It will be hard for AI to read the nuances --- As we start gathering more and more data, it is quite possible that AI will actually be able to read these nuances (e.g. implicit communication and body language). Using something called reinforcement learning for language (which is already used in production in sales chatbots, which try to make customers buy products), an algorithm could be trained to be optimized for "diplomatic reward". If the model does in fact learn a policy (a learning method) to maximize this reward, I believe it will automatically learn to exploit these cues that come naturally to us humans. Though there is a still a long way to go with this, I don't think we should rule it out. It may come sooner than we think.

Source author: Mamta Reid liveandsmile@hotmail.com