Tuesday, June 10, 2014

Computer Fools Interviewers Into Thinking It's Human


How many of you remember this early scene from Blade Runner: The sloppy looking guy is sitting across the table from a man in a suit, who proceeds to prop up some strange monitoring equipment. He begins to administer the “Voight-Kampff” test, which involves asking several off-putting questions designed to evoke an emotional response. Unbeknownst to the innocuous schlub across the table, the test is actually designed to determine whether or not the test taker is an android. It doesn't end well for the interviewer.

Recently a Russian research team attempted to create a program that could pass the real life version of the Voight-Kampff, known as the Turing Test. Developed by Alan Turing in 1950, the test is designed to gauge the intelligence of a computer. It involves a judge who does a kind of IM chat with two possible subjects who he can't physically see. One of them is human and one of them is a computer.

The judges must ask several questions and determine whether or not they think they're talking to a computer. Much like the interview from Blade Runner, the judges aren't really looking for right or wrong answers, rather they are trying to figure out how “human” the answers are. After being quizzed by a number of people, the computer supposedly passes the test if it can fool 30 percent of the judges into thinking it's a person.

In this case, the computer was a chatbot named Eugene Goostman, and was designed to fool the interviewer into thinking that it's a 13 year old boy from Ukraine. Eugene managed to convince 33 percent of the judges that it was a real live boy. Professor Ken Warwick is confident to call this the first time the Turing has been passed:
“Some will claim that the Test has already been passed. The words Turing Test have been applied to similar competitions around the world. However this event involved the most simultaneous comparison tests than ever before, was independently verified and, crucially, the conversations were unrestricted. A true Turing Test does not set the questions or topics prior to the conversations. We are therefore proud to declare that Alan Turing's Test was passed for the first time on Saturday.”
There appears to be a lot of skepticism though, mainly with the choice of using a 13 year old boy who speaks English as a second language. Both of those factors could possibly skew the results, because any answer the bot gives that happens to sound strange or inhuman, could be excused for his lack of language skills. At the very least, it probably wouldn't fool Deckard.

Despite this it appears to be a pretty impressive, albeit frightening technological development. Is this another sign post on the road to a future filled with androids? Or is this just an example of a cleverly programmed machine? Perhaps you can you can decide by asking Eugene a few questions of your own. Personally I didn't think Eugene was all that compelling. Of course I had the hindsight of knowing he was a bot, so the first thing I tried to do was “break” him by asking some oddball questions.

However, we may be rapidly approaching a bizarre future filled with machines that are, at least on the surface, indistinguishable from humans. Our daily lives will probably lack the benefit of hindsight. Until then, I think I'll just try to enjoy the somewhat less bizarre present.

Delivered By Daily Sheeple

No comments:

Post a Comment