Re: Kosko on AI in L.A. Times

From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Tue Jul 10 2001 - 01:53:33 MDT


"J. R. Molloy" wrote:
>
> What makes it so difficult is that neither Kosko nor
> anyone else knows how brain architecture produces computation and pattern
> recognition, not to mention higher cognitive functions. (When Anders Sandberg
> figures it out, he can charge a bundle for it.) So, the connection isn't such
> a puzzle; it's translating the data... like shuttling information between
> dissimilar computer platforms without knowing how either of them works, only
> much more complicated.

If you don't know how to speak hexCalvin, link the human to another human
instead of a computer. If the basic language is the same, the two brains
may adapt and figure out how to talk to one another. It still counts as
intelligence enhancement and you can monitor the traffic.

If you need the language key, start looking in the interface between the
limbic system and the cerebral cortex. Emotions are modular,
identifiable, evolutionarily ancient, and preprogrammed; but they need to
bind to abstract thoughts formed from symbols built from learned sensory
data. So you can use the binding as your Rosetta stone to decode the
format of memories.

AI is still faster.

Sincerely,
Eliezer.

-- -- -- -- --
Eliezer S. Yudkowsky http://singinst.org/
Research Fellow, Singularity Institute for Artificial Intelligence



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:08:37 MST