Also, given that we don’t code AI these days, but rather let it evolve in some specific conditions, which we don’t always control and often cannot explain how the evolved AI works, how are we going to enforce Three Laws of Robotics, or something similar?