On 08-05-2019 21:55, 'Brent Meeker' via Everything List wrote:
On 5/5/2019 7:06 PM, smitra wrote:
Or perhaps it's a mistake to attribute consciousness to particular
physical systems. Instead of saying that this computer or that brain
is or isn't conscious, we could postulate that there exists (in some
sense) consciousness, and that consciousness can find itself in some
location, and where it finds itself in, is implicitly defined by the
content of the consciousness itself.
If we then consider the set of all conscious states, then each element
of that set contains a subjective description of some conscious
thought which also contains in it where and when this thought is
supposed to have happened and who it belongs to. But that description
of itself won't contain enough information to fully specify what
precise algorithm generated it. My conscious thoughts won't contain
all the information necessary to reconstruct my brain in the exact
state it is now. This means that there are a large number of different
physical systems that are each consistent with being me in the
subjective state I'm in right now.
But I can still pin myself down to being in an environment that from a
macroscopic point of view must look the same as what I'm seeing right
now. There is then a large number of different physical local
environments within which brains are located that can be said to
represent me. I'm all of them, not one particular item. There is a
one-to-one relationship between me getting more localized inside this
set and me changing due to adding more information to my conscious
state.
Systems that are a lot less complex than our brain obviously run much
simpler algorithms than our brains are running. These algorithms with
at best less awareness, would then not be able to localize themselves
as precisely as we can. But since the state space of the computer is
much smaller
State space of the computation or of the computer, the "mind" or the
"brain".
, a question like: "is this AI conscious?" implies a far more precise
localization of the AI's consciousness than in case of us.
Hmm. I was with you up till that. Your earlier said that the AI
being simpler would imply it was LESS localized in physical space,
which I agreed with. Now you seem to say the opposite?
Yes, it's indeed going to be less localized but that would mean that it
won't fit into the device we've set up. So, when we point to our device
and imagine the conscious AI to be in there, it's not actually in that
particular device we're looking at.
This coarse grained view goes a long way to address the problems implied
by thought experiments where one replaces the transistors of the
computer or the neurons of the brain by a devices that perform the exact
same action as is actually happening in the instant the AI is supposed
to feel something but would fail to perform the correct action in a
counterfactual situation. You could then replace the brain by a
recording of brain processes and that recording would then be conscious.
The problem is then with attempting to localize conscious in such a fine
grained picture that's too small to accommodate the algorithm that is
actually running. In a more course grained picture there do exist
counterfactuals on nearby branches that the consciousness itself cannot
resolve. So, in the MWI we should picture ourselves as being located on
bundles of branches, not on single branches.
Saibal
--
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 https://groups.google.com/group/everything-list.
To view this discussion on the web visit
https://groups.google.com/d/msgid/everything-list/6b8119f67c2e73e91815e654ef41fb0e%40zonnet.nl.
For more options, visit https://groups.google.com/d/optout.
