Boris,
You portray yourself as a professional GI'er, and suggest that I am a newcomer. Perhaps, but I bring Physics to the table, which is something you need and don't even know. So here is my contribution. The accumulation of entropy, not a lack of theoretical integrity, is the reason for the failure to scale in all AI/ AGI attempts to date. It is necessary to deal with the entropy, or else there will be no AGI. I know how to do it, and that's what the "dumb" permutations do. They are not so dumb, they remove energy from the information, which also causes entropy to be removed and the information to self-organize. The result, in your own words, is a hierarchical pattern discovery process. The permutations start from the beginning, cross-compare sensory input, discover initial patterns (my block systems), which are selectively forwarded to the next level for expanded search. Recursive comparison & elevation of input patterns on successive levels of search will discover increasingly general patterns or concepts. Higher levels generate downward feedback, ultimately a motor action, to focus on additively predictive lower-level sources. The scope and complexity of discoverable patterns grow in a strictly incremental way. You said it, you wanted it, and its all there. By the way, you are using obsolete info, I dropped matrices long ago in favor of causal sets, and I observe self-organization in causal sets. This is the entire AGI machine I propose, there is nothing else. Everything else goes in the data, which is what gives people the false impression that I neglect the data. In the entropy approach, the AGI machine is problem-independent. The machine is always the same, no matter what problem you are dealing with (people are like that). The problem with Alan Grimes is that he couldn't understand that, and he believed (as you do), that the theory ought to be problem-specific (people are not, they learn how to be). Here is an example. Alan said: "you will need to extend your theory to deal with orthogonal sets of orderings." I could have answered (but didn't): There is no need to extend the theory. The machine will deal with them the same way you and everybody else did: by acquiring all the causal relations implied in that notion. You went to school and learned arithmetic and algebra, then went to college and learned high math, etc. One day, you learned about orthogonal orderings. The AGI, too, will have to "go to school" , quite literally. Some day, somebody will convert an entire book of arithmetic into a causal set and feed it to the AGI. The AGI will self-organize the arithmetic, so it can be used for thinking (for self-organizing other things). Just like you did when you were a child. But with machines, this has to be done only once. After that, copies can be made, and all future machines will "know" arithmetic. Then, algebra, etc. You get the point. " He didn't. Do you? And no, you didn't answer any of my questions. And I did explain the data. It just doesn't help. Sergio From: Boris Kazachenko [mailto:[email protected]] Sent: Monday, August 20, 2012 4:43 PM To: AGI Subject: Re: [agi] Uncertainty, causality, entropy, self-organization, and Schroedinger's cat. Sorry Sergio, this is not against you personally, I know you're much nicer guy than I am. It's just that... there is no there there in your theory. "Entropy", "self-organization", "causality", "emergence" are meaningless buzzwords in this context, & I am sure you will pick up more of them. I answered all your questions, even before you asked them, it's just doesn't help. You, on the other hand, refuse to explain what do you actually suggest to do with the data, other than those dumb matrix permutations & scope adjustment? What is your contribution to all that "causality", "emergence", & so on? AGI | <https://www.listbox.com/member/archive/303/=now> Archives <https://www.listbox.com/member/archive/rss/303/18407320-d9907b69> | <https://www.listbox.com/member/?&;> Modify Your Subscription <http://www.listbox.com> AGI | <https://www.listbox.com/member/archive/303/=now> Archives <https://www.listbox.com/member/archive/rss/303/18883996-f0d58d57> | <https://www.listbox.com/member/?&; ad2> Modify Your Subscription <http://www.listbox.com> ------------------------------------------- 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
