I have been learning about deep learning and machine learning in general. One struck me is that all of this mapping from input to output could be done accurately by the machine, in some respect even better than human. It got me thinking that those mapping from input to output, many people would be easily deceived that it is a causation relationship but in fact many of them would be in correlation relationship. this is tricky, because the confusion between the two would be dangerous. Or at least in some applications we could say that is not just one. two or even x inputs that causes a certain output. I think the world kind of confused which one is correlation and which one is causation. It is very time consuming and energy intensive to establish a causation theory. Most people would just give up or mislead their way of thinking.