On Thu, Apr 4, 2013 at 1:49 AM, Piaget Modeler <[email protected]>wrote:
> If you use standard database technologies to create indexes into the flat > file of facts then > you can navigate the file randomly and efficiently. > > I propose that you choose a format for your facts, then add facts to your > "database" however you want. > The important thing is that you have a representation that is accessible > and searchable. > We are still in the chapter where Steve asked for understanding without wishy-washiness, and PM offered "you compile language into a "database of facts" using a propositional representation. In addition, you convert all sensory input to the AGI into the same propositional representation." - right? Isn't the infinity of language a problem here? Analog-to-digital conversion and input granularity is obviously an issue too. How about the deeper structure of input, the possibility of two wires you see to be loaded or not? Isn't the "any" format for the facts wishy-washy? Though I doubt any hard-core computer scientist would think of his "mental" representation as a (SQL or NoSQL) database, it is not hard to imagine building the kinds of indexes and derivative "tables" that would make even generic SQL look like all and any convenient representation. But given that we have a few proposed solutions to these problems, including CYC, maybe an example or two is in order. My take on understanding is probably wishy-washy because I don't think understanding is a strong concept epistemologically. I mean, and it happens in real life too, understanding is an iterative exploration that is not even monotonous, you can spend some time interpreting a sentence or a paragraph (always in its presumed context) and just when you thought you were about to close the deal and declare the sentence 100% clear and understood you face a curve ball and go back to square 1. If we are not going to use embodied entities and that kind of reality check, eventually rejecting agents that fail at tasks therefore de facto lacking understanding, we may need to use very pedantic "understanding" datasets with utterances and their definitive meaning. How is "one plus one equals two" to be understood? "One plus one equals 0.2"? I am tempted to reiterate some well known opinions from philosophy and linguistics, that meaning and therefore understanding is random, ad hoc, meaningless and not at all understandable! It does look like outside very tight logical domains like genealogy "Mary has a daughter Jane and Jane has a son John - OK, I get it, Mary is John's granny", outside those domains understanding is kind of meaningless - as a rule on this list too :) So, let's not forget that communication is, first of all, an attempt to take over a foreign mind-body, a mild form of demonic possession, understanding possibly an excuse explaining away why we lost control! AT ------------------------------------------- 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
