From: Randall Randall (randall@randallsquared.com)
Date: Fri Jan 02 2004 - 14:53:50 MST
On Friday, January 2, 2004, at 03:06 PM, Perry E. Metzger wrote:
>
> Randall Randall <randall@randallsquared.com> writes:
>>> I agree that people who believe strong superintelligence is
>>> impossible are memetically distant enough from Singularity-aware
>>> thought that trying to avoid offending/confusing them is pointless.
>>
>> This is not at all true. I think it's quite arguable that
>> strong superintelligence is impossible. For instance, it
>> might be that physical law doesn't permit complexity above
>> some ceiling. If so, it might be that the smartest possible
>> humans are already very close to that limit.
>
> I tend to think one can logically rule this one out -- strong
> superintelligence seems compatible with physical law. Why? Because we
> already know of ways to make components with much higher speed and
> much smaller size than human neurons have. One could therefore assume
> that it is possible to get at least a few orders of magnitude
> improvement by moving to non-biological substrates. Even if we just
> made a human who could think a few hundred million times faster, that
> would be a pretty amazing improvement -- and we can be (I think)
> reasonably sure that is possible.
I'm not disputing this. It's just that this is usually known
as "weak" superintelligence, since the mind so described cannot
think thought it couldn't have otherwise, and will not have a
higher limit on conceptual understanding than it's much slower
brethren. To use a trite example, a dog that is speeded up a
million times will not be able to think as well as a human that
isn't. Faster doesn't automatically mean more intelligent.
Clearly, a human is more intelligent than a dog in a "can
understand more complex concepts" way. My assertion above
was that it might be that there is a soft limit on the
complexity of concepts that a mind can understand, independently
of the speed of any given mind.
-- Randall Randall randall@randallsquared.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:43 MDT