What do you think?
Rate this book
Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI
An Instant New York Times Bestseller
“A bestselling page-turner that has made waves not just in Silicon Valley but around the world . . . With Empire of AI, Hao is fundamentally shaping many people’s perceptions and understanding of the company at the center of the AI revolution.” —TIME Magazine, “TIME100 AI 2025”
“Excellent and deeply reported.” —Tim Wu, The New York Times
“Startling and intensely researched . . . an essential account of how OpenAI and ChatGPT came to be and the catastrophic places they will likely take us.” —Vulture
“Hao’s reporting inside OpenAI is exceptional, and she’s persuasive in her argument that the public should focus less on A.I.’s putative ‘sentience’ and more on its implications for labor and the environment.” —Benjamin Wallace-Wells, New Yorker
From a brilliant longtime AI insider with intimate access to the world of Sam Altman's OpenAI from the beginning, an eye-opening account of arguably the most fateful tech arms race in history, reshaping the planet in real time, from the cockpit of the company that is driving the frenzy
When AI expert and investigative journalist Karen Hao first began covering OpenAI in 2019, she thought they were the good guys. Founded as a nonprofit with safety enshrined as its core mission, the organization was meant, its leader Sam Altman told us, to act as a check against more purely mercantile, and potentially dangerous, forces. What could go wrong?
Over time, Hao began to wrestle ever more deeply with that question. Increasingly, she realized that the core truth of this massively disruptive sector is that its vision of success requires an almost unprecedented amount of the “compute” power of high-end chips and the processing capacity to create massive large language models, the sheer volume of data that needs to be amassed at scale, the humans “cleaning up” that data for sweatshop wages throughout the Global South, and a truly alarming spike in the usage of energy and water underlying it all. The truth is that we have entered a new and ominous age of only a small handful of globally scaled companies can even enter the field of play. At the head of the pack with its ChatGPT breakthrough, how would OpenAI resist such temptations?
Spoiler it didn’t. Armed with Microsoft’s billions, OpenAI is setting a breakneck pace, chased by a small group of the most valuable companies in human history—toward what end, not even they can define. All this time, Hao has maintained her deep sourcing within the company and the industry, and so she was in intimate contact with the story that shocked the entire tech industry—Altman’s sudden firing and triumphant return. The behind-the-scenes story of what happened, told here in full for the first time, is revelatory of who the people controlling this technology really are. But this isn’t just the story of a single company, however fascinating it is. The g forces pressing down on the people of OpenAI are deforming the judgment of everyone else too—as such forces do. Naked power finds the ideology to cloak itself; no one thinks they’re the bad guy. But in the meantime, as Hao shows through intrepid reporting on the ground around the world, the enormous wheels of extraction grind on. By drawing on the viewpoints of Silicon Valley engineers, Kenyan data laborers, and Chilean water activists, Hao presents the fullest picture of AI and its impact we’ve seen to date, alongside a trenchant analysis of where things are headed.
“A bestselling page-turner that has made waves not just in Silicon Valley but around the world . . . With Empire of AI, Hao is fundamentally shaping many people’s perceptions and understanding of the company at the center of the AI revolution.” —TIME Magazine, “TIME100 AI 2025”
“Excellent and deeply reported.” —Tim Wu, The New York Times
“Startling and intensely researched . . . an essential account of how OpenAI and ChatGPT came to be and the catastrophic places they will likely take us.” —Vulture
“Hao’s reporting inside OpenAI is exceptional, and she’s persuasive in her argument that the public should focus less on A.I.’s putative ‘sentience’ and more on its implications for labor and the environment.” —Benjamin Wallace-Wells, New Yorker
From a brilliant longtime AI insider with intimate access to the world of Sam Altman's OpenAI from the beginning, an eye-opening account of arguably the most fateful tech arms race in history, reshaping the planet in real time, from the cockpit of the company that is driving the frenzy
When AI expert and investigative journalist Karen Hao first began covering OpenAI in 2019, she thought they were the good guys. Founded as a nonprofit with safety enshrined as its core mission, the organization was meant, its leader Sam Altman told us, to act as a check against more purely mercantile, and potentially dangerous, forces. What could go wrong?
Over time, Hao began to wrestle ever more deeply with that question. Increasingly, she realized that the core truth of this massively disruptive sector is that its vision of success requires an almost unprecedented amount of the “compute” power of high-end chips and the processing capacity to create massive large language models, the sheer volume of data that needs to be amassed at scale, the humans “cleaning up” that data for sweatshop wages throughout the Global South, and a truly alarming spike in the usage of energy and water underlying it all. The truth is that we have entered a new and ominous age of only a small handful of globally scaled companies can even enter the field of play. At the head of the pack with its ChatGPT breakthrough, how would OpenAI resist such temptations?
Spoiler it didn’t. Armed with Microsoft’s billions, OpenAI is setting a breakneck pace, chased by a small group of the most valuable companies in human history—toward what end, not even they can define. All this time, Hao has maintained her deep sourcing within the company and the industry, and so she was in intimate contact with the story that shocked the entire tech industry—Altman’s sudden firing and triumphant return. The behind-the-scenes story of what happened, told here in full for the first time, is revelatory of who the people controlling this technology really are. But this isn’t just the story of a single company, however fascinating it is. The g forces pressing down on the people of OpenAI are deforming the judgment of everyone else too—as such forces do. Naked power finds the ideology to cloak itself; no one thinks they’re the bad guy. But in the meantime, as Hao shows through intrepid reporting on the ground around the world, the enormous wheels of extraction grind on. By drawing on the viewpoints of Silicon Valley engineers, Kenyan data laborers, and Chilean water activists, Hao presents the fullest picture of AI and its impact we’ve seen to date, alongside a trenchant analysis of where things are headed.
496 pages, Kindle Edition
First published May 20, 2025
3475 people are currently reading
23499 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 893 reviews
July 4, 2025
I appreciated this book’s spirit but it was so, so long. I think it’s important that journalists like Karen Hao expose the downsides of AI (e.g., people in the Global South having to do degrading work for miniscule wages, the destructive environmental impacts). At the same time, I found this particular book a bit too unfocused and long. There was a lot of information about different “players” in the AI space but I think a stronger and more concise thesis would have helped Empire of AI feel like a more satisfying read.
November 18, 2025
tl;dr -Empire of AI is not the definitive chronicle of the genAI revolution or the sama drama at OpenAI. It is, instead, a cautionary example illustrating how ideological zeal ruins what could have otherwise been a significant journalistic venture.
Karen Hao’s Empire of AI aspires to be an expansive exposé of one of the most influential and enigmatic companies in the tech world. With its dramatic subtitle, the book promises a deep dive into the dreams and nightmares wrought by OpenAI under the leadership of Sam Altman. Yet, while Hao succeeds in marshaling an impressive quantity of interviews (300+ from 260 individuals) and experience (~7 years of reporting on the AI sector), the book ultimately suffers from a tendentious slant, an unwarranted contempt for the technology itself, namely the fervor for AGI, and a frustrating superficiality in its treatment of the internal dynamics and culture of OpenAI (One of the few reasons anyone would pick this book up would be to learn about why sama was briefly fired and re-hired as CEO).
The book's principal flaw lies in Hao’s persistent framing of OpenAI as an almost unambiguously self-serving, exploitative, and reckless empire masquerading as a public-spirited research lab. While directing skepticism at tech goliaths is reasonable, Hao’s argument relies on insinuation, which can often border on a sardonic or reproving tone, rather than rigorous and honest analysis. Whether a reader is persuaded by Hao will depend wholly on whether they share Hao's professed worldview (a popular understand of academic post-colonialism) rather than the facts presented. In Hao's view, the three great sins of the AI industry are believing in scaling laws (a belief that justifies the high capital and energy costs associated with training LLMs), scraping copyrighted training data from the internet without paying royalties (to what extent this actually occurred is not well-researched in this book either; the evidence is anecdotal) and relying on cheap labor from the Third World to assist with data annotation and the fine-tuning of their models, a process referred to as reinforcement learning from human feedback (RLHF).
Hao's hostility to AI as a technology appears largely motivated by the idea that AI tech is the frontier of 21st century colonialism, where Western powers exploit and oppress the Global South to enrich themselves. Whether this idea makes any sense at all is never examined. Instead, it is simply something Hao assumes to be the case, and she doesn't let pesky facts get in the way of highlighting the supposed evils of our current tech companies in our current political economy. And she expects us to be enraged by the responses of OpenAI execs when they're confronted by her about her concerns about the ethical, environmental, or economic costs of developing AGI. She makes so much of the apparent paradox behind the impassioned pursuit of a technology that supposedly presents an existential risk to humanity. But is it really surprising that autists reared on science fiction believe AI has millenarian stakes? To anyone with a broader perspective on the history of technological innovation, the stakes of AI, while still very important, appear much more pedestrian.
This book also promises to provide some devastating insight into Sam Altman, but there is just no there there. Spending so much time on Ann Altman, for instance, was gauche. Hao's depiction of Sam Altman vacillates between that of a scheming Machiavellian and a hapless idealist overwhelmed by the demands of his many ventures and immense responsibility. This is at once a contradictory and thin portrait of Altman. The same takeaways wouldn't be that hard to gather by simply absorbing some of the ambient conversation about OpenAI on the internet (e.g. any AI-related podcast) and reading sama's tweets. The big failure here is that there are really no actual new insights into Altman's brief removal from his CEO position in November 2023, which still appears to be the result of incompetent bumbling from AI doomers, specifically Ilya Sutskever.
Hao's treatment of AI technology itself is disappointingly shallow. Despite covering one of the most technically complex and fast-evolving domains in contemporary science, Empire of AI rarely rises above the most surface-level exposition. Apart from some absolute basics about neural networks, there is scant effort to explain how transformer architectures work, what distinguishes frontier models from earlier iterations, or why scaling laws matter. Additionally, there is no real survey of the history of advancement in AI (Hinton, LeCunn, Minsky, McCarthy, Newell, etc - some get passing mentions at most). In fact, Hao dismisses much of the sophistication and complexity of the technology by saying once the pieces and parts of it are understood that it all seems quaint (fluency illusion). This leaves the reader without a coherent grasp of what exactly OpenAI has achieved or how it differs from competitors like DeepMind, Anthropic (she does describe why Anthropic exists!), Meta, or others. In place of analytical rigor, Hao often defaults to anecdote or second-hand metaphor, which may entertain but does little to enlighten.
Perhaps the most unforgivable aspect of the book is its failure to offer a substantive reconstruction of OpenAI’s internal debates about foundation genAI and business model questions. Hao just wants to tar OpenAI figures as weirdo techno-utopians or money-grubbing tech bros, whichever is most convenient for whatever point she wants to make at that moment. While the reporting frequently alludes to power struggles, ideological rifts, and a revolving door of talent, these conflicts are rendered in vague terms. Many of the figures are thinly sketched apart from, perhaps, Sutskever and Brockman. Hao doesn't give the reader a clear sense of the intellectual stakes in these disputes, nor does she convincingly differentiate between personality clashes and genuine philosophical disagreements. The result is a fatally slanted narrative of compiled grievances rather than a deep, thorough investigation of a central player in genAI.
Empire of AI is both overlong and flawed as a journalistic venture, while also tossing aside any pretense of journalistic objectivity in order to insert a bizarre polemical framing. Any reader looking for a nuanced, rigorous account of the latest in the AI sector or a history of OpenAI as a non-profit entity/company or its leader will come away disappointed. For those already inclined to view Silicon Valley with suspicion, Hao offers confirmation and comfort is a juvenile form of righteous indignation and smug superiority.
Karen Hao’s Empire of AI aspires to be an expansive exposé of one of the most influential and enigmatic companies in the tech world. With its dramatic subtitle, the book promises a deep dive into the dreams and nightmares wrought by OpenAI under the leadership of Sam Altman. Yet, while Hao succeeds in marshaling an impressive quantity of interviews (300+ from 260 individuals) and experience (~7 years of reporting on the AI sector), the book ultimately suffers from a tendentious slant, an unwarranted contempt for the technology itself, namely the fervor for AGI, and a frustrating superficiality in its treatment of the internal dynamics and culture of OpenAI (One of the few reasons anyone would pick this book up would be to learn about why sama was briefly fired and re-hired as CEO).
The book's principal flaw lies in Hao’s persistent framing of OpenAI as an almost unambiguously self-serving, exploitative, and reckless empire masquerading as a public-spirited research lab. While directing skepticism at tech goliaths is reasonable, Hao’s argument relies on insinuation, which can often border on a sardonic or reproving tone, rather than rigorous and honest analysis. Whether a reader is persuaded by Hao will depend wholly on whether they share Hao's professed worldview (a popular understand of academic post-colonialism) rather than the facts presented. In Hao's view, the three great sins of the AI industry are believing in scaling laws (a belief that justifies the high capital and energy costs associated with training LLMs), scraping copyrighted training data from the internet without paying royalties (to what extent this actually occurred is not well-researched in this book either; the evidence is anecdotal) and relying on cheap labor from the Third World to assist with data annotation and the fine-tuning of their models, a process referred to as reinforcement learning from human feedback (RLHF).
Hao's hostility to AI as a technology appears largely motivated by the idea that AI tech is the frontier of 21st century colonialism, where Western powers exploit and oppress the Global South to enrich themselves. Whether this idea makes any sense at all is never examined. Instead, it is simply something Hao assumes to be the case, and she doesn't let pesky facts get in the way of highlighting the supposed evils of our current tech companies in our current political economy. And she expects us to be enraged by the responses of OpenAI execs when they're confronted by her about her concerns about the ethical, environmental, or economic costs of developing AGI. She makes so much of the apparent paradox behind the impassioned pursuit of a technology that supposedly presents an existential risk to humanity. But is it really surprising that autists reared on science fiction believe AI has millenarian stakes? To anyone with a broader perspective on the history of technological innovation, the stakes of AI, while still very important, appear much more pedestrian.
This book also promises to provide some devastating insight into Sam Altman, but there is just no there there. Spending so much time on Ann Altman, for instance, was gauche. Hao's depiction of Sam Altman vacillates between that of a scheming Machiavellian and a hapless idealist overwhelmed by the demands of his many ventures and immense responsibility. This is at once a contradictory and thin portrait of Altman. The same takeaways wouldn't be that hard to gather by simply absorbing some of the ambient conversation about OpenAI on the internet (e.g. any AI-related podcast) and reading sama's tweets. The big failure here is that there are really no actual new insights into Altman's brief removal from his CEO position in November 2023, which still appears to be the result of incompetent bumbling from AI doomers, specifically Ilya Sutskever.
Hao's treatment of AI technology itself is disappointingly shallow. Despite covering one of the most technically complex and fast-evolving domains in contemporary science, Empire of AI rarely rises above the most surface-level exposition. Apart from some absolute basics about neural networks, there is scant effort to explain how transformer architectures work, what distinguishes frontier models from earlier iterations, or why scaling laws matter. Additionally, there is no real survey of the history of advancement in AI (Hinton, LeCunn, Minsky, McCarthy, Newell, etc - some get passing mentions at most). In fact, Hao dismisses much of the sophistication and complexity of the technology by saying once the pieces and parts of it are understood that it all seems quaint (fluency illusion). This leaves the reader without a coherent grasp of what exactly OpenAI has achieved or how it differs from competitors like DeepMind, Anthropic (she does describe why Anthropic exists!), Meta, or others. In place of analytical rigor, Hao often defaults to anecdote or second-hand metaphor, which may entertain but does little to enlighten.
Perhaps the most unforgivable aspect of the book is its failure to offer a substantive reconstruction of OpenAI’s internal debates about foundation genAI and business model questions. Hao just wants to tar OpenAI figures as weirdo techno-utopians or money-grubbing tech bros, whichever is most convenient for whatever point she wants to make at that moment. While the reporting frequently alludes to power struggles, ideological rifts, and a revolving door of talent, these conflicts are rendered in vague terms. Many of the figures are thinly sketched apart from, perhaps, Sutskever and Brockman. Hao doesn't give the reader a clear sense of the intellectual stakes in these disputes, nor does she convincingly differentiate between personality clashes and genuine philosophical disagreements. The result is a fatally slanted narrative of compiled grievances rather than a deep, thorough investigation of a central player in genAI.
Empire of AI is both overlong and flawed as a journalistic venture, while also tossing aside any pretense of journalistic objectivity in order to insert a bizarre polemical framing. Any reader looking for a nuanced, rigorous account of the latest in the AI sector or a history of OpenAI as a non-profit entity/company or its leader will come away disappointed. For those already inclined to view Silicon Valley with suspicion, Hao offers confirmation and comfort is a juvenile form of righteous indignation and smug superiority.
May 25, 2025
Finally, a credible and impartial author debunks the Open/AI frenzy. Karen Hao avoids the tired "hype vs. revolution" framing that dominates most books on the subject. She is neither a cheerleader nor an opponent but instead offers a genuinely thorough investigation into the company’s merits, the technology, and the broader societal implications.
The biggest takeaway for me? The buzz around AI is real—corporate interest alone confirms its impending ubiquity. Yet it’s far from revolutionary. Rather than disrupting the status quo, this technology seeks to reinforce it, even tightening its grip on society.
If, like me, you hoped AI might level the playing field—reducing inequality, promoting diversity, and decentralizing power from a privileged few—think again. The same elites control this technology, and they are using it to consolidate their dominance, not just for today but for generations to come.
The biggest takeaway for me? The buzz around AI is real—corporate interest alone confirms its impending ubiquity. Yet it’s far from revolutionary. Rather than disrupting the status quo, this technology seeks to reinforce it, even tightening its grip on society.
If, like me, you hoped AI might level the playing field—reducing inequality, promoting diversity, and decentralizing power from a privileged few—think again. The same elites control this technology, and they are using it to consolidate their dominance, not just for today but for generations to come.
August 14, 2025
Well, this was enraging and illuminating and rather depressing and then enraging again.
Weirdly, even if I am a pretty well-known 'AI' (more like LLM, amirite?) hater, I actually did know quite a lot of the info presented here, on many of the directions it takes - I like to be aware of my enemy (haha). The ego angle, the eco angle, the slop shat from ingesting actual art angle, the 'this has been killing my field of work for years' and many others' angle. I didn't know that much about the labor angle and I somehow didn't have an idea that all of this data was labeled by poor people paid very little, gig economy style. Of course, I am not surprised, but I am enraged and appalled. I really loved the focus Hao had on the regular life stories of some of these people from Kenya and Venezuela. They absolutely deserve to be heard somewhere.
I am fascinated by the centrist 1-star reviews for this book. This is not the time for 'covering both sides equally'. I feel like that's partly why we got here. This is definitely the time for people with ethical clarity to tell us what we are risking, before it's too late. No lackadaisical bullshit. No neutrality. And Empire of AI is freaking that and I love it for it. And yet, and yet, I feel like Karen Hao was pretty balanced in this massively meticulously researched book and she helped soften a little bit my hard stance on 'AI'. She says quite a bunch of times that she is not against AI models, she is actually against the scale of these projects, the greedy resource extraction, the remorseless exploitation of labor (both of these in already poor environments), the lack of transparency on data sets, infringement of data privacy and the concentration of power in the hands of very few, ego-driven individuals. Same here.
Like, in a world with UBI (something Sam Altman said in the past he wants, but ofc hasn't done much in that direction), smaller scale models that are sustainable, and where people are actually paid well + benefits for their labor labeling data, I wouldn't mind 'AI' at all. It's this world and this system that's the problem.
There's an example in the epilogue (spoiler, I guess?) of a small model in New Zealand built by Maori people to save / preserve / teach their language to future generations, and it's been trained by people with their consent and I could not be against that. That sounds totally lovely!
I think anyone who has ever used ChatGPT or any other form of 'AI' (and even people who haven't) has to read this book and live in the uncomfortable space of knowing what it is built on, who it is built on. I don't think anyone who benefits from it should get to look away.
And I stand in solidarity with Indigenous people from Chile who have fought data centers, with Kenyans and Venezuelans used for cheap labor with data labeling and then discarded (I, too, know what it's like to build so much of your schedule around a queue of tasks that may or may not come, but I would be enraged at their side either way), with Sam Altman's disabled (hEDS) sister, discarded by her family and everyone who listened to their conscience.
Imho, large-scale AI models are an ethical nightmare from multiple perspectives and they mostly offer convenience to well-off people.
Weirdly, even if I am a pretty well-known 'AI' (more like LLM, amirite?) hater, I actually did know quite a lot of the info presented here, on many of the directions it takes - I like to be aware of my enemy (haha). The ego angle, the eco angle, the slop shat from ingesting actual art angle, the 'this has been killing my field of work for years' and many others' angle. I didn't know that much about the labor angle and I somehow didn't have an idea that all of this data was labeled by poor people paid very little, gig economy style. Of course, I am not surprised, but I am enraged and appalled. I really loved the focus Hao had on the regular life stories of some of these people from Kenya and Venezuela. They absolutely deserve to be heard somewhere.
I am fascinated by the centrist 1-star reviews for this book. This is not the time for 'covering both sides equally'. I feel like that's partly why we got here. This is definitely the time for people with ethical clarity to tell us what we are risking, before it's too late. No lackadaisical bullshit. No neutrality. And Empire of AI is freaking that and I love it for it. And yet, and yet, I feel like Karen Hao was pretty balanced in this massively meticulously researched book and she helped soften a little bit my hard stance on 'AI'. She says quite a bunch of times that she is not against AI models, she is actually against the scale of these projects, the greedy resource extraction, the remorseless exploitation of labor (both of these in already poor environments), the lack of transparency on data sets, infringement of data privacy and the concentration of power in the hands of very few, ego-driven individuals. Same here.
Like, in a world with UBI (something Sam Altman said in the past he wants, but ofc hasn't done much in that direction), smaller scale models that are sustainable, and where people are actually paid well + benefits for their labor labeling data, I wouldn't mind 'AI' at all. It's this world and this system that's the problem.
There's an example in the epilogue (spoiler, I guess?) of a small model in New Zealand built by Maori people to save / preserve / teach their language to future generations, and it's been trained by people with their consent and I could not be against that. That sounds totally lovely!
I think anyone who has ever used ChatGPT or any other form of 'AI' (and even people who haven't) has to read this book and live in the uncomfortable space of knowing what it is built on, who it is built on. I don't think anyone who benefits from it should get to look away.
And I stand in solidarity with Indigenous people from Chile who have fought data centers, with Kenyans and Venezuelans used for cheap labor with data labeling and then discarded (I, too, know what it's like to build so much of your schedule around a queue of tasks that may or may not come, but I would be enraged at their side either way), with Sam Altman's disabled (hEDS) sister, discarded by her family and everyone who listened to their conscience.
Imho, large-scale AI models are an ethical nightmare from multiple perspectives and they mostly offer convenience to well-off people.
dnf
May 27, 2025Too heavily biased to get through, even though I agree 100% with the author. I just like to be able to draw my own conclusions with the in-depth reporting I read. Isn’t that the whole point of journalism? Sigh.
July 3, 2025
In 2019, Karen Hao, a tech journalist, was given an exclusive interview at OpenAI, a little known AI company at the time. Since then she has been following OpenAI and generative AI development. Empire of AI is an investigation on OpenAI, AI industry, and AI “supply chain”. The book is loosely organized chronically. It covers the development of OpenAI since its inception. You can find plenty of dramas, including the “divorce” (several founding members broke off from OpenAI and created a new company Anthropic) and the boardroom drama in November 2023 that briefly ousted Sam Altman. You can get the behind the scenes of LLM and ChatGPT.
The book also includes several chapters about other AI companies and aspects of the AI industry. One chapter is dedicated to Google ousted AI scientist Timnit Gebru, another chapter documents the exploitive operations of data annotation and content monitoring by AI companies in the poorer countries, and a chapter discusses the environmental impact of mega data centers especially in Latin American.
What strikes me most is that OpenAI is yet another Silicon Valley company that proves money and power beget each other, and power corrupts. Sam Altman and Elon Musk co-founded OpenAI to counteract Google. They, including chief AI scientist Ilya Sutskever, truly believe that AGI (Artificial General Intelligence, think Matrix or The Terminator) is on the horizon. They say they can not let an evil company like Google control a technology powerful enough to destroy humanity, therefore they must create AGI first, all the while without answering the question of what makes them superior to others in controlling it. It’s ironic that they talk about the future risk of AI, yet ignore the harm their company and AI industry have already inflicted on others. I personally do not believe AGI is anywhere near us. No matter how much data LLM is fed upon, it will not become self-aware. However, AI is a powerful tool and when used by malicious actors, it will cause unthinkable damage.
My second thought is that Sam Altman is as manipulative from inside as he is charming from outside.
There is some back and forth, which can be repetitive, but overall the book is an excellent, in-depth report and sharp criticism of OpenAI, Sam Altman and the AI industry in general.
The book also includes several chapters about other AI companies and aspects of the AI industry. One chapter is dedicated to Google ousted AI scientist Timnit Gebru, another chapter documents the exploitive operations of data annotation and content monitoring by AI companies in the poorer countries, and a chapter discusses the environmental impact of mega data centers especially in Latin American.
What strikes me most is that OpenAI is yet another Silicon Valley company that proves money and power beget each other, and power corrupts. Sam Altman and Elon Musk co-founded OpenAI to counteract Google. They, including chief AI scientist Ilya Sutskever, truly believe that AGI (Artificial General Intelligence, think Matrix or The Terminator) is on the horizon. They say they can not let an evil company like Google control a technology powerful enough to destroy humanity, therefore they must create AGI first, all the while without answering the question of what makes them superior to others in controlling it. It’s ironic that they talk about the future risk of AI, yet ignore the harm their company and AI industry have already inflicted on others. I personally do not believe AGI is anywhere near us. No matter how much data LLM is fed upon, it will not become self-aware. However, AI is a powerful tool and when used by malicious actors, it will cause unthinkable damage.
My second thought is that Sam Altman is as manipulative from inside as he is charming from outside.
There is some back and forth, which can be repetitive, but overall the book is an excellent, in-depth report and sharp criticism of OpenAI, Sam Altman and the AI industry in general.
July 13, 2025
Karen Hao’s journalistic investigation of OpenAI (CHAT GPT) under the leadership of Sam Altman and Ilya Sutskever and other a few other notables including Elon Musk.
Hao reveals how OpenAI’s initially laudable mission to build open-source, safe artificial general intelligence (AGI) morphed into a powerful, secretive, and profit-driven Microsoft-backed for-profit enterprise with tech-arms race dynamics fueling the reckless sprint towards artificial super intelligence (ASI) that we currently find ourselves in.
I’m writing this just a few weeks after the release of AI 2027 (read it if you haven’t yet) and GROCK-4 Heavy (watch the press releases and fanboy hype cycle if you haven’t yet).
Rather than focusing exclusively on the debate between AI DOOMERS (name for people who are calling for safety research, alignment, and who are concerned that ASI represents an existential threat to humanity, including Ilya Sutskever and Elon Musk) and AI ZOOMERS (name for people who are not as concerned with AI safety, and who are enthusiastic to accelerate the development of AGI/ASI, in part out of the belief that advanced AI will solve existential threats to humanity, e.g. cancer, clean energy, and global warming, etc., including people like Sam Altman and).
Hao documents OpenAI’s evolution from an idealistic nonprofit to a Microsoft-backed powerhouse. Internal drama, including Altman’s 2023 ousting and rapid return. And most importantly, the global consequences, such as the exploitation of low-wage labor in Kenya and environmental costs of massive AI infrastructure (hence the title).
At the bottom, Hao’s central critique is that the AI boom mirrors colonial empires in their extraction of data and natural resources (e.g. energy and water at extremely alarming rates), and exploitation of vulnerable, impoverished labor (e.g. Latin American, East Asian, and African educated/poor people) to do the loathsome and often psychologically damaging work of conditioning AI data, frequently for less than a dollar a day, while other AI workers, executives, and investors make literally billions. And (as an added bonus) while conferring immense political power to an elite super rich, super small cadre of AI tech elites.
Hao argues that (in addition to the dizzyingly complex AI ethical problems) the AI empire could undermine global equity, labor rights, and even democratic governance. Particularly if it continues BAU without real democratic oversight.
Great book.
5/5⭐️
Definitely read AI 2027 by Daniel Kokotejio and the AI Futures project too.
Hao reveals how OpenAI’s initially laudable mission to build open-source, safe artificial general intelligence (AGI) morphed into a powerful, secretive, and profit-driven Microsoft-backed for-profit enterprise with tech-arms race dynamics fueling the reckless sprint towards artificial super intelligence (ASI) that we currently find ourselves in.
I’m writing this just a few weeks after the release of AI 2027 (read it if you haven’t yet) and GROCK-4 Heavy (watch the press releases and fanboy hype cycle if you haven’t yet).
Rather than focusing exclusively on the debate between AI DOOMERS (name for people who are calling for safety research, alignment, and who are concerned that ASI represents an existential threat to humanity, including Ilya Sutskever and Elon Musk) and AI ZOOMERS (name for people who are not as concerned with AI safety, and who are enthusiastic to accelerate the development of AGI/ASI, in part out of the belief that advanced AI will solve existential threats to humanity, e.g. cancer, clean energy, and global warming, etc., including people like Sam Altman and).
Hao documents OpenAI’s evolution from an idealistic nonprofit to a Microsoft-backed powerhouse. Internal drama, including Altman’s 2023 ousting and rapid return. And most importantly, the global consequences, such as the exploitation of low-wage labor in Kenya and environmental costs of massive AI infrastructure (hence the title).
At the bottom, Hao’s central critique is that the AI boom mirrors colonial empires in their extraction of data and natural resources (e.g. energy and water at extremely alarming rates), and exploitation of vulnerable, impoverished labor (e.g. Latin American, East Asian, and African educated/poor people) to do the loathsome and often psychologically damaging work of conditioning AI data, frequently for less than a dollar a day, while other AI workers, executives, and investors make literally billions. And (as an added bonus) while conferring immense political power to an elite super rich, super small cadre of AI tech elites.
Hao argues that (in addition to the dizzyingly complex AI ethical problems) the AI empire could undermine global equity, labor rights, and even democratic governance. Particularly if it continues BAU without real democratic oversight.
Great book.
5/5⭐️
Definitely read AI 2027 by Daniel Kokotejio and the AI Futures project too.
June 20, 2025
AI needs giant data complexes, vast amounts energy & water, located in 3rd world countries, sweat shop workers, destroying environments. Like past colonial empires, it just extracts resources for rich nations. Run by a handful of elites like Sam Altman of ChatGPT backed by Microsoft's billions, "OPEN"AI supposed to have been for humanity, but collecting more and more data and just making money became its only end. With no safety limits, this technology, unlike any in history, because humans might not have control over these machines. Different than the hype about how great AI will be for us all.
October 24, 2025
"This book is not a corporate book. While it tells the inside story of OpenAI, that story is meant to be a prism through which to see far beyond this one company. It is a profile of a scientific ambition turned into an aggressive ideological, money-fueled quest; an examination of its multifaceted and expansive footprint; a meditation on power."
More like 4.5 stars but the level of research and work here, it deserves rounding up.
Disclosure: I'm an AI hater so I know I'm biased but I really enjoyed this coverage of how the altruistic ideals behind LLMs and OpenAI specifically as a company (which started as a nonprofit!) have completely gone to the wayside and LLMs/AI have become modern day "Empires" attempting to take over as many resources as possible for themselves without reckoning with the impact they are having
You can watch my full review here: https://youtu.be/ONxxSpn6FJM
June 1, 2025
Let’s be clear from the jump: this book is not a neutral account of AI or its current stewards. The author approaches the topic with a very particular lens—namely, that the world can be neatly divided into oppressors and the oppressed. Guess which one Sam Altman and OpenAI are?
The portrayal of Sam Altman and OpenAI reads like something lifted from a student op-ed in a college paper titled The Daily Marxist. Altman is cast as a kind of techno-capitalist Bond villain, which would be more compelling if it weren’t layered with so much ideological frosting that it’s hard to find the actual cake underneath. OpenAI, meanwhile, is described with such suspicion and scorn you'd think their engineers were developing Skynet in between brunches at Davos.
To be fair, critique of tech power structures is valid and necessary. But what we get here isn’t thoughtful investigation—it’s a litany of grievances that often feel like they were brainstormed in a drum circle. Colonialism? Check. Patriarchy? Check. Capitalism as the root of all evil? Triple check.
Roughly two-thirds of the book is what I can only describe as grievance-filled nonsense, the kind of stuff that gives serious criticism a bad name.
That said, the book isn’t entirely without merit. Buried beneath the sanctimony are a few genuinely good questions about power, bias, and accountability in AI. But these moments are few, fleeting, and often immediately drowned out by another impassioned rant about how unfair life is for everyone other than Sam Altman.
Final Verdict: If you’re looking for a balanced, insightful exploration of AI, this probably isn’t your book. If you’re looking for a political diary with some scattered tech terms tossed in like croutons—then hey, bon appétit. Just don’t expect a fair portrait of Altman, OpenAI, or anything resembling objective analysis. This is less "deep dive into AI ethics" and more "personal TED Talk sponsored by vibes."
Rating: 2/5 stars
One star for the occasional interesting idea, and another for making me laugh—though not always in the way the author intended.
The portrayal of Sam Altman and OpenAI reads like something lifted from a student op-ed in a college paper titled The Daily Marxist. Altman is cast as a kind of techno-capitalist Bond villain, which would be more compelling if it weren’t layered with so much ideological frosting that it’s hard to find the actual cake underneath. OpenAI, meanwhile, is described with such suspicion and scorn you'd think their engineers were developing Skynet in between brunches at Davos.
To be fair, critique of tech power structures is valid and necessary. But what we get here isn’t thoughtful investigation—it’s a litany of grievances that often feel like they were brainstormed in a drum circle. Colonialism? Check. Patriarchy? Check. Capitalism as the root of all evil? Triple check.
Roughly two-thirds of the book is what I can only describe as grievance-filled nonsense, the kind of stuff that gives serious criticism a bad name.
That said, the book isn’t entirely without merit. Buried beneath the sanctimony are a few genuinely good questions about power, bias, and accountability in AI. But these moments are few, fleeting, and often immediately drowned out by another impassioned rant about how unfair life is for everyone other than Sam Altman.
Final Verdict: If you’re looking for a balanced, insightful exploration of AI, this probably isn’t your book. If you’re looking for a political diary with some scattered tech terms tossed in like croutons—then hey, bon appétit. Just don’t expect a fair portrait of Altman, OpenAI, or anything resembling objective analysis. This is less "deep dive into AI ethics" and more "personal TED Talk sponsored by vibes."
Rating: 2/5 stars
One star for the occasional interesting idea, and another for making me laugh—though not always in the way the author intended.
June 13, 2025
Expertly reported investigation into the AI industry and its global goals and impacts, all channeled primarily through a portrait of OpenAI, which under Sam Altman's leadership began as a nonprofit with a vague mission statement about helping humanity but swiftly became a for-profit moral black hole that exploits workers, resources, and others' creativity. Hao makes an incredibly potent comparison between the tech industry and colonial powers, and from now on I don't think I'll ever see the OpenAIs, Googles, Amazons, etc. any other way.
She also embeds deeply with many of these companies' leaders, and unsurprisingly, these are the dumbest "smart" motherfuckers on the planet. A bunch of inhumane, paranoid, self-deluded mini-Musks who are being allowed to create an empire that perpetuates great harm -- not, to be clear, because AI is going to become sentient and take over the world; the product, uh, sucks actually -- even though they themselves have no clothes and no idea what the fuck they're doing.
There are uses of AI that could be a genuine benefit to humanity, and Hao details some of them, but instead everyone -- including, sadly, the company that published this book -- have bought into generative AI's empty bullshit.
It doesn't have to be this way. It doesn't have to be this way.
She also embeds deeply with many of these companies' leaders, and unsurprisingly, these are the dumbest "smart" motherfuckers on the planet. A bunch of inhumane, paranoid, self-deluded mini-Musks who are being allowed to create an empire that perpetuates great harm -- not, to be clear, because AI is going to become sentient and take over the world; the product, uh, sucks actually -- even though they themselves have no clothes and no idea what the fuck they're doing.
There are uses of AI that could be a genuine benefit to humanity, and Hao details some of them, but instead everyone -- including, sadly, the company that published this book -- have bought into generative AI's empty bullshit.
It doesn't have to be this way. It doesn't have to be this way.
June 8, 2025
This book has much to commend it. If you are looking for an explanation of how the crisis that led to Sam Altman’s temporary ouster as CEO of OpenAI came to be, you’ll find it is well handled. Hao provides a lot of detail about something that seemed strange at the time - the board saying Altman wasn’t trustworthy but offering little in the way of details of what that meant. Here, you get those details.
Hao also has a lot to say about the costs of AI that people have probably heard of but maybe not thought of. The costs to environment and the reliance on human labor paid at subsistence-level wages is each given a full chapter. She also does a fine job of highlighting the slow erosion of ideals behind OpenAI in particular, from an organization that starts pursuing AI for the benefit of all humanity to one that appears much more interested in accumulating wealth.
But the book has flaws too. The narrative structure itself was confusing, with Hao hopping back and forth through the timeline to try and tie individual events to points she was making. I also found the overall tone overly cynical. It’s admirable to make sure we understand the costs to people and the environment that these companies are happy for us to ignore. But the author fails to attribute any motivation to the actors beyond greed and a bent toward exploitation. Her chapter on Altman’s sister is a case in point. This is no doubt a complex and difficult situation, but the sister is portrayed solely as a victim who is incapable of agency and someone whose motives are only questioned by people who fail to genuinely care for her. Maybe this is all justifiable, but the I found Hao’s credibility diminished by the one-sided nature of her critique.
I also thought the book was too long. Hao has clearly done her research and has access to a lot of material. But there are several places where the length and amount of quotes felt like they were more to establish the authors bonafides than to offer insight.
This is a book that frankly won’t appeal to most people I suspect. You’ll enjoy it if you like exposes of powerful companies and revel in gossipy corporate stories. You’ll also enjoy it if you enjoy judging people who have accumulated wealth and influence. I read it mostly for the story of how OpenAI came to be where it is. On this front, the results were mixed. I got the story, but it wasn’t told cleanly and was a bit weight down by being hit over the head with the author’s views on Altman.
Hao also has a lot to say about the costs of AI that people have probably heard of but maybe not thought of. The costs to environment and the reliance on human labor paid at subsistence-level wages is each given a full chapter. She also does a fine job of highlighting the slow erosion of ideals behind OpenAI in particular, from an organization that starts pursuing AI for the benefit of all humanity to one that appears much more interested in accumulating wealth.
But the book has flaws too. The narrative structure itself was confusing, with Hao hopping back and forth through the timeline to try and tie individual events to points she was making. I also found the overall tone overly cynical. It’s admirable to make sure we understand the costs to people and the environment that these companies are happy for us to ignore. But the author fails to attribute any motivation to the actors beyond greed and a bent toward exploitation. Her chapter on Altman’s sister is a case in point. This is no doubt a complex and difficult situation, but the sister is portrayed solely as a victim who is incapable of agency and someone whose motives are only questioned by people who fail to genuinely care for her. Maybe this is all justifiable, but the I found Hao’s credibility diminished by the one-sided nature of her critique.
I also thought the book was too long. Hao has clearly done her research and has access to a lot of material. But there are several places where the length and amount of quotes felt like they were more to establish the authors bonafides than to offer insight.
This is a book that frankly won’t appeal to most people I suspect. You’ll enjoy it if you like exposes of powerful companies and revel in gossipy corporate stories. You’ll also enjoy it if you enjoy judging people who have accumulated wealth and influence. I read it mostly for the story of how OpenAI came to be where it is. On this front, the results were mixed. I got the story, but it wasn’t told cleanly and was a bit weight down by being hit over the head with the author’s views on Altman.
June 27, 2025
Just because a technology can be developed doesn’t mean it should be. Research at MIT has already shown that using ChatGPT to write essays makes our brains lazy and us more stupid.
Algorithms — the dumb cousin of generative AI — has been around for ages. Even these primitive AI’d have caused havocs and genocides (Rohingyas of Myanmar as one example) because they have been instructed to increase user engagement above all else. Thus the lies and untruths thrive.
In the words of Yuval Noah Harari, «most information is junk» and the truth is a rare and costly thing. We already have people’s opinion spreading like a stinking fart through all social media. AI will hallucinate answers and uneducated people will spread these lies.
Admittedly, I do use AI almost daily. Never for reviews or when posting anything on SoMe, but certainly to structure ideas and tidy up documents. It has its uses, but will it make the world a better place? Hell no. Just the energy it requires sticks a hole in that theory, completely disregarding the other numerous horrors that will follow.
The reason I think this is because of the damage the tech bros have already cause. They want to increase their billions at the detriment of everyone else. I thought Sam Altman might be more altruistic, but hell no. This book thoroughly reveals him to be as a conniving self serving bastard as the rest of them.
Algorithms — the dumb cousin of generative AI — has been around for ages. Even these primitive AI’d have caused havocs and genocides (Rohingyas of Myanmar as one example) because they have been instructed to increase user engagement above all else. Thus the lies and untruths thrive.
In the words of Yuval Noah Harari, «most information is junk» and the truth is a rare and costly thing. We already have people’s opinion spreading like a stinking fart through all social media. AI will hallucinate answers and uneducated people will spread these lies.
Admittedly, I do use AI almost daily. Never for reviews or when posting anything on SoMe, but certainly to structure ideas and tidy up documents. It has its uses, but will it make the world a better place? Hell no. Just the energy it requires sticks a hole in that theory, completely disregarding the other numerous horrors that will follow.
The reason I think this is because of the damage the tech bros have already cause. They want to increase their billions at the detriment of everyone else. I thought Sam Altman might be more altruistic, but hell no. This book thoroughly reveals him to be as a conniving self serving bastard as the rest of them.
July 7, 2025
AI/Open AI/AGI/ChapGPT.....these are things I hear about but haven't really delved into to understand or "use". This was quite an education. Since I don't know enough about the field to assess accuracy or "slant" or thoroughness, I can only attest to what seemed to be a pretty informative history of the evolution of certain companies and the major players associated with them.
As with all things "progress", there are potential benefits and costs involved. This book identifies many of those aspects: what AI might do to assist humanity, as well as the tremendous resources it requires which might ultimately be destructive to humanity; the potential replacement factors in terms of jobs and having a purpose in life; the extraction of resources from countries and areas who need to earn a living, while benefiting those well-off elsewhere. As with all power, there are skirmishes between major players to determine who has that power and how it will be used. The book also went into characteristics and issues associated with those who are well-known in the field (Altman, Musk, Thiel) and the division between those who see this as a "boom" and those who look more at the "doom" aspects.
These is much to ponder about the direction this is headed, with much uncertainty as to whether this will ultimately be of benefit to people or used for manipulative and self-serving needs of the few. I wish I had more faith in those leading the charge.
As with all things "progress", there are potential benefits and costs involved. This book identifies many of those aspects: what AI might do to assist humanity, as well as the tremendous resources it requires which might ultimately be destructive to humanity; the potential replacement factors in terms of jobs and having a purpose in life; the extraction of resources from countries and areas who need to earn a living, while benefiting those well-off elsewhere. As with all power, there are skirmishes between major players to determine who has that power and how it will be used. The book also went into characteristics and issues associated with those who are well-known in the field (Altman, Musk, Thiel) and the division between those who see this as a "boom" and those who look more at the "doom" aspects.
These is much to ponder about the direction this is headed, with much uncertainty as to whether this will ultimately be of benefit to people or used for manipulative and self-serving needs of the few. I wish I had more faith in those leading the charge.
August 4, 2025
The best of the AI books I’ve read so far.
Hao writes the first draft of history and it ain’t pretty. The monsters are devouring us as our gatekeepers enable them.
The myth of genius contributes to the fake-world the monsters claim to be building. Exploitation and alienation are features not bugs in the demon haunted world created by the masters of the universe. They induce fear out of whole cloth and offer solutions for their imaginary fears.
Hao writes an outstanding book warning us against the madmen who are running the asylum, while we who inhabit the asylum are distracted.
Fitting, the last line of the book, “this book is explicitly prohibited from being used by AI.”
Hao writes the first draft of history and it ain’t pretty. The monsters are devouring us as our gatekeepers enable them.
The myth of genius contributes to the fake-world the monsters claim to be building. Exploitation and alienation are features not bugs in the demon haunted world created by the masters of the universe. They induce fear out of whole cloth and offer solutions for their imaginary fears.
Hao writes an outstanding book warning us against the madmen who are running the asylum, while we who inhabit the asylum are distracted.
Fitting, the last line of the book, “this book is explicitly prohibited from being used by AI.”
Read
September 27, 2025Learnt my GPUs from my AGIs
November 9, 2025
Empire of AI is ambitious, spirited, consequently messy and all over the place. I think OpenAI is already overhyped and overvalued (news this week claiming it to be too big to fail, hope USGOV doesn't tend to bail it out), anyhow Sam Altman would probably end up like Sam Bankman, so no fan of sama drama but Hao's chronicles were too long and disorderly and ironically felt like generated by earlier versions of chatGPT. I really wished she edited this better and made it a definitive guide as was advertised.
May 28, 2025
A must-read for the AI era that's an interesting look into OpenAI -- one of the most defining companies of our time -- but so much more. This is a deeply researched sociological exploration of the power dynamics of the AI industry, its ethics, environmental impact, and the dual possibilities of AI as either an equalizer, or a tool to perpetuate systemic inequality, depending on how the coming years play out. I highly recommend for anyone working in, or simply interested in, AI and its impacts on society.
A prescient quote from the epilogue: "The critiques that I lay out in this book of OpenAI's and Silicon Valley's broader vision are not by any means meant to dismiss AI in its entirety. What I reject is the dangerous notion that broad benefit from AI can only be derived from -- indeed, will ever emerge from -- a vision for the technology that requires the complete capitulation of our privacy, our agency, and our worth, including the value of our labor and art, toward an ultimately imperial centralization project."
A prescient quote from the epilogue: "The critiques that I lay out in this book of OpenAI's and Silicon Valley's broader vision are not by any means meant to dismiss AI in its entirety. What I reject is the dangerous notion that broad benefit from AI can only be derived from -- indeed, will ever emerge from -- a vision for the technology that requires the complete capitulation of our privacy, our agency, and our worth, including the value of our labor and art, toward an ultimately imperial centralization project."
November 1, 2025
I liked this book. I also hated it. Which, according to the internet, means it’s probably important.
It’s about AI, and all the brilliant, sleep-deprived people trying to build a machine that understands humans. Which is funny, considering most of us can’t even understand our neighbours, our families, or why our Wi-Fi keeps cutting out. 🤭
The book dives into data farms in places like Kenya and Venezuela, and honestly, that part just made me want to quit my job in Tech. Because while some guy in Silicon Valley is arguing over ethics in AI, there is someone halfway across the world labeling photos of cats and sex predators for 12 hours straight… and for what? For a machine that might one day write poems and books about loneliness. Progress, right?
And then there’s the corporate drama. The glorious circus of insecure geniuses! Smart men with fragile egos building empires out of code and caffeine. The kind who say things like “We are changing the world”while simultaneously forgetting to answer their interns’ emails.
It is fascinating. It is horrifying. It is the kind of book that makes you go, ‘Ah, so this is how the apocalypse starts. With a startup.’
Well, in the end if you enjoy reading about humanity’s finest brains behaving like middle schoolers with funding, do read this book.
It’s about AI, and all the brilliant, sleep-deprived people trying to build a machine that understands humans. Which is funny, considering most of us can’t even understand our neighbours, our families, or why our Wi-Fi keeps cutting out. 🤭
The book dives into data farms in places like Kenya and Venezuela, and honestly, that part just made me want to quit my job in Tech. Because while some guy in Silicon Valley is arguing over ethics in AI, there is someone halfway across the world labeling photos of cats and sex predators for 12 hours straight… and for what? For a machine that might one day write poems and books about loneliness. Progress, right?
And then there’s the corporate drama. The glorious circus of insecure geniuses! Smart men with fragile egos building empires out of code and caffeine. The kind who say things like “We are changing the world”while simultaneously forgetting to answer their interns’ emails.
It is fascinating. It is horrifying. It is the kind of book that makes you go, ‘Ah, so this is how the apocalypse starts. With a startup.’
Well, in the end if you enjoy reading about humanity’s finest brains behaving like middle schoolers with funding, do read this book.
June 16, 2025
Such a disappointment! Still well documented list of events happening at OAI, written like one article piece that's journalistic. The whole personal life attacks were unwarranted IMO. She could still have made her point about Altman without dragging the family issues into this saga. In that sense too, Hao was exploitative! What's wrong with Americans!
July 23, 2025
Fine and all but I really have to stop reading books by journalists (who tell stories) when I want to read books by academics (who describe and explain).
Actually I have more to say because this isn't exactly true. I did enjoy Number Go Up: Inside Crypto's Wild Rise and Staggering Fall by Zeke Faux, an investigative journalist's book about SBF and other crypto bros. That was interesting because he was a character in his book interacting with the players he was writing about. He also wrote with a palpable level of disbelief, disdain, and most of all humor. I think that's what I was hoping for here. The difference is that Empire of AI aims to tell you just the facts. Quotes are pulled from prior interviews and court recordings instead of first person interactions and it ends up being more Ken Burns than Michael Moore. Which is fine and all, but still it's light on theory and not a revelation nor a takedown either so I didn't feel like I was armed with any new applicable knowledge nor was I ever like cheering or jeering ykwim? It just kind of was.
Actually I have more to say because this isn't exactly true. I did enjoy Number Go Up: Inside Crypto's Wild Rise and Staggering Fall by Zeke Faux, an investigative journalist's book about SBF and other crypto bros. That was interesting because he was a character in his book interacting with the players he was writing about. He also wrote with a palpable level of disbelief, disdain, and most of all humor. I think that's what I was hoping for here. The difference is that Empire of AI aims to tell you just the facts. Quotes are pulled from prior interviews and court recordings instead of first person interactions and it ends up being more Ken Burns than Michael Moore. Which is fine and all, but still it's light on theory and not a revelation nor a takedown either so I didn't feel like I was armed with any new applicable knowledge nor was I ever like cheering or jeering ykwim? It just kind of was.
June 4, 2025
A sobering piece of investigative journalism that tracks OpenAI from inception to 2025, delving into the fractured vision and deeply irresponsible, unethical and unsustainable business practices of the company and CEO. Thoughtfully researched and presented, this is a really useful read for anyone wondering what generative AI sort-of is, what it was sort-of-meant to be, why it's terrible for literally everyone in its current form (including, shockingly, the investors haemorrhaging money into it), and how the hell it was all sprung on us so quickly.
pausado
June 19, 2025On hold for the moment. I thought it was going to be a different type of book and, at least for now, seems to be just a biography made by your nemesis.
I was not aware of the relationship of the author with OpenAI, but reading the nice slogans on the back of the book I thought there would be more of, for example, the environmental impacts of AI and things of the sort. Maybe there is some of this later on, but I need a break for now
I was not aware of the relationship of the author with OpenAI, but reading the nice slogans on the back of the book I thought there would be more of, for example, the environmental impacts of AI and things of the sort. Maybe there is some of this later on, but I need a break for now
Read
June 8, 2025i learned quite a lot from this.
i have quibbles with a few of Hao’s choices (e.g. was cringing at attempt to shoehorn annie altman’s story into the empire framing). and there were a few heavy handed moments of interpretation that i felt sort of undermined her credibility. but it’s mostly fine and good that she’s writing from a specific perspective…welcome to the marketplace of ideas….
still processing takeaways, but for now the thing i’ve been thinking of most is how goofy it is that safety teams are fixated on theoretical future “agi” paper clip factory esque scenarios. meanwhile, non-theoretical harm unfolds in realtime…workers in the global south are paid $1/hour to label hideous content with no labor protections or psychological support to speak of and data centers get cooled with fresh, filtered water as people who live nearby have to go buy bottled water to drink. it’s really giving “effective altruists who devote their entire lives to maximizing utility for humanity while being soooooo toxic to their friends and family” (see: sbf) … cut from the same cloth / a lot of overlap in the two communities as i understand it. it’s almost like it’s easier to deal with the world in abstract than confront messy reality…..
i have quibbles with a few of Hao’s choices (e.g. was cringing at attempt to shoehorn annie altman’s story into the empire framing). and there were a few heavy handed moments of interpretation that i felt sort of undermined her credibility. but it’s mostly fine and good that she’s writing from a specific perspective…welcome to the marketplace of ideas….
still processing takeaways, but for now the thing i’ve been thinking of most is how goofy it is that safety teams are fixated on theoretical future “agi” paper clip factory esque scenarios. meanwhile, non-theoretical harm unfolds in realtime…workers in the global south are paid $1/hour to label hideous content with no labor protections or psychological support to speak of and data centers get cooled with fresh, filtered water as people who live nearby have to go buy bottled water to drink. it’s really giving “effective altruists who devote their entire lives to maximizing utility for humanity while being soooooo toxic to their friends and family” (see: sbf) … cut from the same cloth / a lot of overlap in the two communities as i understand it. it’s almost like it’s easier to deal with the world in abstract than confront messy reality…..
Read
July 21, 2025The inside scoop on OpenAI’s board coup was very revealing. Overall this book is a skewering of the company. The thesis is that "empires" reward a select few and devastate the rest and that AI is the world’s latest rendition of empire.
It’s insane to me that Sam Altman raves about the Manhattan Project and thinks OpenAI is the second coming of it. So much love for the flash and so little care for the burn. I think the leaders of AI tend to view the harms of it in distorted fashion. “AI safety” is a major talking point amongst the frontier model companies, and that includes things like LLMs being used for impersonation, harassment, misinformation, so on. But Hao (the author) is skeptical as to how exactly these risks are stack-ranked against the companies’ business goals, and I think incidents even this week like a productionized Grok impersonating Hitler suggest that, wherever it was we started, a lot of diligence has been lost in our current acceleration.
There are the risks that AI leaders talk about, but there are also the ones they don’t talk about. Sam Altman asked Scarlett Johannsson to license her voice for GPT-4o and, when she said no, they copied her voice anyways. OpenAI also fanned the flames during the Studio Ghibli moment in March of this year where everyone used 4o to turn their profile pictures into Spirited Away. I think their views on how image and voice generation models subvert the very economies that made the technology possible — all that free online art that became training data — are majorly lacking in empathy.
What’s also interesting to me is that in the early days of OpenAI they cited climate change as one of the primary examples of issues AI could solve. Down the line the company decided to focus more and more on large language models and to throw as much compute (read, electricity) as physically possible at successive GPT models. As this happened at greater scale, you can imagine how silently the climate change talking point faded into the background. Nowadays I keep hearing AI leaders theorize that AI will solve the climate crisis before it pushes us over the brink and therefore we don’t need to hark about climate anymore. I find the idea that we should let a handful of uber rich people play brinksmanship with the environment an absolutely ludicrous gamble, and I couldn’t be less in favor.
Overall I learned a lot reading this. I think the lens of startups like OpenAI are so zeroed in on success that they don’t see the implications of their success for others. We can’t assume that the model companies are doing rigorous thinking about what AI will mean for the world.
It’s insane to me that Sam Altman raves about the Manhattan Project and thinks OpenAI is the second coming of it. So much love for the flash and so little care for the burn. I think the leaders of AI tend to view the harms of it in distorted fashion. “AI safety” is a major talking point amongst the frontier model companies, and that includes things like LLMs being used for impersonation, harassment, misinformation, so on. But Hao (the author) is skeptical as to how exactly these risks are stack-ranked against the companies’ business goals, and I think incidents even this week like a productionized Grok impersonating Hitler suggest that, wherever it was we started, a lot of diligence has been lost in our current acceleration.
There are the risks that AI leaders talk about, but there are also the ones they don’t talk about. Sam Altman asked Scarlett Johannsson to license her voice for GPT-4o and, when she said no, they copied her voice anyways. OpenAI also fanned the flames during the Studio Ghibli moment in March of this year where everyone used 4o to turn their profile pictures into Spirited Away. I think their views on how image and voice generation models subvert the very economies that made the technology possible — all that free online art that became training data — are majorly lacking in empathy.
What’s also interesting to me is that in the early days of OpenAI they cited climate change as one of the primary examples of issues AI could solve. Down the line the company decided to focus more and more on large language models and to throw as much compute (read, electricity) as physically possible at successive GPT models. As this happened at greater scale, you can imagine how silently the climate change talking point faded into the background. Nowadays I keep hearing AI leaders theorize that AI will solve the climate crisis before it pushes us over the brink and therefore we don’t need to hark about climate anymore. I find the idea that we should let a handful of uber rich people play brinksmanship with the environment an absolutely ludicrous gamble, and I couldn’t be less in favor.
Overall I learned a lot reading this. I think the lens of startups like OpenAI are so zeroed in on success that they don’t see the implications of their success for others. We can’t assume that the model companies are doing rigorous thinking about what AI will mean for the world.
June 4, 2025
很精彩，深入openai的内部，从该非营利组织的创立到转变为营利性公司的整个过程都进行了深入剖析，阿尔特曼被政变以及几天后的回归的戏剧性过程写得很深入。openai内部的理想主义者们在和商业主义者们的斗争中的步步失败让人颇为唏嘘。
本书对AI发展过程的问题（超级计算中心对南方国家环境的破坏、数据标注工作对南方人民的压榨等等）也写得很到位和清晰。
可惜本书截稿早了半年，以至于没有来得及写到deepseek的斯普尼克冲击到来后的影响。事实上deepseek的冲击是能够回答本书最后一章中的部分问题的。
本书对AI发展过程的问题（超级计算中心对南方国家环境的破坏、数据标注工作对南方人民的压榨等等）也写得很到位和清晰。
可惜本书截稿早了半年，以至于没有来得及写到deepseek的斯普尼克冲击到来后的影响。事实上deepseek的冲击是能够回答本书最后一章中的部分问题的。
July 11, 2025
This book is not to be confused with an objective or neutral look at Silicon Valley's attempts to build AI. Hao eschews the rose-colored lenses for a pessimistic and negative portrayal of OpenAI, Sam Altman, other competitors in the AI field, and the impacts of AI on the greater world. While this book markets itself as focusing on Altman and Open AI, that really isn't true. Altman and AI do occupy a good sized portion of the book, but Hao also spends time discussing what AI is, how it originated, AI's development in the 20th Century, how it has advanced in the 21st Century and how it hasn't, its environmental effects (disastrous), and ultimately the search for AGI (which is what us laypeople generally understand as ultimate AI that we see in movies, tv shows, etc.).
Several chapters of the book were quite informative, especially the early ones discussing AI, the current iterations of it at the major tech companies (LLMs, neural networks, etc.), and how these tech companies have eschewed other forms of AI (and why). But Hao loses the thread so to speak, sometimes going far afield of Open AI and Altman, focusing on the environmental impacts and societal impacts of developing these LLMs (on poorer workers in the Global South). While I appreciate that light has to be shined on these negative externalities of AI, they did not require the lengthy chapters that Hao devoted to them, which ultimately slowed down the book greatly. Additionally, Hao falls into redundancies with a lot of the first couple of chapters repeating at the end, making me wonder if a better editor could've shrunk this book by 50 pages without losing much.
Ultimately though, I found it to be an insightful, educational read. I'd recommend this to anyone who is curious to learn more about AI and has limited knowledge to bring to the table. If you already know a lot about the field and AI in general, I'd skip this.
3.75 stars.
Several chapters of the book were quite informative, especially the early ones discussing AI, the current iterations of it at the major tech companies (LLMs, neural networks, etc.), and how these tech companies have eschewed other forms of AI (and why). But Hao loses the thread so to speak, sometimes going far afield of Open AI and Altman, focusing on the environmental impacts and societal impacts of developing these LLMs (on poorer workers in the Global South). While I appreciate that light has to be shined on these negative externalities of AI, they did not require the lengthy chapters that Hao devoted to them, which ultimately slowed down the book greatly. Additionally, Hao falls into redundancies with a lot of the first couple of chapters repeating at the end, making me wonder if a better editor could've shrunk this book by 50 pages without losing much.
Ultimately though, I found it to be an insightful, educational read. I'd recommend this to anyone who is curious to learn more about AI and has limited knowledge to bring to the table. If you already know a lot about the field and AI in general, I'd skip this.
3.75 stars.
May 30, 2025
A good compiling of all the issues around OpenAI and tearing down of all the marketing veneer.
Hao also goes deeply into the costs of AI on people around the world.
Hao also goes deeply into the costs of AI on people around the world.
August 21, 2025
Brutal
August 2, 2025
When reading an essay or book about someone or some industry that you don’t know much about, you are at the mercy of the author for what she chooses to show you.
So you have to trust your author.
Empire Of AI is written by Karen Hao. I don't trust her. First I don't think she believes in capitalism. And her treatment of the accusations of Sam Altman's sister felt like a cheap shot. We don’t really know the story behind the sister and why the family closed ranks against her. But Hao tries to tar Altman with an unsubstantiated accusation. They could easily be true. But they could be the angry and unhinged attack by a sister (and daughter) who feels betrayed and attacked by her family. A court case is ongoing.
But Hao wants to paint Sam Altman as a lying, manipulative, untrustworthy titan of industry. He may be that, but throwing that accusation in the introduction puts me on guard against this book.
I presume that Sam Altman is a capitalist who wants to make money. I don’t think I ever thought that Open AI was in any way altruistic.
Should Standard Oil have been allowed to create a monopoly in refining oil in the late 19th Century? I don’t think Standard Oil (or John D. Rockefeller) was evil. We had to create laws to reign in his predatory behavior. But Standard Oil helped push the United States to the forefront of oil refining. You have to have big corporations who want to make money to make your economy sound.
Does AI or GAI require more safeguards? Maybe. Should the US be in the forefront of AI (instead of China)? Definitely.
I don’t think Karen Hao believes that.
So you have to trust your author.
Empire Of AI is written by Karen Hao. I don't trust her. First I don't think she believes in capitalism. And her treatment of the accusations of Sam Altman's sister felt like a cheap shot. We don’t really know the story behind the sister and why the family closed ranks against her. But Hao tries to tar Altman with an unsubstantiated accusation. They could easily be true. But they could be the angry and unhinged attack by a sister (and daughter) who feels betrayed and attacked by her family. A court case is ongoing.
But Hao wants to paint Sam Altman as a lying, manipulative, untrustworthy titan of industry. He may be that, but throwing that accusation in the introduction puts me on guard against this book.
I presume that Sam Altman is a capitalist who wants to make money. I don’t think I ever thought that Open AI was in any way altruistic.
Should Standard Oil have been allowed to create a monopoly in refining oil in the late 19th Century? I don’t think Standard Oil (or John D. Rockefeller) was evil. We had to create laws to reign in his predatory behavior. But Standard Oil helped push the United States to the forefront of oil refining. You have to have big corporations who want to make money to make your economy sound.
Does AI or GAI require more safeguards? Maybe. Should the US be in the forefront of AI (instead of China)? Definitely.
I don’t think Karen Hao believes that.
Displaying 1 - 30 of 893 reviews