Guys, Your both right! Well, I think you need both really, but they do compliment and help eachother out, much information is given in text, and to ignore that means at some point in time you will be telling the bot that, over and over, to teach it, and likewise, there are something your just not gonna get nicely out of a book. One simple thing with my system is that a bot can research a Kitchen, and knows what items it expects to find there, like a fridge, and a sink, and a table and a chair, and what % confidence each of those has. So a bot, given the knowledge that you stuck it in a kitchen, could reason about what shoudl or may be in the environment, and look aroudn and try to find a sink, or ask where the fridge is, or conversly, if stuck in an environment that has a fridge and a stove, coudl reason fairly simply that it was in a kitchen... this is a nice little boost to the bot, and gives it information outside what it was given from the environment. Also things like that it can sit in a chair, and things can be on top of a table. A list of possible actions and interactions can be generated and tested.
James Ratcliff Benjamin Goertzel <[EMAIL PROTECTED]> wrote: > A > clearer way of stating how I perceive our approaches as differing would be > to say that I believe that you are learning by discovery while I am > accumulating already discovered knowledge and resolving conflicts. I think > that we are attacking two very different, complementary problems. My hypothesis is that interpretation and conflict-resolution of already-discovered knowledge that is explicitly articulated in text, can be done really effectively only on the basis of a large body of low-level common-sense that has NOT been explicitly articulated in text (but can be gathered via embodied learning) So, my guess is that your approach may well work to build some useful practical domain-specific systems, but will hit a fairly low intelligence ceiling... But I will be pleased if I'm proved wrong ;-) ben g ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?&; _______________________________________ James Ratcliff - http://falazar.com Looking for something... --------------------------------- Ahhh...imagining that irresistible "new car" smell? Check outnew cars at Yahoo! Autos. ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
