No, Jim, you/an AGI need to be able to recognize if something is a “pattern”  
just as you need to recognize a “tree” or a “box”  or a “chair.” It’s 
essentially the same problem – I picked on “pattern” merely because so many 
AGI-ers think concepts are patterned forms which idfentify patterned objects  - 
 (and patterns, taken as a whole, are so clearly not patterned)..  

A child may see only **a couple** of trees, or boxes – and from that is able to 
recognize further examples .

There is no way this could be an operation involving complexity – and remember 
only humans and animals are able to conceptualise/object recognize. The living 
AGI may only have seen a few features as its first example[s] of a concept, and 
yet can work/conceptualise quite happily and successfully from these.

The mistake you’re making is akin to thinking -   “well, computers play chess 
by performing complex operations – so that’s the way it must be played.”

Well, humans obviously don’t use complex operations to play chess – that is an 
absurd suggestion -  chessplayers play by considering visual configurations of 
pieces – and the typical grandmaster looks at maybe only a hundred possible 
configurations to make a move.

If you want an AGI gamesplayer, – a player that can learn and play *any* board 
game – then it will have to work the human way – and that has nothing to do 
with complexity. Narow AI and only narrow AI – which is basically about 
variations on a known and limited set of factors - has a great deal to do with 
complexity. In AGI/conceptualisation there is no known set of factors. 



From: Jim Bromer 
Sent: Sunday, August 26, 2012 6:56 PM
To: AGI 
Subject: Re: [agi] Pattern: definition & incremental syntax.. P.S.

It is pretty easy to come up with a way for programs to recognize certain kinds 
of patterns in certain kinds of situations.  A computer program could be 
written to abstract or find abstractions from a number of data storage types.  
Technically it should be feasible to write a program that could detect a given 
pattern, if it had enough time, as long as the pattern was not too obscure.

The problem with a challenge like this is that it is not really the problem. In 
AGI, not only does a computer program need to be able to recognize patterns, 
but it needs to be able to find the important patterns that would allow it to 
leverage the knowledge that it already had to achieve stronger goals.  It is 
very easy for a program to find some abstractions out of a source of data, but 
it is impossible for a program to find every possible abstraction (if the 
source of the data was large enough - and it would not have to be that large).  
The problem is that the complexity of finding every kind of possible pattern in 
some data is just too great.  There are too many possibilities.

That is one example of how the contemporary AGI problem is a complexity issue.

Jim Bromer


On Sun, Aug 26, 2012 at 1:35 PM, Mike Tintner <[email protected]> wrote: 
  “it's pretty easy to come up with ways to do it in a program.You can't see 
how a pattern is a patterned concept because you don't understand classes, 
subclasses, and instances.”
  Go ahead – give us a hierarchy of classes for “pattern”, & we’ll present your 
program with a pattern and non-pattern or two for recognition. (I think you’re 
totally lost here – we’re talking about what is basically visual/sensory object 
recognition. You/your machine have to be able to recognize a “pattern.” You 
seem to be talking about, basically, database operations).
             


      AGI | Archives  | Modify Your Subscription   



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com

Reply via email to