That would actually be enough to start with. Even without
   knowing the categories - and I don't need to know the categories
   - an AI probably could come up with not you per se but would
   come up with something that you would recognize as something
   understandable to you but that got it wrong, unless it happens
   to generate the appropriate categories. The understandable is
   the part that's intriguing to me. For example, if you had an AI
   use each of the six movies and instructed it: "Keep each of the
   six in a separate category. Generate a distinct category for
   each that minimally overlaps with any of the other five" or
   "Keep each of the six in a separate category. Generate a
   distinct category for each that maximumly overlaps with all of
   the other five" [I think there's several other ways to go about
   it as well] You would have an AI that you understand because
   it's constructed of ideas from movies you will both have in
   common, yet the lessons it draws FROM those movies may be
   different. I wonder what it would be like talking with someone
   who knew everything you knew to the depths that you knew but
   interpreted it differently, making different connections, taking
   different paths. == I think you might end up with someone you
   could continually talk to, as each tries to convince the other
   to see their point of view, and over a period of time, you'd
   come up with mutual understandings, even if retaining their own
   distinct worldviews. Oh, I'm fascinated with this idea. I also
   need coffee but I love a good "what if". ==