Will artificial intelligence help us or hurt us?

Chances are, you’d never come across the term “artificial intelligence” as recently as a decade ago. And even if you had, it was probably either in reading science fiction or working in the tech industry. For the rest of us, AI was far off the radar screen. Yet today, you can’t read or view the news or open a smartphone without coming across some reference to the field. Artificial intelligence is spreading into areas of work and life we could never have anticipated. To understand what it is and how it works, we might turn to one of the many books that seek to explain it all. But none does such a good job of portraying how AI came into prominence as Karen Hao’s explosive new exposé, Empire of AI. The book is both a history of OpenAI and a biography of its cofounder and singular leader, Sam Altman.

A picture of idealism gone awry

Empire of AI is not a primer on artificial intelligence. Nor is it a history of the industry, whose origins lie in the 1950s. In her tight focus on OpenAI and the brash young billionaire who runs it, Hao paints a picture of idealism gone to seed, as the profit motive steadily asserted its primacy. Altman, Elon Musk, and their nine cofounders set up the company in 2015 as a nonprofit. Its founding mission was “to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity.” AGI is the firm’s ultimate goal. But most of the guardrails to prevent its misbehavior are gone. Now, OpenAI pursues ever greater profits like any other company. The advocates of safety above all are overwhelmed by the logic of the first-to-market mindset.

Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI by Karen Hao (2025) 496 pages ★★★★★Microsoft CEO Satya Nadella welcomes Sam Altman back as CEO of OpenAI in 2023 after its staff revolts over his firing. Image: Justin Sullivan – WiredTwo philosophies, two “clans”

Two contrasting philosophical assumptions cleave the AI tech community and OpenAI specifically. Boomers see the technology’s promise to lead humanity to the promised land of prosperity and longer lives. Doomers fear AI’s potential to pose an existential threat to the human race as super-intelligent machines assert their dominance. These two clashing philosophies roughly correspond to the two “clans” Hao views as central to the ongoing struggle within OpenAI. One, sometimes called the Safety Division, seeks to prevent the emergence of threatening behavior by the technology by exhaustively testing for flaws in the software and its training regimen. The other, at times called the Applied Division, rushes to commercialize AI both to speed along the benefit it brings and to maximize the company’s profits.

That formulation may convey the impression that bitter differences divide OpenAi’s staff. They do. As Ilya Sutskever put it in conversation with Hao, Sam Altman and president Greg Brockman preside over “a directionless, chaotic, and backstabbing environment.” And Sutskever, a cofounder, OpenAI’s former chief scientist, and one of the world’s leading experts in the field, held key positions at the company from 2015 to 2024. He triggered the board’s abortive effort to fire Altman because he saw no other remedy for the chaos. After leaving OpenAI, Sutskever cofounded a new company called Safe Superintelligence Inc. Its name reflects his preoccupation with the potential harm of the technology.

Facebook content moderators demonstrating in Kenya outside the office of one of the leading outsourcing companies in the field. Image: Daniel Irungu – CNN BusinessIll effects around the world

Everything I’ve read to date about artificial intelligence and the companies that are pursuing it has focused on the United States and China alone. Empire of AI is different. As the title implies, Hao writes about the industry’s impact around the world. She reports on two under-appreciated trends. The exploitation of people hired in the Third World to cull through mountains of AI output and identify racist, abusive, and violent content. And the construction of ever-larger data centers in such countries as Chile, where they plunder local water and energy resources. Hao calls it latter-day colonialism. And it’s hard to dispute the charge. Hence, the book’s title, Empire of AI.

A solidly researched history of OpenAI, well written

Hao comes to the task of profiling Altman and the company he leads with impeccable credentials. She was the first journalist allowed to interview OpenAI staffers for a profile of the company. She gave it a mixed review in 2021 in her article for the MIT Technology Review. Although executives barred her from entering the premises in her research for this book, she nonetheless managed to interview scores of current and former OpenAI employees. Obviously, the company’s executives are unhappy with the result. But her research is solid, and she knows whereof she writes. This book is a superior point of entry into the reality underlying AI research today.

An experiment: using AI to review this book

Out of curiosity, I asked both Open AI’s free online product, ChatGPT, and Anthropic’s Claude AI to write 500-word reviews of Empire of AI. ChatGPT delivered more than 600 words of general text about the threat and the promise of artificial intelligence without once mentioning either OpenAI or Sam Altman. Claude AI did much better. Much, much better. Here’s what it delivered. It’s verbatim except for the occasional section heading, which I added to make reading easier. As you can see even without reading the book, Claude’s review is right on target.

Claude AI’s review of Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI

Karen Hao’s “Empire of AI” arrives at a crucial moment in the artificial intelligence revolution, delivering a meticulously researched exposé that pulls back the curtain on one of the most influential companies in modern tech history. As an award-winning journalist who covered OpenAI years before ChatGPT became a household name, Hao brings unique insider access and perspective to this compelling narrative.

The book’s central thesis is provocative yet persuasive: that the AI industry, led by companies like OpenAI, operates with the same extractive logic as colonial empires of the past. Hao draws deliberate parallels between historical colonialism and what she terms “AI colonialism,” arguing that these companies seize and extract precious resources—artistic works, personal data, and human labor—to fuel their ambitious visions of artificial general intelligence.

A central focus on Sam Altman

At the heart of the narrative is Sam Altman himself, portrayed as a charismatic yet polarizing figure whose optimism about AI’s transformative potential clashes with his pragmatic, sometimes ruthless business approach. Hao’s reporting, based on interviews with around 260 people plus correspondence and internal documents, reveals the complex dynamics within OpenAI’s leadership and the tension between the company’s founding nonprofit mission and its commercial ambitions.

One of the book’s greatest strengths is its global perspective. Rather than confining her analysis to Silicon Valley boardrooms, Hao examines the hidden human and environmental costs of AI development, from data centers in Chile to data workers in Kenya. This approach provides crucial context often missing from tech journalism, highlighting how the AI boom creates new forms of inequality and exploitation across the Global South.

OpenAI’s shift from nonprofit to capitalism

The timing of the book’s release in May 2025 allows Hao to incorporate the full arc of OpenAI’s transformation, from its nonprofit origins through the ChatGPT explosion and subsequent corporate restructuring. Her narrative captures the breathless pace of AI development while questioning whether this speed comes at the expense of safety, ethics, and global equity.

Hao’s writing is both accessible and urgent, making complex technical and geopolitical concepts understandable without sacrificing depth. The book succeeds in being both a corporate biography and a broader critique of how technological power is consolidated and wielded in the 21st century.

A strained central metaphor

However, some readers may find the colonial empire metaphor occasionally strained, and the book’s focus on OpenAI, while deep, leaves questions about how other AI giants like Google and Microsoft fit into this framework. Additionally, the rapidly evolving nature of AI means some developments may already feel dated by the time readers encounter them.

“Empire of AI” stands as essential reading for anyone seeking to understand not just how we arrived at our current AI moment, but where we might be heading. Hao’s reporting reveals that the choices made by a small group of tech leaders will have profound consequences for global power structures, labor markets, and human agency itself. The book serves as both a warning and a call to action, suggesting that alternative, more equitable approaches to AI development remain possible—if we act soon enough.

This is investigative journalism at its best: rigorous, revelatory, and urgent. In an era when AI hype often drowns out critical analysis, Hao’s work provides the sober assessment we desperately need.

About the authorKaren Hao giving a bookstore talk in 2025. Image: Wikipedia

Karen Hao has written about artificial intelligence since 2018. She worked as a senior artificial intelligence editor at the MIT Technology Review, then moved to the Wall Street Journal and The Atlantic. The Journal hired her in 2022 to report on Chinese technology from Hong Kong for a year. Today she freelances for many of the country’s most prominent magazines. She holds a BA in Mechanical Engineering from MIT. Hao is a native speaker of Mandarin Chinese as well as English.

For related reading

You’ll find many other views on AI at 30 good books about artificial intelligence.

For books about other companies and other business leaders and scientists, see:

My 10 favorite books about business historyScience explained in 10 excellent popular books12 great biographies

And you can always find the most popular of my 2,300 reviews, and the most recent ones, on the Home Page.

The post Will artificial intelligence help us or hurt us? appeared first on Mal Warwick on Books: Insightful Reviews and Recommendations.