At this stage, it’s the humans that train robots via reinforcement learning in Matrix -esque simulations or by playing the ...