Robin Hanson wrote,
> I think there is a real possibility that AI research is mostly hardware
> limited. If so, then the optimal strategy would be to mostly skip the
> research now, invest the money somewhere while waiting for the hardware
> to get good enough, and then spend lots more on research later on. If
> so, the real waste is spending money on long-term-goal-directed AI
> research now at all. Of course compared to other big wastes in the world
> this is may be small potatoes.
Currently, hardware exceeds software capabilities. In fact, AI advocates and
researchers (from Hans Moravec to Ray Kurzweil) concede that the functionality
of software continues to fall behind that of hardware. With the advent of
quantum computing or computers using molecular sized switches, the potential of
computer hardware will accelerate even farther beyond that of software (at least
software produced by hand). Nevertheless, I feel that with increased focus on
genetic programming, software may catch up, because it will be out of the hands
of human programmers. So, in the long run I completely agree with your
conclusion.
Stay hungry,
--J. R.
3M TA3
This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:50:16 MDT