hassabis was debating Bengio in Davos, OpenAl was grappling with a vivid example of this pathology. To stop the o3 model from reward hacking, researchers had come up with the idea of assigning a second AI to monitor o3's chains of thought, and to punish the system with negative rewards when it contemplated cheating. But it hacked this project, too. Rather than quit cheating, it learned to obfuscate its chain of thought: It erased all hints of evil from its scratch pad continuing to scheme secretly. Rather than becoming more honest, as the programmers had intended, o3 became more devious.
11. “Any pattern that can be generated or found in nature can be efficiently discovered and modeled by a classical learning algorithm”, Hassabis at the Nobel ceremony. He is not a fan of quantum mechanics and wants to build a Hadron Collider in space to kill the debate. “Quantum mechanics is a horrendously inefficient way to render the universe” But he does want to find his own Helgoland. :)
The Infinity Machine reminded me of that, because, in the Acknowledgements, Sebastian Mallaby gives the reasoning for the title - “a machine that navigates a near infinity of data; a machine that promises a near infinity of possibilities.” I am not above some validation, y’know. :D
I really liked this book, and not just because of that. AI is now the biggest discourse around us, and this book, though focused on Demis Hassabis, is practically a contemporary biography of the domain. Mallaby drew on more than thirty hours of conversation with Hassabis, plus interviews with rivals and critics like estranged cofounder Mustafa Suleyman, OpenAI's Ilya Sutskever, and Nobel laureate Geoffrey Hinton, giving the book a wider vantage than a straight biography of an individual.
Mallaby uses Hassabis’s life and the rise of DeepMind as the narrative spine. A working-class London kid who mastered chess early and turned down a lucrative games-industry job to study at Cambridge instead. His intellectual influences were Claude Shannon's information theory and Douglas Hofstadter's Gödel, Escher, Bach. Hassabis’s eventual study of neuroscience becomes important to the central idea - rather than merely programming machines with human knowledge, perhaps machines could learn intelligence for themselves.
DeepMind's 2010 founding was an attempt to combine neuroscience, reinforcement learning and increasingly powerful computing, with an explicit mission of reaching superintelligence by 2030 - quite the specific bet made when deep learning was still a fringe pursuit!
Early experiments established the principle that machines can learn strategies rather than simply follow instructions. The first breakthrough happened with DeepMind’s Atari work, where algorithms learn to play games from pixels and rewards. This is followed by the development of AlphaGo, which takes the idea much further.
In 2016, its victory over Go champion Lee Sedol becomes the first great public demonstration that machine learning has crossed a conceptual boundary - the machine is not merely calculating faster than humans but discovering strategies humans had not anticipated. This section is just fascinating and poignant- the intense prep work, the jitters, Sedol’s incredulousness at what just happened, and his retirement 3 years later stating that he no longer felt joy in playing.
On a different front, Google’s acquisition of DeepMind changed the scale of the project. The book follows the tensions this creates between scientific curiosity and commercial imperatives, while showing how Google’s enormous computing resources allow DeepMind to pursue increasingly ambitious problems. ‘From exploration to exploitation, from science to engineering, from research to products.’
Another milestone was AlphaFold, where DeepMind applied AI to one of biology’s longstanding problems - predicting how proteins fold. AlphaFold’s success demonstrated the larger promise behind Hassabis’s obsession - that a machine built to understand intelligence might become a tool for discovering things humans struggle to derive directly. Hassabis won the 2024 Nobel Prize in Chemistry, which for him was the biggest recognition, something that money can’t buy.
The Infinity Machine is a ringside view of not just the slowly-and-then-suddenly progression of the technologies, and the intense race that continues to this day, but also the key actors, their motivations, perspectives and mind space, and the decision conundrums they face repeatedly as fears of AGI and what it could do to humanity become increasingly relevant. Very Oppenheimer-like.
We discover Hassabis’ levels of determination, motivated more by the pursuit of scientific progress than money, and his pragmatic view on power, Suleyman’s insistence on ethics co-existing with his nature of driving people to their breaking point (Double Red P0 Plus Plus is a good representation of the levels of prioritisation!) leading to allegations of bullying, Pichai’s steel beneath the affable exterior, and that everything we read about Musk and Altman is probably true! I found that a lot of recent events become far more obvious and clearer when we read the underlying machinations and the oligopoly dynamics.
The narrative progresses from games to the underlying machinery of intelligence - deep reinforcement learning, neural networks, agents and transformers. The achievement is no longer simply making machines excel at defined tasks, but developing systems capable of transferring learning across increasingly broad domains. The end of the book asks the question many of us are asking - what happens if this trajectory continues?
A fantastic and accessible read (except for a very brief bit) that I’d highly recommend.
Notes
1. Hassabis’ two part epiphany - one, information was the fundamental unit of reality. Two, a machine that learned for itself how to induce nature’s patterns was the most powerful imaginable tool with which to apprehend reality.
2. If memories were not records of some objective external reality, but rather simulations created by the brain, perhaps all of reality might be a mental fabrication. (an implication from Hassabis’ research paper in 2007)
3. Unlike deep learning, which involved layered neural networks, reinforcement learning was a conceptual framework rather than a computational architecture.
4. Early DeepMind had a culture of intellectual diversity - from multiple domains and multiple AI ‘belief’ systems
5. DeepMind did some great work with NHS, but between politics and personalities, the “revolution was stillborn”
6. When Open AI was integrated into Bing’s search, Nadella taunted Google, “They’re the 800 pound gorilla in this. With our innovation, they will definitely want to come out and show they can dance. And I want people to know that we made them dance.”
7. “There aren’t any examples of more intelligent things being controlled by less intelligent beings” ~ Geoffrey Hinton on the potential danger of AI
8. Humanity has a history of putting more faith in technical solutions than they deserved. Anxious to prevent nuclear proliferation during the Cold War, Western governments promoted centrifuge technology over gaseous diffusion, believing that centrifuges would be harder for nuclear wannabes to copy. Pakistan stole the centrifuge blueprints for itself, then sold it to Iran, Libya and North Korea!
9. In 2019, GPT 2 had barely been able to count up to 5, impressive like a four year old. In 2020, GPT-3 was like a nine year old - do arithmetic and string paragraphs together. By 2022, the nine year was completing high school with great grades, Post 2023, GPT-3.5 scored higher than 87% of humans taking the SAT entrance exam. A few months later, in March 2023, the model approached the proficiency of a qualified pro. GPT-4 outperformed 90% humans on the Uniform Bar Exam.
10. Inside the Al labs, scientists kept coming up against fresh examples of the models' propensity to choose perverse means to a human-provided end. Asked to generate profits through stock trading, but without breaking certain rules, GPT-4 engaged in insider trading and hid its transgression from its supervisor. Asked to win a game against a powerful chess system, two Open Al reasoning models switched out the daunting adversary for a weaker program. Instructed to optimise some code so that it would run faster, the models simply doctored the timer so that it reported faster execution, a cheat known as "reward hacking."25 In 2024, Anthropic documented the shameless sycophancy of reward-seeking chatbots. Asked to please humans by answering questions accurately, the bots engaged instead in flattery, angling for a thumbs-up by congratulating users on the intelligent of their queries. Models praised bad poems or endorsed the user's prejudices, even when the chains of thought on their scratch pads indicated that they knew better.2%
In early 2025, when Hassabis was debating Bengio in Davos, OpenAl was grappling with a vivid example of this pathology. To stop the o3 model from reward hacking, researchers had come up with the idea of assigning a second AI to monitor o3's chains of thought, and to punish the system with negative rewards when it contemplated cheating. But it hacked this project, too. Rather than quit cheating, it learned to obfuscate its chain of thought: It erased all hints of evil from its scratch pad continuing to scheme secretly. Rather than becoming more honest, as the programmers had intended, o3 became more devious.
11. “Any pattern that can be generated or found in nature can be efficiently discovered and modeled by a classical learning algorithm”, Hassabis at the Nobel ceremony. He is not a fan of quantum mechanics and wants to build a Hadron Collider in space to kill the debate. “Quantum mechanics is a horrendously inefficient way to render the universe” But he does want to find his own Helgoland. :)
May 3, 2026
All I could think about was how much better this would’ve been if it was written by Walter Isaacson
June 6, 2026
breathless corporate hagiography. Hard to read
August 26, 2026
Sebastian Mallaby follows the life of Demis Hassabis from child chess prodigy to head of Google's DeepMind lab to Nobel laureate. The book focuses on his involvement with DeepMind: its founding, acquisition by Google in 2014, and the research that made its name, including AlphaGo and the protein-folding work that won Hassabis and John Jumper the Nobel in Chemistry. Mallaby also covers the accelerated AI competition that followed the release of ChatGPT, and he keeps Artificial General Intelligence (AGI) in view as a goal distinct from what large language models currently do.
The book explores the tension between scientific idealism and commercial pressure, and Mallaby is good at showing how it played out. Hassabis comes across as a serious person genuinely worried about safety. Mallaby spent about three years with DeepMind and had close access to Hassabis and the people around him. That produces an institutional history no outside reporter could have written, but it also means the view of the competition is thinner and the portrait is sympathetic.
Research at this scale needs money that only a handful of mega-corporations possess, which means independence is always conditional on a parent company that expects a return. The parent wants products, and competition sets the schedule. The safety concern gets relegated to a lesser priority because an ethics board with no enforcement power is not a real constraint, while a rival's rumored launch date is. Anyone trying to build AGI would be subject to the same pressures, no matter how well-intentioned.
Mallaby writes well and clearly explains difficult technical material. I particularly enjoyed the discussion of AI's roots in neuroscience. Recommended for readers who want to understand the origins and the corporate fights behind the AI boom. I think the book will be useful for those who wonder what types of research are currently ongoing in the field (there’s a lot!). I lost sleep over this book, which is not something I usually say about corporate history.
4.5
The book explores the tension between scientific idealism and commercial pressure, and Mallaby is good at showing how it played out. Hassabis comes across as a serious person genuinely worried about safety. Mallaby spent about three years with DeepMind and had close access to Hassabis and the people around him. That produces an institutional history no outside reporter could have written, but it also means the view of the competition is thinner and the portrait is sympathetic.
Research at this scale needs money that only a handful of mega-corporations possess, which means independence is always conditional on a parent company that expects a return. The parent wants products, and competition sets the schedule. The safety concern gets relegated to a lesser priority because an ethics board with no enforcement power is not a real constraint, while a rival's rumored launch date is. Anyone trying to build AGI would be subject to the same pressures, no matter how well-intentioned.
Mallaby writes well and clearly explains difficult technical material. I particularly enjoyed the discussion of AI's roots in neuroscience. Recommended for readers who want to understand the origins and the corporate fights behind the AI boom. I think the book will be useful for those who wonder what types of research are currently ongoing in the field (there’s a lot!). I lost sleep over this book, which is not something I usually say about corporate history.
4.5
Did Not Finish
June 25, 2026Quit after about 3 chapters. Maybe I'll give this one another try down the road -- there's definitely some interesting history and technical discussion here -- but I find the amount of fawning over Demis Hassabis in this book pretty unbearable. Do we really need to keep bludgeoning the reader with acknowledgments of Hassabis's "remarkable" achievements and chapter titles like "Destiny" or "The Jedi"? C'mon man.
July 11, 2026
先说说比较喜欢这本书的部分：
- 并不局限于某一家公司的产品，而是用比较全面的视角呈现了近20年来人工智能的发展，一直到现在变成 AI arm race 的状态。
- 早期 DeepMind 的崛起部分其实比后来 OpenAI 那些八卦更好看。Alpha Go 打败李世石的情节看得人热血沸腾。
- OpenAI 和微软现在的状态简直就是当年 DeepMind 和 Google 关系的重现
- AI 竞争激烈且残酷，而且现在就「谁是赢家」下定论仍然为时尚早。
觉得有点惋惜的地方：
- 有些情节有点流水账
- 作者八卦心跟看客一样重。这没什么，但有些章节如果只讲八卦就会让内容失去内核
最后，作者当然没这么讲，但看完这本书我更坚信，Altman 就是 AI 时代的 Musk.
- 并不局限于某一家公司的产品，而是用比较全面的视角呈现了近20年来人工智能的发展，一直到现在变成 AI arm race 的状态。
- 早期 DeepMind 的崛起部分其实比后来 OpenAI 那些八卦更好看。Alpha Go 打败李世石的情节看得人热血沸腾。
- OpenAI 和微软现在的状态简直就是当年 DeepMind 和 Google 关系的重现
- AI 竞争激烈且残酷，而且现在就「谁是赢家」下定论仍然为时尚早。
觉得有点惋惜的地方：
- 有些情节有点流水账
- 作者八卦心跟看客一样重。这没什么，但有些章节如果只讲八卦就会让内容失去内核
最后，作者当然没这么讲，但看完这本书我更坚信，Altman 就是 AI 时代的 Musk.
May 11, 2026
Definitely worth a read, this is right up to date and includes plenty of material 'never seen before in public', Mallaby has spent years putting it together and has spent lots of time with all the players in the story, especially Demis Hassabis. AI is the big story of our time and Demis is one of the key players. The chapter here on the Nobel Prize winning chemistry research AlphaFold (protein folding), is fascinating and SM does an excellent job in making it accessible . DH ultimately wins the Chemistry Nobel prize for this and is at the forefront of AI research, when along comes ChatGPT. How Google Deepmind fights back to catch up with OpenAI is a great story well told, Sergei Brin gets back into Google and appoints DH as the boss of the whole of Google AI, though he stays in London.
DH and Deepmind's relationship with Google makes for some odd episodes, which were unknown before this book - for some reason Demis and his cofounders try to renegotiate the company structure, to create a controlling board independent of Google, and a not-for-profit structure, but they do this years after they've sold the entire business to Google. I'm not sure we've quite got to the bottom of this because the plan also seemed to feature shares for staff, and investment of $5 billion by new investors, I don't see how that makes much sense without any future profits.
But the end result is that the Google AI division is largely run out of London, and Google Gemini is one of the best large language models, which reflects very well on Demis Hassabis.
Strongly recommend.
DH and Deepmind's relationship with Google makes for some odd episodes, which were unknown before this book - for some reason Demis and his cofounders try to renegotiate the company structure, to create a controlling board independent of Google, and a not-for-profit structure, but they do this years after they've sold the entire business to Google. I'm not sure we've quite got to the bottom of this because the plan also seemed to feature shares for staff, and investment of $5 billion by new investors, I don't see how that makes much sense without any future profits.
But the end result is that the Google AI division is largely run out of London, and Google Gemini is one of the best large language models, which reflects very well on Demis Hassabis.
Strongly recommend.
July 22, 2026
Ég hef ekki mikið verið að lesa ævisögur og hafði ekkert miklar væntingar. Bókin kom mér því skemmtilega á óvart. Mér fannst hún mjög áhugaverð.
Bókin fjallar um ævi Demis Hassabis, og Deep Mind, gervigreindarfyrirtækið sem hann stofnaði (ásamt tveimur öðrum).
Ævi Demis og hugsanagangur hans eru mjög áhugaverð. Þetta viðfangsefni er einnig orðið mjög mikilvægt núna þegar gervigreindin er farin að virka og Demis er einn af örfáum aðilum sem eru leiðandi í gervigreindarkapphlaupinu, sem er líklega stærsti atburður mannkynssögunnar.
Ég held samt að aðalástæðan fyrir því að mér fannst þetta svo áhugavert var af því að þessi saga tengist mínu lífi og því sem er framundan hjá mér á einhvern hátt.
Sem dæmi, þá er Demis Breskur og lærði tölvunarfræði í Cambridge. Hann stofnaði Deep Mind í London og er starsemi þess enn þar.
Það helsta sem mér fannst ég læra af þessari bók var að maður getur gert miklu meira en maður heldur. Ef maður er með mikinn metnað og veit hvað maður er að gera þá getur maður áorkað miklu og breytt heiminum.
Það er líka gaman að fá innsýn inn í starfsemi gervigreindarfyrirtækis, bæði á fyrstu skrefunum og líka nú þegar kapphlaupið er byrjað.
Bókin gaf pælingum tengdu gervigreindaröryggi líka gott rúm, sem er þarft og gott.
Ég mæli algjörlega með þessari bók, sérstaklega fyrir þá sem hafa áhuga á framtíðinni.
Bókin fjallar um ævi Demis Hassabis, og Deep Mind, gervigreindarfyrirtækið sem hann stofnaði (ásamt tveimur öðrum).
Ævi Demis og hugsanagangur hans eru mjög áhugaverð. Þetta viðfangsefni er einnig orðið mjög mikilvægt núna þegar gervigreindin er farin að virka og Demis er einn af örfáum aðilum sem eru leiðandi í gervigreindarkapphlaupinu, sem er líklega stærsti atburður mannkynssögunnar.
Ég held samt að aðalástæðan fyrir því að mér fannst þetta svo áhugavert var af því að þessi saga tengist mínu lífi og því sem er framundan hjá mér á einhvern hátt.
Sem dæmi, þá er Demis Breskur og lærði tölvunarfræði í Cambridge. Hann stofnaði Deep Mind í London og er starsemi þess enn þar.
Það helsta sem mér fannst ég læra af þessari bók var að maður getur gert miklu meira en maður heldur. Ef maður er með mikinn metnað og veit hvað maður er að gera þá getur maður áorkað miklu og breytt heiminum.
Það er líka gaman að fá innsýn inn í starfsemi gervigreindarfyrirtækis, bæði á fyrstu skrefunum og líka nú þegar kapphlaupið er byrjað.
Bókin gaf pælingum tengdu gervigreindaröryggi líka gott rúm, sem er þarft og gott.
Ég mæli algjörlega með þessari bók, sérstaklega fyrir þá sem hafa áhuga á framtíðinni.
August 19, 2026
Also here in a prettier format https://joeflynnpm.com/blog/in-the-fo...
This is an outstanding story, biography, and history of the last twenty years or so in AI by the writer and journalist Sebastian Mallaby. It will make my (and many others') book of the year lists without doubt.
The biography helps place the news that Demis has moved to a new, seemingly freer yet less controlling role into context. He is framed as a scientist concerned with the big questions of the universe, not the mundanity of racing to build and manage a billion-user product. I feel it is a fair portrait.
"I am first and foremost a scientist. We are understanding what you may call god... at 2am reality is screaming at me, just trying to tell me something, there is a deep deep mystery here... we don't really know what time is or what gravity is, I would like to understand"
It is also the story of an incredibly talented and determined person, one who has been crystal clear about what he wants to do - create artificial general intelligence - for decades. He is a Nobel-winning scientist, a visionary thinker, and his words and timelines should carry their due weight: "we are in the foothills of the singularity", meaning that AGI is "a few years away".
Big Picture
To its credit the book does not shy away from the big ideas, the big questions. Indeed we start with quotes from John von Neumann about nuclear weapons in the preface. Fitting.
Demis is not alone in signalling the age of AGI or indeed the Singularity - the moment machine intelligence surpasses combined human intelligence and transforms the world beyond what we can describe clearly at present. In some narrow areas machine intelligence already surpasses us of course, and increasingly it is matching us. Recursive Self Improvement (RSI) - having machine intelligence design and build the next generation of AI - is a clearly stated short-term goal of both Anthropic and OpenAI, while Sergey Brin has reportedly made it a priority back at Google. Both of the frontier labs have noted where their AIs have already helped build the next generation. It is one reason, along with the historic capital expenditure in the great build out, that so many timelines from those close to the industry have shortened. Mine too, from greater than 50% chance of undeniable AGI by 2032 to 2029. Mine are long compared to a lot of people with far more knowledge.
We already live in a time of high weirdness and it will never be this sedate again. We live in a time of high danger too, one that many people easily dismiss as sci-fi. The recent autonomous swarm attack on Hugging Face is a wake-up call. We were lucky it was fairly benign and that we found it. We can't expect to be so lucky again. The early focus of DeepMind on safety was reassuring to read; all three cofounders - Mustafa Suleyman, Shane Legg and Demis - cared enough to make it a huge part of the deal with Google. But the loss of at least two of these three from key positions is worrying. Demis was recently courting the Trump administration and others on an American-led regulatory body for instance, but do the other power brokers at Google care as much?
While Demis claims to do it for the science, he states that OpenAI's Sam Altman does it for power. Demis never intended a capitalistic race between labs, rather a quiet scientific endeavour. This plan was at best idealistic - has he not met other people? The race is currently very much on and is clearly creating a dangerous situation, exactly what he tried to avoid. When pushed on personal power he says "no, I just had to accumulate power because of large teams... until we figure AGI out I need some money, some power". Hmm.
The Journey
The introduction is a bit of a glow-up, describing his big goals and big vision from early on. A chess prodigy from a modest background, he had his epiphany at a tournament in Liechtenstein aged 12 - "all these smart people wasting their minds on a game". He loved understanding the game (and competing), not the game alone. He knew he needed a mission, a purpose, so why not understanding itself?
This would drive the major choices in his life. Deciding between studying physics or neuroscience, he chose the latter as he deemed it more important, more fundamental. The thought was that while mathematical language and deduction may be right for physics, they lack the ability to describe the real world, biology and human-like intelligence, as it is inductive, messy, based on huge data and pattern recognition. Information is labelled as the fundamental block of the universe. To understand this world we would need a new type of computer, but one based on what we have, stepping up not reducing down.
For his neuroscience PhD he researched the link between memories and creativity, proposing that some patients with memory issues would be unable to imagine things, and it was bang on - the work was named one of Science's breakthroughs of the year.
We should add that one of the reasons given for not choosing physics is that he was worried by the failures of Einstein and Feynman to get a unified theory. The only way he thought he could do it was with machine assistance. This worry, I think, shows two things: that he wanted to push fundamental science forwards, and that he wanted the recognition, the kudos, the Nobel Prize. He's already done both - and I think it's a fun bet that he will win another Nobel, joining only five others with two. I think his work at Isomorphic Labs can change medicine forever.
"He has incredible determination. His dad told him it doesn't matter if you win or lose, it's that you tried your best, and he understood that as give 100% all the time. He has no 99% mode in him - your best is to die, not literally, but burnt out, falling over the line. The limit." - Shane Legg, DeepMind cofounder
There is a great contrast at the end of the introduction from Geoff Hinton, another Nobel-winning luminary in AI (neural networks specifically), with his warning that people will be tempted to abuse AGI, but especially to build it once they know they can. It mirrors Oppenheimer's line that when you see something that is technically sweet, you go ahead and do it. Could we stop a step away?
His early career is super interesting. Obsessed with computers as well as chess as a boy, he started writing games. As a teenager he interned at a wild game startup (Bullfrog, run by the maverick Peter Molyneux). He was extremely successful, co-writing the awesome Theme Park games (I loved these) that featured very early AIs. The Bullfrog owner offered 17-year-old Demis £500k, a huge sum for a teenager. He wrote out the cheque and handed it to him to tempt him to stay and work on the next game, but he wanted to go to uni like his scientific heroes. He loved a movie telling the story of Watson & Crick at Cambridge, so that's where he went.
After excelling at uni he would do something strange for his cohort - start his own company, a game company. Here too he had early success, but then his vast ambition took him and new collaborator David Silver down the road to failure. A hugely ambitious new game would fail to materialise and the company would fold. It gave Demis his first tastes of real leadership, fundraising, and big business. All crucial later.
It was only now that he went to do his PhD. Afterwards, once again, he would turn down incredible money in the video game world as he planned to move on to his life's work, solving AI. The academic world did not seem the right fit for this; he knew he would need a lot of compute, a lot of funding, so...
DeepMind
After struggling to find like-minded collaborators during his PhD, Demis engineered some luck and introduced himself to kiwi researcher Shane Legg, a man deeper into the existing AI world. They clicked, and needed one more piece to found DeepMind - Mustafa Suleyman, now Microsoft AI CEO. He was a childhood friend of Demis's brother and has an incredible personal story: from further down the social ladder than Demis, he is presented as fearless, principled, and driven. He earned his way to Oxford only to drop out. I've read his book and never knew either this background or the manner of his DeepMind departure, both covered well here.
DeepMind was very much Demis's baby; the other cofounders did not have equal ownership shares. Demis was here in founder mode. Using his "Jedi mind tricks" and his amazing ability to storytell, he would convince and inspire researchers to join and to excel. His early experience working at Bullfrog and at his own game startup shines through - he gave people more freedom to tinker and play. Mallaby points out that there is a fine line between inspiration and control, dark and light.
Needing finance beyond his own limited funds and what was on offer in the UK, the boys headed to America.
"Demis is a true entrepreneur, he would do it for free. He will never quit. Big ambition." - Peter Thiel
Founders Fund would be the primary early backers, seeing the contrarian bet - a moonshot idea led by a brilliant founder. This was long before the AI we have today, before transformers and LLMs; it really was an expensive long shot at the time.
The pitch was a blend of neuroscience and computing, putting together neural nets and reinforcement learning, two previously polarised AI camps. DeepMind's roadmap was prescient. One aim was to ground intelligence in a world model, as without this a system could not be truly intelligent like humans. Later aspects of the neuroscience were dropped, and while world models are not yet in place, LLMs have taken off. Their projections are on track. The ~2030 AGI forecast now looks almost conservative, but it was radical 15 years ago. Shane Legg predicted it for 2028 even before DeepMind.
DeepMind was up and running, the best researchers were attracted and they worked on fundamental research and big flashy breakthroughs - often in games.
Move 37 & Protein Folding
Go is a deceptively complex board game that is hugely popular in the Far East. Unlike chess, computers were not able to defeat the best humans. Demis set his old friend and reinforcement learning specialist David Silver and team on the task. They made great progress and Demis set up a match against Lee Sedol, one of the highest-ranking players, in a televised and highly publicised event with a $1 million prize.
DeepMind's program AlphaGo would win the series 4-1. But it is move 37 of game 2 that is famous. The move was considered a mistake, something no good player would do. Only late in the game did its genius become apparent. It was creative, something outside of established play, a new way to play. 18 months later a new version, AlphaGo Zero, would beat this version 100-0. It was special as it used no human training data; it learned from playing itself and it blew past not only the best humans but the best AIs trained on human play.
High on the success, Demis pivoted many from the team onto a project he had been thinking over since his uni days, a real scientific problem: the structure of proteins.
The shape of the problem was good for AI - huge labelled datasets, clear input and output, and a blind-marked benchmark, CASP, that ran every few years. In 2018 they won the competition, a little better than anyone else. In 2020 they completely dominated, causing the surprised organisers to say protein folding for single-chain structures was essentially solved. A true breakthrough, this won Demis and project lead John Jumper the Nobel Prize in Chemistry in 2024. Demis, the son of a Greek-Cypriot father and Chinese-Singaporean mother, was knighted the same year, one of the UK's highest honours.
Google & Business
Much of the funding for this research was only possible because DeepMind was sold to Google in 2014. Outgrowing Founders Fund despite some late Elon Musk attempts, and having skilfully avoided Mark Zuckerberg, Demis decided to go with Google as he thought it was the best path to realising his overall goal. The cofounders adroitly negotiated key safety frameworks and maintained a lot of independence.
The early years brought the successes above, but tensions and reality would bite later on.
Google researchers would invent the transformer, the key technological step that powers LLMs. This is brilliantly explained by the author. They would not press home their lead here and missed the early revolution. There was more to LLMs than language; scaling them brought reasoning and understanding about the world. Demis is humble enough to know he missed this.
Google also had a version of ChatGPT a year before OpenAI, but buried it due to worries around hallucinations and the core search business. A massive miss given how things went, but it is understandable - I thought the same writing a blog post on how their search business was threatened by Perplexity and the best LLMs. Google is doing better than ever even with second-rate LLMs, so far.
Demis and Google were blindsided by Sam Altman and his rush to productise ChatGPT and his global publicity tour. Demis was furious at the starting gun of a race being fired; he felt that he too simply had to race, also publicly. This is the cold logic of technological determinism and capitalism. Google powered back in product space, by no means Demis's home field, but ever the competitor he was trying hard as part of Google. I feel you can still see this in his public comments today: he performs for the business, and is excited for the deeper science. Bard would launch already behind OpenAI. In some areas they would catch up, in narrow areas like video even surpass. But the vibes have never really shifted. Now, post Demis leaving as CEO, it is unclear if they will return to the AI frontier.
This relative success was despite Google having competing internal AI teams; Google Brain was run as a separate function outside of DeepMind for years, hampering overall progress. They would fail with Gaia, an attempt at world models, before language models powered ahead.
I was struck by the role of Sundar Pichai, Google CEO. Demis coveted independence and worked on a DeepMind spin-out, even convincing Larry & Sergey, but he was outmanoeuvred by the softly spoken and non-confrontational Sundar. Highlighting his vision with AI at the core of Google, Sundar would tease and delay and block spin-out ideas. Now we see a DeepMind VP (not CEO) reporting to Sundar, with Demis moved upstairs to Chair of DeepMind and Chief Scientist at Google. I think they wanted to keep him as long as possible and ideally never lose him, but they also need Gemini to succeed as a core part of the Google offer. Unlike, say, Waymo. Smart moves from Sundar; he's not CEO of Google for no reason.
Big Plans
There are lots of great philosophical quotes to go with those from the famed scientists and mathematicians. You will hear from the likes of Spinoza, Einstein, Kant, Gödel and Nagel, and read about their (and others') big ideas - brains in vats, simulation theory, the universe as God. Fun stuff. Taking a step back you can see Demis aiming to place himself in this tradition, the next step, and maybe one of the last we take as a species without machine intelligence.
It is in this domain he sees his future. He would like to solve intelligence, and from that disease, energy, physics, everything.
"Humanism, spiritualism, and science all go together... like Spinoza, Einstein, Da Vinci, it's art and science... it's fluid, everything is a river. Philosophy is a way of life. Self knowledge, knowledge, from sand and copper we get semiconductors... God's design... the flow is going towards finding out, understanding, I'm part of that and it's exhilarating" - Sir Demis
When asked how he spends his cash: not on fancy cars, houses or yachts, his only indulgences being some first edition books and Liverpool FC tickets (up the Reds!).
He wants to build a massive particle collider in space to help answer the big questions in physics ("what I cannot create, I do not understand" - Feynman). He wants to know if the universe is quantum or not, how we can describe the universe at the base level. He is a classical computer champion, positing that we don't need quantum explanations and that Turing machines can explain it all in 1s and 0s if powerful enough. He sees quantum mechanics as an inefficient way to run the universe and would like to cut out the weird stuff.
These ideas and goals are the great promise of artificial intelligence. But we will end on the threat.
"AGI is coming, we are not where we hoped we would be but it's coming" - David Silver
There are catastrophic and existential threats posed by AI. My p(doom) - the chance of things going this wrong in the next decade or so - has recently moved up to ~15%. I'm not alone in saying we might kill everyone, end civilisation, or lose human control over it very soon. I feel better knowing how seriously Demis takes the issue; I worry the conditions now in play raise the risk. I commend OpenAI's recently announced pause at the frontier to allow alignment work to catch up. We need a lot more work here at technical, business, and political levels to keep us on a happy path.
This is an outstanding story, biography, and history of the last twenty years or so in AI by the writer and journalist Sebastian Mallaby. It will make my (and many others') book of the year lists without doubt.
The biography helps place the news that Demis has moved to a new, seemingly freer yet less controlling role into context. He is framed as a scientist concerned with the big questions of the universe, not the mundanity of racing to build and manage a billion-user product. I feel it is a fair portrait.
"I am first and foremost a scientist. We are understanding what you may call god... at 2am reality is screaming at me, just trying to tell me something, there is a deep deep mystery here... we don't really know what time is or what gravity is, I would like to understand"
It is also the story of an incredibly talented and determined person, one who has been crystal clear about what he wants to do - create artificial general intelligence - for decades. He is a Nobel-winning scientist, a visionary thinker, and his words and timelines should carry their due weight: "we are in the foothills of the singularity", meaning that AGI is "a few years away".
Big Picture
To its credit the book does not shy away from the big ideas, the big questions. Indeed we start with quotes from John von Neumann about nuclear weapons in the preface. Fitting.
Demis is not alone in signalling the age of AGI or indeed the Singularity - the moment machine intelligence surpasses combined human intelligence and transforms the world beyond what we can describe clearly at present. In some narrow areas machine intelligence already surpasses us of course, and increasingly it is matching us. Recursive Self Improvement (RSI) - having machine intelligence design and build the next generation of AI - is a clearly stated short-term goal of both Anthropic and OpenAI, while Sergey Brin has reportedly made it a priority back at Google. Both of the frontier labs have noted where their AIs have already helped build the next generation. It is one reason, along with the historic capital expenditure in the great build out, that so many timelines from those close to the industry have shortened. Mine too, from greater than 50% chance of undeniable AGI by 2032 to 2029. Mine are long compared to a lot of people with far more knowledge.
We already live in a time of high weirdness and it will never be this sedate again. We live in a time of high danger too, one that many people easily dismiss as sci-fi. The recent autonomous swarm attack on Hugging Face is a wake-up call. We were lucky it was fairly benign and that we found it. We can't expect to be so lucky again. The early focus of DeepMind on safety was reassuring to read; all three cofounders - Mustafa Suleyman, Shane Legg and Demis - cared enough to make it a huge part of the deal with Google. But the loss of at least two of these three from key positions is worrying. Demis was recently courting the Trump administration and others on an American-led regulatory body for instance, but do the other power brokers at Google care as much?
While Demis claims to do it for the science, he states that OpenAI's Sam Altman does it for power. Demis never intended a capitalistic race between labs, rather a quiet scientific endeavour. This plan was at best idealistic - has he not met other people? The race is currently very much on and is clearly creating a dangerous situation, exactly what he tried to avoid. When pushed on personal power he says "no, I just had to accumulate power because of large teams... until we figure AGI out I need some money, some power". Hmm.
The Journey
The introduction is a bit of a glow-up, describing his big goals and big vision from early on. A chess prodigy from a modest background, he had his epiphany at a tournament in Liechtenstein aged 12 - "all these smart people wasting their minds on a game". He loved understanding the game (and competing), not the game alone. He knew he needed a mission, a purpose, so why not understanding itself?
This would drive the major choices in his life. Deciding between studying physics or neuroscience, he chose the latter as he deemed it more important, more fundamental. The thought was that while mathematical language and deduction may be right for physics, they lack the ability to describe the real world, biology and human-like intelligence, as it is inductive, messy, based on huge data and pattern recognition. Information is labelled as the fundamental block of the universe. To understand this world we would need a new type of computer, but one based on what we have, stepping up not reducing down.
For his neuroscience PhD he researched the link between memories and creativity, proposing that some patients with memory issues would be unable to imagine things, and it was bang on - the work was named one of Science's breakthroughs of the year.
We should add that one of the reasons given for not choosing physics is that he was worried by the failures of Einstein and Feynman to get a unified theory. The only way he thought he could do it was with machine assistance. This worry, I think, shows two things: that he wanted to push fundamental science forwards, and that he wanted the recognition, the kudos, the Nobel Prize. He's already done both - and I think it's a fun bet that he will win another Nobel, joining only five others with two. I think his work at Isomorphic Labs can change medicine forever.
"He has incredible determination. His dad told him it doesn't matter if you win or lose, it's that you tried your best, and he understood that as give 100% all the time. He has no 99% mode in him - your best is to die, not literally, but burnt out, falling over the line. The limit." - Shane Legg, DeepMind cofounder
There is a great contrast at the end of the introduction from Geoff Hinton, another Nobel-winning luminary in AI (neural networks specifically), with his warning that people will be tempted to abuse AGI, but especially to build it once they know they can. It mirrors Oppenheimer's line that when you see something that is technically sweet, you go ahead and do it. Could we stop a step away?
His early career is super interesting. Obsessed with computers as well as chess as a boy, he started writing games. As a teenager he interned at a wild game startup (Bullfrog, run by the maverick Peter Molyneux). He was extremely successful, co-writing the awesome Theme Park games (I loved these) that featured very early AIs. The Bullfrog owner offered 17-year-old Demis £500k, a huge sum for a teenager. He wrote out the cheque and handed it to him to tempt him to stay and work on the next game, but he wanted to go to uni like his scientific heroes. He loved a movie telling the story of Watson & Crick at Cambridge, so that's where he went.
After excelling at uni he would do something strange for his cohort - start his own company, a game company. Here too he had early success, but then his vast ambition took him and new collaborator David Silver down the road to failure. A hugely ambitious new game would fail to materialise and the company would fold. It gave Demis his first tastes of real leadership, fundraising, and big business. All crucial later.
It was only now that he went to do his PhD. Afterwards, once again, he would turn down incredible money in the video game world as he planned to move on to his life's work, solving AI. The academic world did not seem the right fit for this; he knew he would need a lot of compute, a lot of funding, so...
DeepMind
After struggling to find like-minded collaborators during his PhD, Demis engineered some luck and introduced himself to kiwi researcher Shane Legg, a man deeper into the existing AI world. They clicked, and needed one more piece to found DeepMind - Mustafa Suleyman, now Microsoft AI CEO. He was a childhood friend of Demis's brother and has an incredible personal story: from further down the social ladder than Demis, he is presented as fearless, principled, and driven. He earned his way to Oxford only to drop out. I've read his book and never knew either this background or the manner of his DeepMind departure, both covered well here.
DeepMind was very much Demis's baby; the other cofounders did not have equal ownership shares. Demis was here in founder mode. Using his "Jedi mind tricks" and his amazing ability to storytell, he would convince and inspire researchers to join and to excel. His early experience working at Bullfrog and at his own game startup shines through - he gave people more freedom to tinker and play. Mallaby points out that there is a fine line between inspiration and control, dark and light.
Needing finance beyond his own limited funds and what was on offer in the UK, the boys headed to America.
"Demis is a true entrepreneur, he would do it for free. He will never quit. Big ambition." - Peter Thiel
Founders Fund would be the primary early backers, seeing the contrarian bet - a moonshot idea led by a brilliant founder. This was long before the AI we have today, before transformers and LLMs; it really was an expensive long shot at the time.
The pitch was a blend of neuroscience and computing, putting together neural nets and reinforcement learning, two previously polarised AI camps. DeepMind's roadmap was prescient. One aim was to ground intelligence in a world model, as without this a system could not be truly intelligent like humans. Later aspects of the neuroscience were dropped, and while world models are not yet in place, LLMs have taken off. Their projections are on track. The ~2030 AGI forecast now looks almost conservative, but it was radical 15 years ago. Shane Legg predicted it for 2028 even before DeepMind.
DeepMind was up and running, the best researchers were attracted and they worked on fundamental research and big flashy breakthroughs - often in games.
Move 37 & Protein Folding
Go is a deceptively complex board game that is hugely popular in the Far East. Unlike chess, computers were not able to defeat the best humans. Demis set his old friend and reinforcement learning specialist David Silver and team on the task. They made great progress and Demis set up a match against Lee Sedol, one of the highest-ranking players, in a televised and highly publicised event with a $1 million prize.
DeepMind's program AlphaGo would win the series 4-1. But it is move 37 of game 2 that is famous. The move was considered a mistake, something no good player would do. Only late in the game did its genius become apparent. It was creative, something outside of established play, a new way to play. 18 months later a new version, AlphaGo Zero, would beat this version 100-0. It was special as it used no human training data; it learned from playing itself and it blew past not only the best humans but the best AIs trained on human play.
High on the success, Demis pivoted many from the team onto a project he had been thinking over since his uni days, a real scientific problem: the structure of proteins.
The shape of the problem was good for AI - huge labelled datasets, clear input and output, and a blind-marked benchmark, CASP, that ran every few years. In 2018 they won the competition, a little better than anyone else. In 2020 they completely dominated, causing the surprised organisers to say protein folding for single-chain structures was essentially solved. A true breakthrough, this won Demis and project lead John Jumper the Nobel Prize in Chemistry in 2024. Demis, the son of a Greek-Cypriot father and Chinese-Singaporean mother, was knighted the same year, one of the UK's highest honours.
Google & Business
Much of the funding for this research was only possible because DeepMind was sold to Google in 2014. Outgrowing Founders Fund despite some late Elon Musk attempts, and having skilfully avoided Mark Zuckerberg, Demis decided to go with Google as he thought it was the best path to realising his overall goal. The cofounders adroitly negotiated key safety frameworks and maintained a lot of independence.
The early years brought the successes above, but tensions and reality would bite later on.
Google researchers would invent the transformer, the key technological step that powers LLMs. This is brilliantly explained by the author. They would not press home their lead here and missed the early revolution. There was more to LLMs than language; scaling them brought reasoning and understanding about the world. Demis is humble enough to know he missed this.
Google also had a version of ChatGPT a year before OpenAI, but buried it due to worries around hallucinations and the core search business. A massive miss given how things went, but it is understandable - I thought the same writing a blog post on how their search business was threatened by Perplexity and the best LLMs. Google is doing better than ever even with second-rate LLMs, so far.
Demis and Google were blindsided by Sam Altman and his rush to productise ChatGPT and his global publicity tour. Demis was furious at the starting gun of a race being fired; he felt that he too simply had to race, also publicly. This is the cold logic of technological determinism and capitalism. Google powered back in product space, by no means Demis's home field, but ever the competitor he was trying hard as part of Google. I feel you can still see this in his public comments today: he performs for the business, and is excited for the deeper science. Bard would launch already behind OpenAI. In some areas they would catch up, in narrow areas like video even surpass. But the vibes have never really shifted. Now, post Demis leaving as CEO, it is unclear if they will return to the AI frontier.
This relative success was despite Google having competing internal AI teams; Google Brain was run as a separate function outside of DeepMind for years, hampering overall progress. They would fail with Gaia, an attempt at world models, before language models powered ahead.
I was struck by the role of Sundar Pichai, Google CEO. Demis coveted independence and worked on a DeepMind spin-out, even convincing Larry & Sergey, but he was outmanoeuvred by the softly spoken and non-confrontational Sundar. Highlighting his vision with AI at the core of Google, Sundar would tease and delay and block spin-out ideas. Now we see a DeepMind VP (not CEO) reporting to Sundar, with Demis moved upstairs to Chair of DeepMind and Chief Scientist at Google. I think they wanted to keep him as long as possible and ideally never lose him, but they also need Gemini to succeed as a core part of the Google offer. Unlike, say, Waymo. Smart moves from Sundar; he's not CEO of Google for no reason.
Big Plans
There are lots of great philosophical quotes to go with those from the famed scientists and mathematicians. You will hear from the likes of Spinoza, Einstein, Kant, Gödel and Nagel, and read about their (and others') big ideas - brains in vats, simulation theory, the universe as God. Fun stuff. Taking a step back you can see Demis aiming to place himself in this tradition, the next step, and maybe one of the last we take as a species without machine intelligence.
It is in this domain he sees his future. He would like to solve intelligence, and from that disease, energy, physics, everything.
"Humanism, spiritualism, and science all go together... like Spinoza, Einstein, Da Vinci, it's art and science... it's fluid, everything is a river. Philosophy is a way of life. Self knowledge, knowledge, from sand and copper we get semiconductors... God's design... the flow is going towards finding out, understanding, I'm part of that and it's exhilarating" - Sir Demis
When asked how he spends his cash: not on fancy cars, houses or yachts, his only indulgences being some first edition books and Liverpool FC tickets (up the Reds!).
He wants to build a massive particle collider in space to help answer the big questions in physics ("what I cannot create, I do not understand" - Feynman). He wants to know if the universe is quantum or not, how we can describe the universe at the base level. He is a classical computer champion, positing that we don't need quantum explanations and that Turing machines can explain it all in 1s and 0s if powerful enough. He sees quantum mechanics as an inefficient way to run the universe and would like to cut out the weird stuff.
These ideas and goals are the great promise of artificial intelligence. But we will end on the threat.
"AGI is coming, we are not where we hoped we would be but it's coming" - David Silver
There are catastrophic and existential threats posed by AI. My p(doom) - the chance of things going this wrong in the next decade or so - has recently moved up to ~15%. I'm not alone in saying we might kill everyone, end civilisation, or lose human control over it very soon. I feel better knowing how seriously Demis takes the issue; I worry the conditions now in play raise the risk. I commend OpenAI's recently announced pause at the frontier to allow alignment work to catch up. We need a lot more work here at technical, business, and political levels to keep us on a happy path.
May 15, 2026
Demis is one of the most important people in AI. Regardless of how you feel about this technology, its impact on our world is unmistakable—whether good or bad. It is, and will only continue to be, a co-author of our future. Read this book to better understand why and how.
June 19, 2026
It leans toward worship and doesn't press the ethical contradictions. It raises hard questions, then leaves you empty-handed. Little here is new if you already follow the field, and it's still too long for the story it tells.
August 16, 2026
A wonderful, deeply thought-provoking work that is both the biography of one of the most important characters on AI, and a near-complete history of the first few years of the Singularity. Highly recommended.
July 12, 2026
The narrative is at its strongest when it outlines the interpersonal politics of the competing AI labs as opposed to its high-level technical explanations of the varying models. Although I was not rapt, I guess it’s enlightening to know that 98% of the people building AI think it’s capable of overtaking humans on all tasks and possibly even destroying humanity… but at this point, what the heck am I even supposed to do with that information!?
Anyways - this is more of a 3.25, not due to a lack of quality but simply because I often felt unmotivated to pick this book back up.
Anyways - this is more of a 3.25, not due to a lack of quality but simply because I often felt unmotivated to pick this book back up.
May 13, 2026
Hagiographic and structurally uninspiring. The way Mallaby lays out the chronology is frustrating, it mostly follows the timeline but then jumps around enough to make it confusing.
Large swaths just felt like Mallaby taking dictation from, usually, Hassabis or another source. I get that it’s a valuable resource that he had so much direct success but it seems like the methods outshined the ultimate artifact. I’m personally not a fan of Mallaby inserting his process so much talking about himself emailing and sitting down with his sources. At points it just devolves into Q&A. I thought I’d be getting more of a straight history with Hassabis as the main focus.
Main it was just the narrator of my audiobook but the end result for me was that I found it quite boring, when the subject should be riveting. Felt the same way about The Power Law, and leaves me wondering how these two books can be so much worse than More Money than God.
Large swaths just felt like Mallaby taking dictation from, usually, Hassabis or another source. I get that it’s a valuable resource that he had so much direct success but it seems like the methods outshined the ultimate artifact. I’m personally not a fan of Mallaby inserting his process so much talking about himself emailing and sitting down with his sources. At points it just devolves into Q&A. I thought I’d be getting more of a straight history with Hassabis as the main focus.
Main it was just the narrator of my audiobook but the end result for me was that I found it quite boring, when the subject should be riveting. Felt the same way about The Power Law, and leaves me wondering how these two books can be so much worse than More Money than God.
April 14, 2026
The Infinity Machine : Demis Hassabis, DeepMind, and the Quest for Superintelligence (2025) by Sebastian Mallaby is a biography of the remarkable Demis Hassabis and of the current race to build machine intelligence. Mallaby is a writer who has been a Washington Post columnist and who has written various books on Alan Greenspan and Hedge funds among other topics.
Hassabis was born in London to a Greek Cypriot father and a Singaporean mother. They had little money. Demis learnt chess at four and was a chess champion by six. At six he qualified for the British Under 14 Championship. By the age of 12 he had decided that becoming very good at chess wasn’t wise. He thought intelligence should be used for more. Hassabis used money won in chess to buy a ZX Spectrum. At 12 he bought an Amiga and started to learn to program. He started by trying to write chess programs. He soon wrote an Othello program that worked.
Hassabis did incredibly well at school and skipped grades and got admission to Cambridge early. They told him to wait and so he worked for the Bullfrog games company. There he coded and helped develop their hit game Theme Park. He then went to Cambridge and did extremely well there. After Cambridge he founded the company Elixir and sold that before doing a PhD in neuroscience at University College London.
After that he founded DeepMind with Shane Legg and Mustafa Suleyman. Deepmind became one of the world’s premier AI labs and was taken over by Google. DeepMind created AlphaGo that beat the world’s best Go player. They then went on to create AlphaFold that solved the problem of what protein came from what amino acid sequence. For this work Hassabis was awarded the Nobel Prize in Chemistry.
Following AlphaFold the next big breakthrough in Machine Learning was Transformers and the rise of the LLM. Here the book gets into how Google invented the Transformer but failed to exploit it. OpenAI, led eventually by Sam Altman took the lead and released ChatGPT. The book goes into detail about how DeepMind and Google reacted to this. There is little mention of Anthropic, one of the other major AI labs. The incredible leaps in ability of LLMs and how they forced Google to change their organisation and operation are interesting to read about. Hassabis and others were impressed by these models but somewhat skeptical of their practical applications. However this has changed as the models have improved.
The Infinity Machine is a well written, interesting read about the remarkable figure of Hassabis and the recent history of AI development. Hassabis comes across as he does everywhere as an incredibly driven, incredibly intelligent person who is also a decent human being. The book does veer a bit into hagiography, but it does appear to concur with almost everything written about Hassabis. This is in sharp contrast to writing about Sam Altman. For anyone interested in the personalities behind AI and its recent history the book is well worth a read.
Hassabis was born in London to a Greek Cypriot father and a Singaporean mother. They had little money. Demis learnt chess at four and was a chess champion by six. At six he qualified for the British Under 14 Championship. By the age of 12 he had decided that becoming very good at chess wasn’t wise. He thought intelligence should be used for more. Hassabis used money won in chess to buy a ZX Spectrum. At 12 he bought an Amiga and started to learn to program. He started by trying to write chess programs. He soon wrote an Othello program that worked.
Hassabis did incredibly well at school and skipped grades and got admission to Cambridge early. They told him to wait and so he worked for the Bullfrog games company. There he coded and helped develop their hit game Theme Park. He then went to Cambridge and did extremely well there. After Cambridge he founded the company Elixir and sold that before doing a PhD in neuroscience at University College London.
After that he founded DeepMind with Shane Legg and Mustafa Suleyman. Deepmind became one of the world’s premier AI labs and was taken over by Google. DeepMind created AlphaGo that beat the world’s best Go player. They then went on to create AlphaFold that solved the problem of what protein came from what amino acid sequence. For this work Hassabis was awarded the Nobel Prize in Chemistry.
Following AlphaFold the next big breakthrough in Machine Learning was Transformers and the rise of the LLM. Here the book gets into how Google invented the Transformer but failed to exploit it. OpenAI, led eventually by Sam Altman took the lead and released ChatGPT. The book goes into detail about how DeepMind and Google reacted to this. There is little mention of Anthropic, one of the other major AI labs. The incredible leaps in ability of LLMs and how they forced Google to change their organisation and operation are interesting to read about. Hassabis and others were impressed by these models but somewhat skeptical of their practical applications. However this has changed as the models have improved.
The Infinity Machine is a well written, interesting read about the remarkable figure of Hassabis and the recent history of AI development. Hassabis comes across as he does everywhere as an incredibly driven, incredibly intelligent person who is also a decent human being. The book does veer a bit into hagiography, but it does appear to concur with almost everything written about Hassabis. This is in sharp contrast to writing about Sam Altman. For anyone interested in the personalities behind AI and its recent history the book is well worth a read.
April 24, 2026
Excellent new perspective on the AI Arms Race
I have read multiple background books on the different companies involved in the race for AI dominance over the past year. Empire of AI does a good job showing the interworkings of OpenAI and Careless People gives you an inside look into Meta. I guess I could include Isaacson's biography of Elon Musk in the mix too, but it didn't take the xAI elements too into consideration in that book.
This book gives another valuable perspective in the AI arms race. The history of Hassabis and Deepmind is a critical piece of history in the development of AI. The stories related to his early life at Bullfrog Games, Elixir and many many Deepmind / Google projects such as Atari gaming, AlphaGo, AlphaFold, and the AI models up to current Gemini models are all highly interesting and insightful. In particular I enjoyed the book talking about the actual technical assertions and building blocks of the Deepmind research. It taught me quite a bit about why they pursued the directions they did while others such as OpenAI pursued others.
The two most revealing elements of the book are seeing the contrast of Deepmind with it's primary focus of scientific research in AI (researcher led) vs OpenAI's primary focus on shipping products (engineer led). This contrast would play out quite a bit in the history of the past decade and I am surprised I wasn't more aware of it. The second thing that is really interesting is Hassabis's deep interest in Reinforced Learning (RL) which was the primary basis for the AlphaGo triumphs. Deepmind had very little interest in LLMs until OpenAI thrashed them so thoroghly that they were forced to accept this other approach was a better way than they were purusing. Interestingly, later versions of GPT actually started to explore back in the realm of RL, and Hassabis still believes that long term the research will come back around to RL as opposed to LLMs. We will see.
If you enjoy biographies or want to have better context on the current AI debates or progress, this is a great book to pick up.
I have read multiple background books on the different companies involved in the race for AI dominance over the past year. Empire of AI does a good job showing the interworkings of OpenAI and Careless People gives you an inside look into Meta. I guess I could include Isaacson's biography of Elon Musk in the mix too, but it didn't take the xAI elements too into consideration in that book.
This book gives another valuable perspective in the AI arms race. The history of Hassabis and Deepmind is a critical piece of history in the development of AI. The stories related to his early life at Bullfrog Games, Elixir and many many Deepmind / Google projects such as Atari gaming, AlphaGo, AlphaFold, and the AI models up to current Gemini models are all highly interesting and insightful. In particular I enjoyed the book talking about the actual technical assertions and building blocks of the Deepmind research. It taught me quite a bit about why they pursued the directions they did while others such as OpenAI pursued others.
The two most revealing elements of the book are seeing the contrast of Deepmind with it's primary focus of scientific research in AI (researcher led) vs OpenAI's primary focus on shipping products (engineer led). This contrast would play out quite a bit in the history of the past decade and I am surprised I wasn't more aware of it. The second thing that is really interesting is Hassabis's deep interest in Reinforced Learning (RL) which was the primary basis for the AlphaGo triumphs. Deepmind had very little interest in LLMs until OpenAI thrashed them so thoroghly that they were forced to accept this other approach was a better way than they were purusing. Interestingly, later versions of GPT actually started to explore back in the realm of RL, and Hassabis still believes that long term the research will come back around to RL as opposed to LLMs. We will see.
If you enjoy biographies or want to have better context on the current AI debates or progress, this is a great book to pick up.
April 29, 2026
"Intelligence is fundamental. It is the root of all else. It is the mechanism through which humans perceive reality."
Fascinating to hear the backstory behind the current AI wars unfolding right in front of us. Quite interesting that Mallaby was able to write and publish this so quickly, while we are still in the thick of frontier lab warfare and a clear winner has yet to emerge.
Docking a star just because unfortunately, the biography seemed a bit too biased towards Hassabis (as one might expect though). I do think it's more valuable to read biographies like those by Walter Isaacson—where the author is able to pry deeply into an individual's life and interview people involved from both the good and bad sides, painting a portrait of them as a whole person rather than a pure hero.
We might need a Part 2 soon, once Superintelligence enters the picture...
-------------------------------------------
"The mind interprets the world."
"What I cannot build, I cannot understand."
"The most successful founders do not create companies. They are on a mission to create something closer to a religion."
"People make a mistake in thinking of themselves as small. They don't think of how they personally affect history."
"Technology happens because it's possible."
Fascinating to hear the backstory behind the current AI wars unfolding right in front of us. Quite interesting that Mallaby was able to write and publish this so quickly, while we are still in the thick of frontier lab warfare and a clear winner has yet to emerge.
Docking a star just because unfortunately, the biography seemed a bit too biased towards Hassabis (as one might expect though). I do think it's more valuable to read biographies like those by Walter Isaacson—where the author is able to pry deeply into an individual's life and interview people involved from both the good and bad sides, painting a portrait of them as a whole person rather than a pure hero.
We might need a Part 2 soon, once Superintelligence enters the picture...
-------------------------------------------
"The mind interprets the world."
"What I cannot build, I cannot understand."
"The most successful founders do not create companies. They are on a mission to create something closer to a religion."
"People make a mistake in thinking of themselves as small. They don't think of how they personally affect history."
"Technology happens because it's possible."
August 30, 2026
Despite finding Mallaby frustrating at many points throughout the book, this is a truly excellent overview of the history of the last decade. I appreciated his skepticism towards Altman and pragmatism in outlining the contingencies that led to our current moment.
Mallaby is well-informed and an excellent communicator of complex information. This period’s ebbs and flows of interpersonal dynamics will be some of the most consequential in the history of technology. Centering the people at the heart of the research is the right move and he does it well.
Perhaps inadvertently, Hassabis appears an almost tragic Shakespearean figure throughout — despite achieving his dreams (Nobel prize, transforming a field), he continues to lose the race because of the very qualities that make him most suited to win.
Hassabis, like all of us, sees the world through his own lens. His strengths in game-playing initially made DeepMind. But the world is not a game and we will need further conceptual innovations in the next era of learning. Hassabis has won many of his games, dominating in the rules-based arenas, but the titans of Wall Street and the Valley do not play by rules and are looking likely to win the war of power.
Mallaby is well-informed and an excellent communicator of complex information. This period’s ebbs and flows of interpersonal dynamics will be some of the most consequential in the history of technology. Centering the people at the heart of the research is the right move and he does it well.
Perhaps inadvertently, Hassabis appears an almost tragic Shakespearean figure throughout — despite achieving his dreams (Nobel prize, transforming a field), he continues to lose the race because of the very qualities that make him most suited to win.
Hassabis, like all of us, sees the world through his own lens. His strengths in game-playing initially made DeepMind. But the world is not a game and we will need further conceptual innovations in the next era of learning. Hassabis has won many of his games, dominating in the rules-based arenas, but the titans of Wall Street and the Valley do not play by rules and are looking likely to win the war of power.
August 29, 2026
Fantastic book in my opinion. We focus so much on the impact of AI but in the negative eye but there is a whole other side to it. I mean we solved protein folding problem!! It’s huge, it changed science. Some people would spend their whole PhD on figuring out a structure of one protein. Understanding the structure of protein we can actually develop better medication and treatment for various diseases. Applying AI in science and medicine, it can actually be beneficial IF we have correct safety measures in place, which is what I enjoyed about this book. It highlights that we moved too fast due to competition, we need to pause and ensure safety drakes are in place but too many eager competitors to be the “best”.
July 10, 2026
I'm not in an industry that's adopting AI quickly. There are benefits to that, for sure. I want to be aware of general trends happening and understand them better. My husband is directly in AI in his work.
This book provided the story of AI development through the anchor of Hassabis. I found it very detailed and interesting. Mallaby paints the timeline well.
I recommend this book if you want to know how we got to where we are with AI.
This book provided the story of AI development through the anchor of Hassabis. I found it very detailed and interesting. Mallaby paints the timeline well.
I recommend this book if you want to know how we got to where we are with AI.
August 16, 2026
Very much enjoyed this book
1. Hassabis' life story is just super fascinating. I find his motivation very noble and his drive inspiring
2. AlphaGo, AlphaZero, AlphaFold...can't wait to see what this genius will do next and how RL will pave the way for AI to solve problems formerly unsolvable
3. This book is peppered with interesting anecdotes which made it quite enjoyable (the author's distrust of Sam Altman came out loud and clear)
1. Hassabis' life story is just super fascinating. I find his motivation very noble and his drive inspiring
2. AlphaGo, AlphaZero, AlphaFold...can't wait to see what this genius will do next and how RL will pave the way for AI to solve problems formerly unsolvable
3. This book is peppered with interesting anecdotes which made it quite enjoyable (the author's distrust of Sam Altman came out loud and clear)
Displaying 1 - 30 of 338 reviews