[ExI] The digital nature of brains

[Ben Zaiboc]

Spencer Campbell <lacertilian at gmail.com> wrote:

> >From my perspective, Gordon has been very consistent
> when it comes to
> what will and will not pass the Turing test. His arguments,
> implicitly
> or explicitly, state that the Turing test does not measure
> consciousness. This is one point on which he and I agree.


The Turing test was designed to answer the question "can machines think?".

It doesn't measure consciousness directly (we don't know of anything that can), but it does measure something which can only be the product of consciousness: The ability of a system to convince a human that it is itself human.  This is equivalent to convincing them that is is conscious.

If this wasn't the case, people would have no real reason to believe that other people were conscious.  

For this reason, I'd say that anything which can convincingly pass the Turing test should be regarded as conscious.  

Obviously, you'd want to take this seriously, and not be satisfied with a five-minute conversation.  It'd have to be over a period of time, involving many different domains of knowledge before you'd be fully convinced, but if and when you were convinced that you were actually talking to a human, You'd have to admit that either you think you were talking to a conscious being, or that you think other humans aren't conscious.


Ben Zaiboc