Back on the problem of "understanding"

more below

_______________________________________

James Ratcliff - http://falazar.com

Looking for something...

--- On Wed, 8/6/08, Terren Suydam <[EMAIL PROTECTED]> wrote:
From: Terren Suydam <[EMAIL PROTECTED]>
Subject: Re: [agi] Groundless reasoning --> Chinese Room
To: [email protected]
Date: Wednesday, August 6, 2008, 1:50 PM

Abram,

I think a simulated, grounded, embodied approach is the one exception to the
otherwise correct Chinese Room (CR) argument. It is the keyhole through which we
must pass to achieve strong AI.

The Novamente example I gave may qualify as such an exception (although the
hybrid nature of grounded and ungrounded knowledge used in the design is a
question mark for me), and does not invalidate the arguments against ungrounded
approaches.

The CR argument works for ungrounded approaches, because without grounding, the
symbols to be manipulated have no meaning, except within an external context
that is totally independent of and inaccessible to the processing engine. 
--> Meaning and understanding here I dont believe are just a true false value.
In this instance the Agent WOULD have some level of meaning known, if given a 
database of facts
about cats it would be able to answer some questions about cats, and woudl 
understand cats to a certain extent.

I believe for this to be further constructive, you have to show either 1) how
an ungrounded symbolic approach does not apply to the CR argument, or 2) why,
specifically, the argument fails to show that ungrounded approaches cannot
achieve comprehension.

Unfortunately, I have to take a break from the list (why are people
cheering??).  I will answer any further posts addressed to me in due time, but I
have other commitments for the time being.

Terren

---------------------------------------------------------

--James Reply
1. Given that a Chinese Room VS an AI in a box, the agent replying to the 
chinese questions
has no "understanding" of chinese.  To all extents and purposes it is replying 
in a 
coherent way to all questions, and by the Turing test is unable to be different 
acting than a
human.  That meets my burden of being an AGI, if it replies always in 
reasonable manner.
Whether it understands anything or not seems to be a totally different question.

2. Understanding, using any of the definitions, seems to be judgeable on a 
scale, 
emphasis on judgeable, in that there is no measure of understanding that can be 
done in a vacuum.
So to say, does the AGI understand is nonsensical without that context.
In school, we determine understanding by testing, and asking questions, and 
performing tasks.
So an AGI it would seem would need to be handled in a similar fashion.
A un-grounded AGI without a body when quizzed about certain items would show a 
certain level 
of understanding depending on the depth and correctness of its knowledgebases 
and routines.
Is it truly "understanding" the concept any further than reading it, and 
answering the question?
A grounded AGI may perform better because it is able to interact and gather 
more and better details about 
the topics.
But in the end the grounded AGI simply has a larger lookup database of 
experiences it can use.
When handed a question on a sheet of paper, it looks it up in the larger DB.
A embodied robot AGI would have the added ability of interacting physically 
with the objects, 
therefor when handed a cup, it could look-up what to do with it, 
and "understand" that it could fill it up with a liquid, and follow a plan for 
that.
In this sense it would be able to "prove" to an outsider that it understood 
what a cup was.

Please Answer: Now how can we really say how this is different from human 
understanding?

I receive a question, I rack my brain for stored facts, if relevant, and 
any experiences I have had if relevant, and respond, either with words or an 
action.





      


-------------------------------------------
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=8660244&id_secret=108809214-a0d121
Powered by Listbox: http://www.listbox.com

Reply via email to