Saturday Musings: Watson, Jeapordy and Singularity

Saturday Musings: Watson, Jeapordy and Singularity

>

I have been trying to get back in the habit of doing these  Saturday Musings posts where I take some time to think of an issue that’s a little bit bigger than (though often related to) the worlds of IT, startups and Pittsburgh.  I missed it yesterday because I was busy at a wedding, but I’m going to make it up today and, as a bonus, you’ll still be getting your weekly dose of Kraut!

Today’s big issue is artificial intelligence and the total lack of a threat it is to human superiority.  The issue comes up this week because of the Jeapordy! challenge this week.  For those of you who didn’t see it, an IBM super computer named Watson took on Ken Jennings and another past champion in an exhibition game of Jeapordy.  Watson not only won, he destroyed his opponents, earning $77,147 while his opponents earned a combined $45,600.

As a technologist I have faced the same question from my less technical friends and family all week, “Does this mean that computers are smarter then humans now?”  Some in the technology press/blogosphere have even asked that question from their “pulpits.”  The one that stood out to me was Alexis Tsotsis’ reference to a possible advancement past Vernor Vinge’s singularity.  Suggesting that Watson may have surpassed even its programmers’ expectations by playing a joke on Alex Trebek (that didn’t happen).  You can read her post here for the conspiracy details, but I wouldn’t recommend it.  The truth is Watson is not intelligent and any suggesting by an expert that he is, is pandering at best and irresponsible worst.

Watson is nothing more then a giant word association machine.  All he does is sift through documents written by humans to correlate one word to another word.  In short, when he hears a clue about a city with airports named after world war II battles and generals.  It starts looking for how often a city’s name appears in articles that mention “general”, “battle”, “world war II” and “airports”.  It does this with ridiculous efficient speed and efficiency (actually finding an answer before human counterparts can search their own brain), but it does it with less intelligence then a chimp using a stick to catch ants in an ant hill.  You see a chimp putting a stick in an ant hill isn’t that impressive of a use of technology, but it IS an original idea to solve a need of the chimp.  While it is amazing that Watson can do what he does, he is only doing what his programmers told him to, he has no idea why or for what purpose.