[ExI] Is Artificial Life Conscious?

Thu May 5 00:47:26 UTC 2022

On Thu, 5 May 2022 at 09:26, Brent Allsop via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

>
> Hi Stathis,
> On Tue, May 3, 2022 at 4:28 PM Stathis Papaioannou via extropy-chat <
> extropy-chat at lists.extropy.org> wrote:
>
>> On Wed, 4 May 2022 at 07:06, Brent Allsop via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>> Hi Jason,
>>> We continue to talk past each other.  I agree with what you are saying
>>> but...
>>> [image: 3_robots_tiny.png]
>>> First off, you seem to be saying you don't care about the fact that the
>>> first two systems represent the abstract notion of red with different
>>> qualities, and that they achieve their Turing completeness in different
>>> ways.
>>> If that is the case, why are we talking?  I want to know what your
>>> redness knowledge is like, you don't seem to care about anything other than
>>> all these systems can tell you the strawberry is red, and are all turing
>>> complete?
>>>
>>> In addition to turing completeness, what I am interested in is the
>>> efficiency by which computation can be accomplished by different models.
>>> Is the amount of hardware used in one model more than is required in
>>> another?
>>> The reason there are only a few registers in a CPU, is because of the
>>> extreme brute force way you must do computational operations like addition
>>> and comparison when using discrete logic.  It takes far too much hardware
>>> to have any more than a handful of registers, which can be computationally
>>> bound to each other at any one time.  Whereas if knowledge composed of
>>> redness and greenness is a standing wave in neural tissue EM fields, every
>>> last pixel of knowledge can be much more efficiently meaningfully bound to
>>> all the other pixels in a 3D standing wave.  If standing waves require far
>>> less hardware to do the same amount of parallel computational binding, this
>>> is what I'm interested in.  They are both turing complete, one is far more
>>> efficient than the other.
>>>
>>> Similarly, in order to achieve substrate independence, like the 3rd
>>> system in the image,  you need additional dictionaries to tell you whether
>>> redness or greenness or +5 volts, or anything else is representing the
>>> binary 1, or the word 'red'. Virtual machines, capable of running on
>>> different lower level hardware, are less efficient than machines running on
>>> nacked hardware.  This is because they require the additional translation
>>> layer to enable virtual operation on different types of hardware.  The
>>> first two systems representing information directly on qualities does not
>>> require the additional dictionaries required to achieve the substrate
>>> independence as architected in the 3rd system.  So, again, the first two
>>> systems are more efficient, since they require less mapping hardware.
>>>
>>
>>
>> Substrate independence is not something that is “achieved”, it is just
>> the way it works. Hamming is substrate independent because you can make a
>> hammer out of many different things, even though a particular set of
>> materials may be more durable and easier to work with, because it is
>> impossible to separate hammering from the behaviour associated with
>> hammering. Similarly, it is impossible to separate qualia from the
>> behaviour associated with qualia (the abstract properties, as you call
>> them), because otherwise you could make a partial zombie, and you have
>> agreed that is absurd.
>>
>
> I've been reading this over and over again, since you sent it several days
> ago, trying to figure out how you could be thinking about qualia, and the
> "behavior associated with qualia".  Because this must be radically
> different from what I understand qualia to be.
>
> First off, I don't believe I've ever talked about anything close to "the
> abstract properties, as [Brent] calls them".  I always talk about an
> abstract word like "redness" which is meaningless until we define it by
> pointing to a particular physical (or subjective) quality or property.  And
> I've also talked about abstract descriptions of such physical properties,
> like descriptions of how glutamate reacts in a synapse, stressing how
> abstract descriptions of this physical behavior tells us nothing of the
> intrinsic physical qualities that could be being described.  The actual
> physical properties being described are not "abstract properties", they can
> just be described with abstract descriptions, which tells you nothing of
> what they are like.  The fact that you conflate my descriptions of abstract
> descriptions of properties with "abstract properties" whatever those are,
> seems to tell me volumes about how you think about reality, descriptions of
> reality, and knowledge of reality.  I'm probably mistaken, but it seems to
> me that anyone that doesn't clearly distinguish between these things won't
> be able to understand what intrinsic colorness qualities, or knowledge of
> color, and perception (through senses) could be.
>

I apologise if I misapplied your use of the word "abstract". Part of the
difficulty is that we use different terminology. I am referring to the
physical properties, the properties that can be "abstractly" described (if
I am using that term correctly now), the observable properties, the
functional properties. These are contrasted with the qualia, the phenomenal
properties, the private properties, the consciousness. The functionalist
position is that you cannot reproduce the first type of properties without
also reproducing the second type of properties, because if you could, it
would result in absurdity. That is the entire argument summarised in a
sentence.

> Then when you compare hammering behavior with something like an intrinsic
> redness quality of either the strawberry or intrinsic qualities of our
> knowledge of the strawberry (or the behavior of such which can be described
> with abstract descriptions which tells us nothing of the quality of what we
> are describing...?)  When you describe hamering behavior (or dancing or any
> other similar behavior) to a blind person, you will be able to 100%
> communicate to them what you are describing.  but when you describe any
> "behavior of redness" to a blind person, you will fail, completely.  The
> differences between these two things is what is all important, this is what
> you seem to be ignorring.
>

I struggle to find an analogy and this one is imperfect, but just as if you
reproduce the hammering behaviour you necessarily reproduce the hammering,
so if you reproduce the glutamate behaviour you necessarily reproduce any
qualia associated with glutamate. If you can't reproduce the glutamate
behaviour that doesn't mean that qualia are substrate specific any more
than if you can't reproduce the hammering behaviour it means that hammering
is substrate specific.

> What do you think colorness qualities are?  Why can't you describe them to
> blind people?  What is it, in this world, that has all these colorness
> qualities?  How can anyone think that colorness qualities are anything like
> hammering behavior?  If someone was honestly claiming they experience a
> colorness quality you can't experience, and if they called that colorness
> quality grue.  What would that mean, to you?
> And I've asked you this before, and I don't recall your answer.  It's easy
> to understand how someone could come up with a description of some new
> behavior, something different than a hammering behavior.  But how would you
> come up with (or discover) a new color nobody has ever experienced before.
> How would you know what a description of that colorness quality was like?
>

I think colourness qualities are what the human behaviour associated with
distinguishing between colours, describing them, reacting to them
emotionally etc. is seen from inside the system. If you make a physical
change to the system and perfectly reproduce this behaviour, you will also
necessarily perfectly reproduce the colourness qualities.

-- 
Stathis Papaioannou
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20220505/c7bb6594/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3_robots_tiny.png
Type: image/png
Size: 26214 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20220505/c7bb6594/attachment.png>