Because the situation seems genuinely serious and urgent.

If you carefully consider a topic, you can sometimes see one of history’s zigs or zags coming.

In 1933, a physicist named Leo Szilard was the first person to realize that nuclear chain reactions are possible.^* He thereby gained the ability to predict one of history’s zigs earlier than anyone else.

We think that if you look at AI from the right vantage point today, you can see one of history’s zags coming. And we think that events are set to go poorly if humanity does not change course.

AI labs are racing to build machines that are smarter than any human, and they’re making what appears to be significant progress in advancing the frontier. As we’ll discuss in the coming chapters, modern AIs are more grown than crafted. They exhibit behavior that nobody asked for and nobody wanted, and they’re on track to become more capable than any human. This looks like an extremely dangerous situation to us.

Top scientists in the field came together to sign an open letter warning the public that the threat of AI should be treated as a “global priority alongside other societal-scale risks such as pandemics and nuclear war.” This is not an isolated concern; it’s shared by nearly half the field. Even if you are initially skeptical of the dangers, we hope that the level of stated concern by AI experts, and the high stakes if these concerns turn out to be correct, make it clear why this is a topic that deserves to be seriously talked out.

This is a topic where we should weigh the arguments rather than just following our first intuition. If the letters and the warnings are correct, then the world has gotten itself into an incredibly dangerous position. And we’ll spend the rest of the book laying out the arguments and evidence behind those warnings.

We do not think the situation is hopeless. We wrote this book with the hope of changing the trajectory humanity seems to be on, because we think there’s hope that we can solve this.

The first step toward solving a problem is to understand it.