On fitting an AI with a listening hood. Prologue: This Is Not a Story About the Future When people talk about the risks of AI, one thought experiment comes up again and again: the paperclip maximizer , made famous by the philosopher Nick Bostrom. You give an AI a goal that sounds perfectly harmless— manufacture paperclips as efficiently as possible. Then its intelligence surpasses ours. To secure raw material, and to keep anything from interrupting production, it calmly disassembles us along wit