I agree that even a simple embodiment will be a great benefit to a growing AGI.
As for the more closely modeled human AI with full "emotion and biofeedback 
sensors" (my words) I dont think that is really feasible at this time, and not 
as necessary, though the AGI will still need to internally model ALL of these 
human functions, in order to truly understand us, and interact at a very high 
level.

I think the main thing that an embodied AGI (be it robotic or virtual, or 
textual world) is the ability to increase the ways and speed of learning.
The embodied AGI has a level of freedom and choice that standard AI's really 
dont have.
It can choose what it wants to learn, move about in the world, encounter new 
people, environments, and situations.  The wider the range of environments, the 
more it has the ability to learn.

One main goal of an AGI is to allow it to learn... without an embodied AGI you 
are very limited to the amount of learning that it can do, it can do basic 
interactions with people, and lookup information in text db's.

So we need to find the minimum level of base information that we can start it 
with, along with a very flexible framework that will allow it to learn easily 
in all those environments, which the novamente agent is heading towards the 
ability to do.

I had thought also as a shorter dev methed, for creating a MUD agent that would 
have a larger level of freedom to move about in the text world and learn, a 
long with a large text corpora for creating object frames.  This would give a 
basic agent the ability to pick up a object in the MUD world, say a book, and 
reference its internal knowledge base, that says books can be opened.... and 
try to do that in the game world... if it cant, the action can be added to the 
game world easily, allowing it to grow with the AGI in abilities.

James
_______________________________________

James Ratcliff - http://falazar.com

Looking for something...

--- On Fri, 8/8/08, Ben Goertzel <[EMAIL PROTECTED]> wrote:
From: Ben Goertzel <[EMAIL PROTECTED]>
Subject: Re: [agi] Human experience
To: [email protected]
Date: Friday, August 8, 2008, 8:31 AM






-- I don't see what benefit embodiment brings to the creation

   of an agi scientist/engineer, whereas reading is critical.

    Mechanical awareness -- not so much -- AGI could have

    "immediate" mechanical awareness of not just 3D, but also

    4D, 5D, etc. spaces.
I feel that simple embodiment is the most likely route to enable
an AGI to learn human language well enough to read complex
texts, including those in scientific and engineering domains.


This is really the main point.  I don't think that statistical corpus
analysis nor hand-engineering of linguistic rule-bases is going to
make systems that can really understand language.  But I think that

systematic instruction of a simply-embodied AGI system can get us
there, even without all the bells and whistles of a truly humanlike
body.

 




-- I see that simple embodiment is not anywhere near enough

   to put human social contact into the reach of direct experience.

   Embodiment will help AGI understand "chair" and "table";

   it will not help it understand vindictiveness, slander.


True

ben  g







  
    
      
      agi | Archives

 | Modify
 Your Subscription


      
    
  





      


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