On 5/15/2015 11:14 PM, Bruce Kellett wrote:
meekerdb wrote:
On 5/15/2015 10:29 PM, Bruce Kellett wrote:

The AI that I envisage will probably be based on a learning program of some sort, that will have to learn in much the same way as an infant human learns. I doubt that we will ever be able to create an AI that is essentially an intelligent adult human when it is first turned on.

I agree with that, but once an AI is realized it will be possible to copy it. And if it's digital it will be possible to implement it using different hardware. If it's not digital, it will (in principle) be able to implement it arbitrarily closely with a digital device. And we will have the same question - what is that makes that hardware device conscious? I don't see any plausible answer except "Running the program it instantiates."

But that does not imply that consciousness is itself a computation.

I didn't draw that conclusion. That's the conclusion Bruno wants to draw - or close to it (he talks about consciousness supervening on an infinite number of computational threads).

There is not some subroutine in your AI the is labelled "this subroutine computes consciousness". Consciousness is a function of the whole functioning system, not of some particular feature.

Right.  And the system includes the environment that the consciousness refers 
to.

That is why I think identifying consciousness with computation is in fact adding some additional magic to the machine.

I don't see that it's adding anything, magic or otherwise. As you say, any process can regarded as a computation.


Consciousness arose in nature by a process of natural evolution. Proto-consciousness gave some evolutionary advantage, so it grew and developed.

Yes, it seems like a natural extension of modeling one's surroundings as part of decision making, to add yourself into the model in order to imagine yourself making different choices.

Nature did not at some point add the fact that it was a computation, and then it suddenly become conscious. Consciousness is a computation only in the trivial sense that any physical process can be regarded as a computation, or mapping taking some input to some output. There is not some special, magical class of computations that are unique to consciousness. Consciousness is an evolved bulk property, not just one specific feature of that bulk.

But computation is also not just one specific feature of a process, it's a wholistic concept. So I disagree that there is not some special class of programs that implement consciousness; specifically those that model the device as part of it's own deicision processes.

Brent


Bruce


--
You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.

Reply via email to