Audience: pl place your questions in the chatbox. I will read them to the speaker.

How far did we come from PARRY or LUNAR or BASEBALL? Idea wise?

In the spirit of Checklist based Explainable AI sanity checks; what other checks do you see for QA systems (linguistic or otherwise)?

Prof. Hovy, I have always wondered whether we have focused on the wrong end of the QA problem. Should we not look at the semantic content of a passage and form a graph and then try to find the possible set of questions that I can answer and then find a reduct set of possible questions and then match NL questions to that.

Dear Prof, can you define/describe the latitude longitude QA problem which you mentioned briefly?

@Prof PB: There are a couple of questions on top

Is there an inherent assumption that there is an answer (somewhere) for the Q asked?

Thank you so much Prof. Hovy for the great talk! :)