My point of view is that complexity (complicatedness) is the reason we don't have a lot of primitive AGI programs to work with. I don't have a solution for complexity but I do have some ideas about it. For example, a massive indexing system is necessary.
However, the use of a massive indexing system would introduce additional complexity to the problem, so the indexing system has to be effective. It would have to be a hierarchal system where it goes from initial impression to gradually more detailed selections (based on the input that was being analyzed and previous learning that had been acquired for input that was similar). However, this would require a step by step analysis and comparison (of reasons) so this means that this indexing system would tend to be like a distributed search tree. It could have some of the same problems So what do I do about that? I would design the 'tree'-like indexing to have extensive cross-referencing through cross-categorical relations. While my attempt to deal with complexity seems reasonable, it comes at an expense of introducing more complexity to solve complexity. Well is that like using fire to fight fire? Maybe - but it is not a sure thing. But I could try my indexing theory with some partial simulations. If those simulations produced some interesting results I could then come back and say - hey, it will work because my simulations seem to say that that part of the problem can be resolved. Jim Bromer ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
