Supremacy by Parmy Olson - Book Review
ChatGPT was introduced to the public in November 2022, on a webpage with a simple text box. It felt, for those old enough to remember, reminiscent of early Google.com, which was a reaction to Yahoo’s bloated homepage. Instead of a sea of images and distractions, here was a simple text box that promised to answer all your questions.
Except, instead of directing you to the right webpage, as Google did, ChatGPT responded with human-like long-form answers. The reaction was revolutionary, and today, in late 2025, already one in ten humans on Earth uses ChatGPT every week. The AI category, in general, has seen trillions of dollars invested, accounts for a growing share of the global electric grid, and is making inroads into formerly white-collar professions, including coding, teaching, law, finance, medicine, and more.
But who is actually behind these breathtaking machines? Supremacy by Parmy Olson takes a deep look at two of the biggest players: Google’s DeepMind and OpenAI. Both started off as nonprofits with vague goals of helping humankind, before swiftly pivoting into money-gobbling corporate for-profit machines. Both leaders, Sam Altman and Demis Hassabis, speak in messianic terms about the power of AI and the risks of it falling into the wrong (aka not their) hands.
Personally, I found this book a riveting albeit disquieting read. That so much power is concentrated in the hands of two men has never boded well for humanity’s future. Today, multiple AI firms are attempting to build frontier-type models that ingest the entire internet. However, for a long time, these two firms and their founders were duking it out in the relative shadows.
One small critique of the book is that while Hassabis is a genuine technologist and academic who has rolled up his hands and done pioneering AI work, Altman is much more of a sales and marketing showman. Other technical geniuses behind OpenAI’s rise deserve more prominence, though Altman is a willing and eager spokesman for all of it. If you read Empire of AI by Karen Hao, you see a much more personal portrayal of Altman and how he clawed his way to the top of OpenAI despite not being involved for a long time.
That quibble aside, this book was utterly fascinating. Some quotes that stood out in particular to me from the book:
“The kindest thing you can say about these people is that they were naive in thinking they could be nonprofits but take corporate money, doing things for the good of all. But these are otherwise ruthless individuals, and I suspect it is more accurate to say that they were early to recognize the potential power of AGI, but felt that they alone were qualified to control it and manage the numerous ethical, legal, and economic quagmires that result. It is also worth noting that they all got funded and ultimately quickly. It was more about power and ego than ‘the greater good.’”
“Effective altruism was always a smokescreen of vanity that these men–all white men–were obligated to make money and control the resulting AGI.”
The author also points out, as have others, that these LLMs are built on universal copyright theft and “what the models learn from the open web entrenches the status quo.” Because, despite the promise of actual intelligence in AI, at the end of the day, all these models are next-token prediction machines, trained for now on very human data.
The engineers in charge of the models take post-training steps to make them appear neutral, rounding out sharp edges wherever possible. However, these models are a mirror held up to the web community, which is not representative of human beings as a whole and never has been. These models and their growing ubiquity make those entrenched positions, as well as seats of power, ever harder to overthrow.
My short review of this book is that it should be mandatory reading for anyone interested in AI and LLMs in particular. You need to understand the humans behind the machines to truly grasp what is happening. And in the case of these two men, it’s not always the most flattering picture.