From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Mon Aug 26 2002 - 09:24:59 MDT
Ben Goertzel wrote:
>> > For instance, as I've argued, there needs to be, in any finite system
>> > within the universe, some non-prob-inference process for adapting the
>> > universal set U used in the system's prob-inference processes.
>>
>>Does it work nonaccidentally? Or is it entirely random? If it works
>>nonaccidentally, some form of rationality is behind it
>
> As I clarified in another e-mail, the crux of the matter seems to be that,
> in order to be effective overall (what you'd call Rational overall), a
> system may need to have some subcomponent that does not, viewed in
> isolation, appear to be operating rationally at all...
Okay. Drag it out and let's have the BPT-detectors give it a sniff. But
bear in mind that I never said *anything* about whether something should
be "viewed in isolation"; as far as I'm concerned "rationality"
(likelihood of a pattern correlating to a given aspect of external
reality) is a property of a total physical process which derives from its
complete causal history. Nor do I think this is an inappropriate usage of
the definition; there is no reason why physical processes and causal
histories would be exempted from rational scrutiny or why a rational
entity could not make use of such information in reasoning.
-- Eliezer S. Yudkowsky http://intelligence.org/ Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT