Re: answers I'd like from an SI

From: Thomas McCabe (pphysics141@gmail.com)
Date: Mon Nov 12 2007 - 17:30:11 MST


> Getting back to the topic at hand, I think a superintelligent AI is
> not going to understand GR all that better than the experts today do,
> as there is just not that much more to understand about it. It's easy
> to say an SI will be thousands of times better at explaining something
> that many people consider hard to learn. But it only seems hard and
> mysterious until you fully learn it. Do you think an SI is going to
> be able to explain how to cook a turkey 100 times faster than a normal
> person? Sure... they can say the words 100 times faster, but the
> person listening who doesn't know how to cook the turkey, is still
> going to have to hear it explained at the same rate.
>
> Jeff
>

This is the result of a universal bandwidth limitation on the human
species, much like the universal limitations on serial processing
speed, memory, cognitive capacity, etc. Outgoing human communication
is ~300 baud. Incoming communication can be much faster, but only for
highly specialized data (sounds, pictures, etc.). Even a primitive BCI
can do better than this. A superintelligence could just use implanted
nanotechnology to beam the ideas into our heads, quite possibly in
less than a microsecond.

 - Tom



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:00 MDT