Hi all,
I have spent some time recently mulling over the details of a
partially-new language for communicating between humans and AI's. The
language is (tentatively) called Lojban++ and is described here:
http://www.goertzel.org/papers/lojbanplusplus.pdf
Of course, I don't think that a language like this solves the
fundamental problems of AGI design/dynamics/teaching. By no means.
However, I think it can be a valuable tool, in terms of making the
teaching process easier and smoother. Humans come with a lot of
inbuilt inductive bias that helps us learn natural languages. AGI's
don't have this particular inductive bias, unless one explicitly
builds it in, which is very hard. Thus it makes sense, for AGI
teaching purposes, to use a language that can be mastered without any
particular inductive bias, because it's closer to the "thought" level
without so much arbitrariness. Lojban++, a pidgin combining some
English vocabulary with the already existing logic-based language
Lojban, seems to me to fit the bill...
Building a Lojban++ parser and semantic-mapper will require a fair bit
of work, and if anyone on this list is interested in taking on this
project (on an open-source basis, most likely) I'd be eager to talk
about it...
-- Ben G
-------
To unsubscribe, change your address, or temporarily deactivate your subscription,
please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
