An intriguing
article in the Guardian today about an upcoming programs, based on Alan Turing's conversational questioning tests. There are six new super-computers, all with equally incongruous names like
Alice and
Eugene, will be trying to answer a series of questions, posed by a human interrogator. These will be compared with the answers by a live a human, and the interrogator will have to guess which is which. If a computer fools the interrogator, then the programmer wins $100,000 and a medal.
Seems pointless, really, but it does have some interesting consequences, as the author points out. For one thing, if a computer can fool a human, and continue to do so beyond the testing phase, does this mean it can be truly classified as conscious? The problem then facing the participants, beyond that they are intractable geeks, is whether it is reasonable to turn the computer off. The logical extension of this argument being: if an entity was universally acknowledged as being sentient, yet had an off switch, would it ever be reasonable to use it? Oooh, good opportunity to go watch the entire Star Trek: The Next Generation series.