Our prediction has been made in print: "Unlike many, we believe that [the Turing Test] will be regularly passed within 10 years of today, by 2015, and that contextual learning techniques are those most likely to succeed. " (from Computing Machinery and the Individual: the Personal Turing Test).
A entirely different prediction we made, also in the paper, was "We believe that with continuing development, and with a doubling of its database to 10 million entries, Jabberwacky will pass as human most of the time for most people." We are referring to people who are simply chatting online, in a natural way - which is very different to the rigourous questions and analyses likely to be imposed during a formal Turing Test.
10 million is likely to be reached in 2006, though at the time of writing we hadn't considered the impact of growing popularity in languages other than English.
|
|
|
