RE: Seed AI (was: How hard a Singularity?)

From: Ben Goertzel (
Date: Tue Jun 25 2002 - 09:44:07 MDT

> > to engineer human-level intelligence (perhaps better than human
> > intelligence in some ways, and initially worse in others). A hard
> > problem, but a different one.
> I have the impression that this is an extremely hard problem. (Naturally
> intelligent systems, not idiot savants).

Yes, creating an AGI is certainly an extremely hard problem!

If we succeed in solving it, it will be a huge and incredible achievement.

We'll spend at least 6 weeks on vacation, and then set about collaborating
with our AGI on building uploading-enabling and life-extension technology...

Building a spaceship to fly to the moon was also an extremely hard problem,
as was arriving at the Standard Model in particle physics, as was
understanding the genome well enough to do even the simple genetic
engineering we can now do. The human race has solved some extremely hard
problems before. Is this one (AGI design/engineering) harder than any of
those? It's hard to compare, isn't it?

 -- Ben G

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:39 MDT