I have a hard time understanding the purpose of Meena, other than Google saying "Look, how good we are by scraping all the social media conversations available with our large scraping engine called Googlebot and feeding them into our giant ai computer system". This bot will be able to do small-talk but not any specific authoritative question/answer work.

I have been looking into chatbots half a year ago to add to one of my projects but didn't implement it because of the huge training you need to do before you get any sensible results that are actually useful for the user.

Yes, it did real like a smug-sounding "how good are we" message. How3ever, Chatbots are here, whether we like them or not.

Besides the technology issues of machine learning, I believe there are going to be potential legal angles because the "answers" could be misleading, or even entirely incorrect. If the wrong answers causes someone to have a financial loss, or worse, who would be liable?
Headine, "Chatbot in the dock over alleged incorrect advice."

The responsibility for wrong decisions by AI is still an open question. That issue will only be cleared if events with a major impact on human beings happen and the courts have to weigh in. My own prediction is that the first significant AI-related court cases will be related to car accidents with casualties due to self-driving cars' decisions. I don't expect chatbots to be in the frontline of court cases, but their behavior may be adjusted after the first decisions in other AI-related cases have been made.