Friday, October 17, 2008

Machines Fool 25% Of Human Investigators!


















(A.C.E conversing with human interrogator)


In 1950, the brilliant mathematician Alan Turing, devised the Turing test. He postulated that, "If, during text-based conversation, a machine is indistinguishable from a human, then it could be said to be 'thinking' and, therefore, could be attributed with intelligence" (note the phraseology: 'attributed with intelligence').
Since 1991 teams of researchers have been annually competing for a $100,000 Loebner prize, awarded to the first team to pass the Turing test.
This week in Reading, England, two teams of researchers presented A.C.Es (artificial conversational entities) that fooled the human investigators 25% of the time. That means that if the ACE conversed with 4 people, one of those people, based on the nature of the conversation, thought that they were conversing with a human – a huge acheivment considering the complexity and unpredictability of human conversation.
As teams of researchers edge closer to capturing that Loebner prize and passing the Turing test, web-based avatars like those created on lifenaut.com, will become more and more advanced conversationally – there will come a point where your friends won't be able to tell whether or not they are chatting with you or your avatar. But the passing the Turing test won't just be a nifty accomplishment that makes your lifenaut avatar more life-like, it will represent a huge step toward creating machines that possess intelligence. Indeed, that's why Alan Turing proposed the test in the first place.

1 comment:

  1. Check out this Web 2.0 approach to chatbots: http://chatbotgame.com.

    Just as Deep Blue brute-forced it in chess with speed, the idea behind the Chatbot Game is to brute-force it with a huge number of user-submitted Google-like chat rules.

    ReplyDelete