[extropy-chat] 12,000 IQ and nothing on?

[Anders Sandberg]

Lee Corbin wrote:
> You put the existence of societal "problems"
> relating to low IQ in one category, and overall
> differences (i.e. distributions) in what I was
> calling "general accomplishment" or the potential
> to accomplish, in a separate category, which is
> also very interesting, and seems probably right
> to me.

Accomplishment often means the creation of positive wealth while problems
signals wealth destruction or lowered chances of a flourishing life.

> Of course, in this thread we are more concerned
> with the latter (as a possible reflection on the nature
> of superhuman intelligence) than on the former,
> despite the greater sociological and economic
> importance of those "problems" traceable to less
> than 100 IQ.

I think there is also a big opportunity in enhancing the normally bright.
Moving many millions from 95 to a 100 means a lot of reduction in problems
and a lot more accomplishments. Just the little thing of being able to do
jobs with written instruction (limit around 95 or so) opens a whole range
of opportunities.

> To me the fundamental problem is how closely
> the nature of extreme capability (as opposed to
> the notion of cognitive ability---in order to avoid
> begging the question) follows any kind of
> approximately linear scale. By now (I go along
> with the psychometricians) we have for humans
> that there is an approximately linear scale for
> human cognitive ability.  But this may break
> down---as you mentioned in your previous
> email---for very advanced entities.

I'm not certain there is any way of callibrating it other than enormous
competitions. But making a competition that works for ants, mice, chimps,
humans, transhumans, AIs, posthumans and fnorgnitzbs would be nearly
impossible. There seems to be at least a kind of understanding horizon
beneath humans that means that simpler creatures cannot understand general
concepts, but more complex creatures probably can emulate each other a la
the Church-Turing thesis.

> I now add a third possibility---namely that
> some component may be measurable by
> mathematical achievement. (This goes in hand
> with a contention I've had for a long time that
> perhaps extreme intelligence rather trivially
> solves all non-mathematical problems
> comparitively early in its development.)

Math might be a good choice since it includes many different levels of
complexity and one can find a wide variety of domains. Some quite similar
to the everyday world, some utterly different. Of course, some of us would
say the everyday world is just embedded in math.


-- 
Anders Sandberg,
Oxford Uehiro Centre for Practical Ethics
Philosophy Faculty of Oxford University