On Fri, 29 Sep 2000, Eugene Leitl wrote:
> Franklin Wayne Poley writes:
> > Maybe
> > even enough AI to improve upon its own hardware and software and evolve
>
> Uh, apart from hardware (duh), you have to impose tight limits on
> software's variability. Evolving evasion maneuvres is one thing,
> evolving friend-foe recognition is perhaps a bit silly.
Silly? Why? If there are objective-real-world-measurable criteria for
selecting friend vs. foe (there are) then you can bet your boots the
military will try to itemize them and program them. Humans err in this,
hence "friendly fire". If programs can do better than humans in reducing
such errors they will be used. Correction...not IF...WHEN.
FWP
This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 17:39:27 MDT