Bryan Johnson imagines telling his 8-year-old nephew about superintelligence arriving in the next decade, comparing it to water turning to steam at a phase transition. He proposes Blueprint, an AI system guided by one principle: don't die. By measuring what kills people and building AI to prevent it, he argues we can navigate radical uncertainty safely.
Hey, Unc! Yes, my immortal nephew. What will happen when we give birth to superintelligence? Huh, what a thoughtful question for an 8-year-old, and I've got just the answer. Chapter 1: Society is about to enter a phase transition. The difference between 2026 and 2036 is like water. Water at 99°C is hot, stable, and liquid, but at 101°C, it becomes a gas unpredictable, and following a different set of physical laws. Now imagine you're an expert swimmer, perfect form in water. Then suddenly the pool hits 101°C and it turns into steam. You try to swim, but you don't move, and every skill you've mastered was built for the old environment. That's what life could feel like in 10 years with superintelligence. So what do we do? Well, in this new reality, the move is not to have better plans, but to build better systems. What kind of systems? An algorithmic system that helps you make healthy choices and keeps up as technology changes. I call it Blueprint. But does that mean AI will replace us? We don't know what AI will do, and that's overwhelming. But there's a new game we can play. It's called Don't die. You see, right now you're spawned into a world designed for death. Everything you eat, drink, and consume. Bringing superintelligence into that world is risky. But what if there was another world? A world where everything is measured and we know what brings death and what brings life. Then we align AI to help us avoid the things that bring death. Don't die. That's exactly right. Because no matter how we feel about AI, We all agree on one thing: we don't want to die right now. Well, I think that's—
