Ok thank you for your answer. So many things are claimed that it is hard to track what is real and what is just hype. I agree that it's something big. Training

Ok thank you for your answer. So many things are claimed that it is hard to track what is real and what is just hype. I agree that it's something big. Training Alpha Go on itself means something bigger than "just" optimizing a statistical model on human data. I think recognition of logic elements to strategy planning are parts of what will make ML really close to IA. (With memory, and cleverness to learn) And are the next big steps.