Elon Musk gives us an 80% shot at a good outcome—but what about the other 20%? As AI moves toward superintelligence, the risk of catastrophic failure grows. From rogue machines to algorithmic dictators, the danger isn’t in the future. It’s knocking now.