If you did not see IBM’s new “robot” beat the two best human players in a televised Jeopardy (Ken Jennings and Brad Rutter) two weeks ago, I highly recommend watching it

Whole watching Watson win I was reminded of the other super know-it-all robot called Google, but whereas Google acts a bit like a trained dog who responds to “fetch”, Watson feels so much more human because he fetches and then speaks aloud the best response. So I got to thinking about how well Watson would work as a tutor for humans. There are all sorts of interesting challenges here. Among these are:

First, domain novices usually do not possess domain-specific vocabulary which experts use. So Watson would have to interpret the novice questions and form a best guess response. This is straightforward if the question has a known, factual answer, but could easily be more difficult if the student presented a “what-if” scenario.

Second, tutors often have to do more than merely verbalize a response to a question. For example, they may have to draw pictures, tables, or diagrams. This is complicated because Watson not only has to figure out the question’s intent but may actually have to recall parts of relevant pictures it has mined from the web and then use pieces to make a new drawing.

Third, good human tutors form rich models of the tutee’s knowledge and understanding and constantly ask the tutee questions to update the model. If you’ve ever been lucky enough to be a tutor or get tutored, you know this this is crucial because it allows the tutor to interactively eliminate misapprehensions by constructing relevant examples. This is a very creative process, perhaps it should rightfully be called an art, and to get good at it, Watson would require lots of data from past tutoring sessions with other students. But if enough children use Watson, this might be possible.

All these challenges seem hard to me, and I am reminded of a quotation attributed to Jon Von Neumann who invented the basic architecture of modern computers over 70 years ago. He said, “You insist that there is something that a machine can’t do. If you will tell me precisely what it is that a machine cannot do, then I can always make a machine which will do just that.”

NO COMMENTS
Post a comment

*