From: Mitchell Porter (mitchtemporarily@hotmail.com)
Date: Mon Feb 05 2007 - 18:07:10 MST
If you the programmer ('you' being an AI, I assume) already have the
concept of probability, and you can prove that a possible program will
estimate probabilities more accurately than you do, you should be able
to prove that it would provide an increase in utility, to a degree
depending on the superiority of its estimates and the structure of
your utility function. (A trivial observation, but that's usually where
you have to start.)
If you the programmer have to *invent* the notion of probabilities -
that's a little harder. I can imagine that a predictor which uses the
shortest-program principle to model its environment might discover
induction - the 'model' would then be 'use the model you get by
applying these heuristics of induction to the saved data'.
_________________________________________________________________
Advertisement: Meet Sexy Singles Today @ Lavalife - Click here
http://a.ninemsn.com.au/b.aspx?URL=http%3A%2F%2Flavalife9%2Eninemsn%2Ecom%2Eau%2Fclickthru%2Fclickthru%2Eact%3Fid%3Dninemsn%26context%3Dan99%26locale%3Den%5FAU%26a%3D23769&_t=754951090&_r=endtext_lavalife_dec_meet&_m=EXTเข
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT