I really enjoy Allison Parrish‘s Deep question bot. It’s a bot that asks questions based off of ConceptNet, which contains information about the world that computers should know:

What if there were a tricycle… that did not have a pedal?

— Deep Question Bot (@deepquestionbot) January 11, 2016

Have you ever considered a typewriter that is young instead of old?

— Deep Question Bot (@deepquestionbot) January 4, 2016

Why are mugs considered to be dishes? Why are humans found in universes?

— Deep Question Bot (@deepquestionbot) January 18, 2016

I came across this bot a while ago, and was happy to find it again through Allison Parrish’s twitter.  I love it because of several reasons.  One, it helps people understand how ConceptNet can be interpreted by robots.  It reminds me of a child learning about the world (with good grammar), and asking questions that almost make sense, but not quite.  When questions come up that do make sense, they often make me laugh, because those that do make sense tend to be related to more simple things (i.e. tricycles and mugs).  It also brings to question how convoluted and complicated the english language is.  At times though, the questions make me imagine different ways the world could be.  Maybe this bot could provide someone with a source of inspiration!  I also like thinking of this robot from a more objective point of view.  It asks questions without truly knowing the relations of many things in the world, and yet it can ask these questions.  The human reader projects a lot of information onto these questions so that they make sense.  I wonder how much of the bots personality is created by the reader rather than the questions asked.  The way the questions are asked are often so humanlike in quality because of the added ellipses, “right”, and other well placed colloquial terms, make the bot extremely relatable.