From: Jimmy Wales (jwales@bomis.com)
Date: Wed Aug 01 2001 - 12:19:30 MDT
James Rogers wrote:
> I tend to look at IQ as more of a logarithmic measure of effective
> intelligence. The difference between an IQ of 50 and one of 100 is more
> exponential than linear, though the problem when applied to humans is that
> it is more a measure of capability than results.
Oh, yes, I agree completely. And there are surely good questions to be raised
about both the way IQ is measured *and* about whether the multi-faceted thing
we mean by "intelligence" can be measured on a univariate scale, *period*.
The only real point is that simply adding up dumb machines isn't necessarily going
to get intelligence. I don't think anyone disagrees with that. You could take
enough pocket calculators to equal the CPU and memory of the human brain, and you'd
still not have a human brain. You can take 1,000 mentally retarded humans and not
get the thinking power of 1 normal person.
As we approach "human equivalence", we will presumably see smarter and smarter machines
with sub-human intelligence. They will likely be very good at some tasks, and very bad
at others. Already, the typical Intel-PC is better at Chess than the average person, but
can't tell the difference between spam and mail from a friend.
It's convenient for theorizing (Moravec-style equivalence estimating) to think of
intelligence on a univariate scale, but really that isn't true.
--Jimbo
-- ************************************************* * http://www.nupedia.com/ * * The Ever Expanding Free Encyclopedia * *************************************************
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT