Suzette: Loebner Prize winner Chatbot

Advances in artificial intelligence daily, still partly pointing to that challenge several developers dream to meet: the overcoming of the famous Turing Test. The fact of an artificial system can demonstrate a level of intelligence able to mislead a human not only speaks wonders that system, but also the minds that created it.This year the Loebner Prize for artificial intelligence chatbot was called Suzette, coming in first place after attaining fool one of the judges.

Anyone who has tried to chat with a chatbot knows that in one way or another, the conversation acquires certain characteristics that us quickly identify it as a machine or software which seeks to masquerade as a human being. This is the essence of the Turing Test: that an artificial system demonstrably intelligence so that a human being is unable to determine if they really talk to another human being or not. Indeed, systems are still quite far away, something which in part confirms the position of certain intelligence researchers artificial to ignore completely the Turing Test. One of the foundations for this decision is based on the Turing Test evaluates “human behaviour” of a machine, and not “intelligent behavior”. And the difference between both is important.When the “PC Therapist” won the Loebner Prize in 1992, did it thanks to the introduction of errors in his writing.In other words, the machine won a test of intelligence seems to be so clumsy as a human.

“Suzette” had sufficient merit to be the best, but part of responsibility in the deception would have a human factor

In 2010, the Loebner Prize was won by a chatbot known as “Suzette”, developed by Bruce Wilcox, who has been involved in some titles as Army Men and Green Rogue AI programmer. The prize is symbolized by the “modest” sum of three thousand dollars, and Suzette managed to overcome rivals of having misled to one of the judges with their chat. On average, are often used five minutes discussions to determine what as good is a machine imitating a human, but in this Loebner Prize session was extended by a total of 25 minutes, something that requires a level of attention than both by the side of the judge who developed the chatbot-side.

However, it is worth mentioning that in some cases, the chatbot not earns itself only. All in all, the chatbot represents half of the equation, since the judge must have a conversation with a human being.If for some reason that human decides to exhibit a behavior “robotic” of their conversation, could assist the chatbot indirectly to win the prize.In this case, the judges were teachers and students participating.Perhaps only wanted to complicate life teachers on an elaborate hoax, or it may have been other interests.Although it already would enter pure speculation, cannot be ruled out completely.After all, there still are people within the chess circle who believe that Garry Kasparov left to lose against Deep Blue in 1997…