From: James Rogers (firstname.lastname@example.org)
Date: Thu Aug 14 2003 - 12:22:12 MDT
Ben Goertzel wrote:
> James Rogers wrote:
> > In all likelihood there is probably
> > only one pragmatic solution to AGI surrounded by a space of
> > credible close approximations.
> My intuition differs ... I suspect there are many
> qualitatively different solutions, corresponding to different
> sorts of minds.
My view wasn't really an "intuitive" one, it was more a pragmatic economic
I think it is quite rational to assume that there is only one theoretical
solution in the general abstract. With respect to real implementations, the
efficacy of a particular design will drop quite rapidly as approximations of the
theoretical ideal become poorer. Therefore, I would say that it is a reasonable
assumption that all high-efficacy designs will look quite similar to each other.
Yes, there is a huge range of qualitatively different solutions that will work
in theory. But given finite resource limitations, most of these solutions will
be essentially worthless when competing with high-efficacy designs. Most
functional differences between high-efficacy AGIs will be window dressing.
So to put it another way: You can design a theoretically viable AGI in any
number of ways, but when you throw IQ/$ into the equation (a real-world
viability factor) one type of AGI design will completely overshadow all others.
Unless, of course, all the competing designs really suck (i.e. the prevailing
market at the moment).
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT