This book leans heavily on extreme, speculative scenarios to argue that artificial superintelligence inevitably leads to human extinction. But that conclusion rests on assumptions that don’t hold up.
First, any advanced AI would, for a long time, depend on human-built infrastructure—power, data centers, networks, and maintenance. That dependence gives humanity real leverage. If a system were clearly dangerous, it could be shut down. The idea that it would instantly escape all control ignores how deeply these systems are tied to human operations.
Second, the book treats outcomes as binary—either perfectly aligned or instantly catastrophic. In reality, every powerful technology humanity has created has been both useful and imperfect. We don’t abandon them; we manage them, improve them, and adapt. AI would likely follow the same pattern.
Third, the authors jump from possibility to inevitability. Yes, you can imagine a scenario where everything goes wrong—but that’s not the same as showing it will happen, especially when it requires a chain of unlikely assumptions about total loss of control, perfect coordination by the AI, and zero response from humans.
A more grounded view is that humanity will build, test, and scale these systems while they are still dependent on us—and will retain the ability to intervene. By the time deeper automation arrives, we will have already learned what these systems are capable of and whether they are safe to continue deploying.
This book is just a collection of worst-case thought experiments presented as inevitable future.