More Human Than Human

Not only did I totally drop the ball on commemorating Woody Guthrie’s centenary, it appears that I also failed to note the 100th birthday of Alan Turing, English mathematician and philosopher, and the man probably most responsible for the modern ability to masturbate to pornography streaming over an iPhone.  Turing was born on June 23, 1912, and we should all think of him when we do that iPhone thing I just mentioned.  Or maybe not.  Looks like he was gay. The Atlantic did a nice series of articles on him if you’d like to read more.

Turing is also credited with developing the Turing Test which tries to determine whether or not machines can think, i.e. is there such a thing as artificial intelligence?  I wrote a bit about artificial intelligence here, and it’s been an interest of mine for awhile, ever since I got into a heated discussion with a socially awkward CS major about the definition of intelligence.  I was a nineteen year old squishy liberal artsy English major with an answer for everything and defined intelligence as “creative thinking,” the ability to make up unique ideas and expressions.  I still think that’s a pretty good definition, but realize its limitations: it would probably exclude my cat, Noodles, who has figured out a way not only to get me to feed her at her direction, but also has me maintaining a litter box in pristine condition for her to poop in.  And that definition would probably include an elephant who has been trained to splash paint onto a canvas with her tail when she’s not busy shitting all over the goddam place.

Turing devised a practical test for determining artificial intelligence that is way more useful  than merely endlessly arguing about the meanings of abstract terms like “creativity” and “intelligence” and “how worthless is an English major anyway?” and “this is the precise reason why you’ll never have sex!”  Turing proposed that a machine can be considered intelligent if it can fool a real live human being into thinking it is also a real live human during a conversation.

Since then, computer folks have been trying to create a program that can carry on a conversation well enough to pass the Turing Test, although they haven’t met with much success.  Here’s a link to Rosette, the 2011 winner of  the Loebner Prize Competition, which awards a nice chunk of money to whoever comes closest to Turing’s idea (other chatbots can be found here, and here).  I have a hard time believing these chatbots would fool anyone (some of the researchers claim to have convinced 30-60% of participants) and while it’s fun to play with them for a bit, the conversation soon becomes nonsensical and boring, which, come to think of it, isn’t so unlike most of my human to human convos.

Advertisements