A software created by a British computer genius has passed one of the key tests of artificial intelligence - a machine that could 'think' and fool human beings into thinking that they are talking to a person, not a machine.
At the Techniche computer festival in Guwahati in India this week, software called Cleverbot succeeded in fooling 59.3 per cent of 1,334 voters that it was a human being - far more than had participated in any previous official 'Turing test'.
The audience watched as 30 volunteers conducted a four-minute typed conversation with an online chatter who could either be a Cleverbot, or a real person behind a screen - the conversations were displayed on a big screen for the audience to vote on.
Cleverbot did succeed in fooling more than half the audience that it was human - but the human volunteers were still more convincingly 'human'.
Voters thought that 63.3 of the human-to-human conversations were with a real person. Cleverbot might have fooled more people than any previous test.
"I suppose Cleverbot has passed a 'chatty' variation on the Turing test," the software's British creator, Rollo Carpenter told the Daily Mail.
"It may not have been indistinguishable from a human being - but 59 per cent and 63 per cent are very, very close," he added.