Blind Man's Bluff and the Turing Test

Posted 6 Apr 2004 at 01:07 UTC by steve

A recently released paper by Andrew Clifton proposes that the Turing test does not provide a valid criterion for the presence of consciousness. Imagine a "Turing Test" in which the interrogators must be convinced that the participant is a normally sighted individual. A blind person might pass by successfully lying about the visual sensations experienced by sighted persons. According to Clifton, this means an intelligent enough computer could pass a Turing test by lying about being conscious. He then goes on to attempt to define consciousness and propose a test for it that he calls the "Introspection Game".

The paper seems to argue that..., posted 7 Apr 2004 at 01:22 UTC by sigfpe » (Journeyer)

...in order to tell whether or not someone or something is sentient you need to examine its hardware and/or source code. So for legal reasons, from now on, I'm just going to have to assume that everyone else is non-sentient.

So they are zombies, posted 7 Apr 2004 at 18:04 UTC by outsider » (Journeyer)

What you suggest by 'non-sentient' persons, is Dave Bennet's theory of zombies. In fact, you cannot distinguish a zombie from a sentient person...

I meant David Chalmers, posted 7 Apr 2004 at 18:06 UTC by outsider » (Journeyer)

I said Dave Bennet when it must be David Chalmers.

That's the point!, posted 8 Apr 2004 at 00:26 UTC by sigfpe » (Journeyer)

you cannot distinguish a zombie from a sentient person...

I think the point of the paper is that you are exactly right when looking from the outside, but you can tell the difference if you get to peek at the source code.