When is Google going to come up with a system that shows how a question or query was interpreted by its AI in terms of a semantic/knowledge/NLP graph that we can then adjust interactively in order to train its deep learning system?
Yes. Wolfram Alpha tells you how it interpreted your query, so if you get back an answer of "42", you know how it got that answer. Google likes to be more opaque than that.
As Google gets more into answering questions rather than returning links, this will become more of a problem.
