Simply, no. Most of big advances in the past five years have been around Natural Language 'Parsing' of unstructured information into structured information (th

Simply, no. Most of big advances in the past five years have been around Natural Language 'Parsing' of unstructured information into structured information (this is what Deep Learning is very good at, so far at least); classifying unstructured utterances into structured intents, mapping unstructured utterance 'questions' to unstructured answers in a corpus (fundamentally similar search), literal parsing of utterances to shallow or deep (semantic) parses, and so on. The thoroughly end-to-end, multi-modal and multi-scoped task of participating in a conversation remains largely unsolved. As any non-primitive interaction with Siri or Alexa should indicate. Word of the day is "brittle".