From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Thu Aug 09 2001 - 16:54:57 MDT
Samantha Atkins wrote:
>
> I've known people with a whole string of nines in their
> percentile IQ rating. Some were very effective and involved and
> full of vision and some were as ineffectual, confused and lost
> as any of those with a less lofty score. IQ simply doesn't
> directly correlate with being effective or fully engaged in
> life.
Then either the IQ test is measuring the wrong factor, or a whole string
of nines on the normal gaussian distribution is still insufficient for
intelligence to be the central mountain on the mind's landscape. Because
you cannot just raise and raise and raise intelligence, including the
amount of brainpower devoted to cognitive characteristics such as
self-awareness, without affecting the personality.
I personally find it quite plausible that the IQ test is useful for most
of the gaussian distribution but fails to measure anything useful or
important at the fringes, i.e., the alleged distinction between Marilyn
vos Savant and someone with an IQ of 170 would not show up in any major
way if we opened up their brainware and started measuring the effective
processing power of different cognitive subsystems.
> Does this have implications for creating a SI?
It has implications for how to find the people needed to create seed AI.
It has little or no implications for superintelligence, except insofar as
the interaction between ever-increasing intelligence levels and an
installed base of human-universal emotional complexity has implications
for Friendly AI.
-- -- -- -- --
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 - 08:09:39 MST