Actually my ideal system is one that uses induction, abduction, and truth maintenance. It allows the AGI to start as a tabula rasa baby, and then learn language incrementally. So, web-surfing lay people can teach it. The only problem is that the algorithms involved are very complex...
OK, well now that sounds like Novamente, except 1) NM has the capability to utilize human-coded rules as a guide for its learning 2) We have chosen to have folks talk to NM in the context of a mutually perceived sim world rather than in a "vacuum" However, as you note, the algorithms (and data structures, and required underlying software architectre) are complex, so getting the whole thing to work takes a lot of time and effort.... We are starting w/ learning based on perception and action and want to do language learning after that, whereas you want to start w/ language learning. But in fact one **could** start with language learning w/in the NM architecture, if one wanted to. I just think that isn't the optimal way to go... -- 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/?member_id=231415&user_secret=fabd7936
