Sunday, April 22, 2012

Cleverbot



Alan Turing, the man responsible for cracking the German Enigma code also created a test for machines that could think and act like humans. Recently in Guwahati, India at the Techniche Computer Festival, Cleverbot successfully fooled 59.3% of 1,334 voters thinking it was human. The amount of participants was far greater than any other previous Turing Test. The creator of this robot is Rollo Carpenter. He said that, “You could argue that by fooling more than 50 per cent of the people, Cleverbot has passed a Turing test. But there is no universal test.”

The way the test was conducted was that the robot was tested alongside humans, for 'control' purposes. Thirty volunteers directed a short typed conversations with an online chatter. This chatter could either be a real person or Cleverbot. These conversations were then projected onto a big screen for the audience to vote wether it was the robot or a human. It's crazy to think that a machine is capable of imitating humans so closely and what a couple years of technology will bring. And as silly as it sounds it seems quite possible that machines could one day supersede humans.

Even though Cleverbot passed the test, it is still an endless debate. Carpenter also stated that “ You could argue that it failed because to truly 'pass', it should be indistinguishable from a human, and it wasn't. Even in 1997, Jabberwacky, a robot succeeded in fooling more than half of the audience but the human chatter were considerably more human. 

1 comment:

Hye-Jin said...

I think the significance of the result of the Cleverbot test is that the difference between the percentage of people who thought Cleverbot was a human being and the real human-to-human conversation was with a real person is very small; it is only 4 percent difference. To be honest, fooling 59.3 percent of people is not very impressive, but the fact that those people even thought a real human being was a machine tells that the Cleverbot was not very different from real person. It would have been interesting if they had voted between the Cleverbot and the human being for who would be a real person, because that would show more effectively how many people were fooled by the Cleverbot.