> On 12.04.2014, at 05:22, "Ben Goertzel" <[email protected]> wrote: > > I think it might make sense to separate things so that we have > > -- requests that refer to a specific Atomspace that has already been > instantiated and populated
While I think you may just have encountered one of the reasons we don't have AGI APIs, quite a few of the "selections" and "references" are AGI-complete one way or another, for example in a face-to-face conversation the utterance "that's not it" has to be evaluated in several (all?) contexts before the "correct" interpretation is recovered: is it about what I am doing now, is it about what you are doing, is it about the 2-3 topics we've been discussing. You could perhaps "pretend" that starting this conversation created a top-level context, but I do not see how you can prevent an explosion of nested contexts. One way out is to accept there is only one gigantic context called "me", another would be to limit the contexts to something like sales, billing, tech, but then there goes the G out of the AI. Well anyway, one API is better than no API :) 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
