The most dangerous assumption in AI safety is that we must control an AI that can outsmart us. The second is that we can control how fast we build it.