Bookmarks: "If Anyone Builds It, Everyone Dies" by Eliezer Yudkowsky & Nate Soares
Published September 2025 by Little, Brown and Company
By Drew Foster
Near the end of Eliezer Yudkowsky and Nate Soares’ sobering new nonfiction book detailing the existential threat posed by artificial superintelligence, the authors describe a scene on the Titanic when the ship begins to sink. Some concerned passengers clamber into lifeboats; many others refuse to heed emergency announcements, a few going as far as to belittle the folks seeking safety. The ship’s unsinkable! You’re overreacting! There’s nothing to worry about! We all know how that one ended.
Yudkowsky and Soares, both veterans in the AI field who are co-leaders of the nonprofit Machine Intelligence Research Institute, believe humanity has reached a similar moment regarding artificial intelligence. Alarm sirens are blaring. The ship’s taking on water. It’s time for emergency maneuvers. "If Anyone Builds It, Everyone Dies" is their call to action.
The good news is that artificial superintelligence does not yet exist. The bad news is that no one knows when it could come about. One year? Five years? Ten years? It is not known, but the authors don’t believe it’s far off. And we, as a society, seem to be chugging full steam ahead toward its creation. A New York Times article from April 1 noted that artificial intelligence companies raked in a record-smashing $297 billion in funding through the first three months of this year. So much for pumping the brakes.
Yudkowsky and Soares lay out their straightforward argument early in the book:
"If any company or group, anywhere on the planet, builds an artificial superintelligence using anything remotely like current techniques, based on anything remotely like the present understanding of AI, then everyone, everywhere on Earth, will die."
Yikes.
Their case unfolds in three parts:
Part One tells the reader how artificial intelligence is produced and why it’s so difficult to understand. The authors ask if AIs can have wants, and what they might desire. They predict "strange, alien preferences that [AIs] pursue to the point of human extinction."
Part Two imagines how a well-intended artificial superintelligence could go terribly awry. I was expecting nuclear war, but I was wrong. The authors formulate more nuanced, more chilling scenarios.
Part Three examines the challenges ahead, reviews what AI companies are doing about them, and scrutinizes why the world isn’t taking stronger action. It also provides suggestions for individuals to help prevent AI catastrophe.
"If Anyone Builds It, Everyone Dies" is an important work. The authors mention at the book’s end that 69% of U.S. voters surveyed in 2023 thought AI should be regulated as a dangerous and powerful technology. I recommend this book for that 69%. I also recommend it for the remaining 31%. On the fence about the potential of AI? Read this book to get a better understanding. Enthusiastic about the possibilities AI holds for our future? Read this book to learn about alternative scenarios. It’s available through the Hawaiʻi State Public Library System.
Yudkowsky and Soares cite Nobel Prize-winning "Godfather of AI" Geoffrey Hinton’s belief that there’s a better-than-50% chance artificial superintelligence renders humanity extinct. That’s a coin flip that we’re all going to die. The authors ask: Do we climb into the lifeboat or laugh at the alarms?
