To recognize forms as “programmable as patterns” is even harder and second-order than just recognizing them as patterns in the first place – and depends on that.
The problem you haven’t grasped is that while all pattern sets can be described as having common elements in common positions/ relationships - the elements *and* relationships change from one pattern set to another and are capable of infinite evolution. At some point, s.o. discovered for example that you could use living creatures as pattern elements, at another that you could use random variations of the sizes of parts (as in cellular automata). There will be ever more discoveries/inventions, and patterns will continue to evolve. Individual pattern sets however cannot evolve – or they cease to be patterns. From: [email protected] Sent: Thursday, August 23, 2012 10:23 PM To: AGI Subject: Re: [agi] Boris Explains His Theory This is where the incompleteness of representations comes in handy. To recognize all of them as patterns, we need only create a program that can recognize that they can be represented as Ben described. And there is no need to do this in advance for every possible pattern. We just create something that looks for patterns, and adds them to the list as we find them. You don't expect a person to instantly see every pattern. It's better at recognizing cetain types right off the bat, while others take longer due to inherent biases in the way the human mind represents things. So why would you make that demand of an AI? -------------------------------------------------------------------------------- On Aug 23, 2012 3:59 PM, Mike Tintner <[email protected]> wrote: Oh jeez, Ben – they are compressible into DIFFERENT programs – a ZILLION different programs. Specialist programs, basically one per pattern. In order to recognize all of them as “patterns” , they would all – all zillion patterns - have to be compressible into just ONE program, just one and the SAME program that could identify/generate every different example of a pattern. Including ones not yet invented. Not a zillion programs, just one. That’s not “easy” Ben, that’s the unsolved problem of AGI. .You sure you’re really interested in it? From: Ben Goertzel Sent: Thursday, August 23, 2012 9:28 PM To: AGI Subject: Re: [agi] Boris Explains His Theory All the examples you mentioned On Thu, Aug 23, 2012 at 3:54 PM, Mike Tintner <[email protected]> wrote: Koch curve, Mandelbrot, herringbone, cellular automato are compressible into short (say, LISP) programs consisting of repeated iterations of simple equations. That's a pretty obvious commonality among all of them... ;p -- Ben Goertzel, PhD http://goertzel.org "My humanity is a constant self-overcoming" -- Friedrich Nietzsche AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription 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
