Re: AI
From: Alan E and Carmel J Brain <aebrain@d...>
Date: Mon, 28 Feb 2000 21:29:14 +1000
Subject: Re: AI
Roger Books wrote:
>
> On 25-Feb-00 at 18:05, sportyspam@harm.dhs.org
(sportyspam@harm.dhs.org) wrote:
>
> > Hmmm, another example might be tanks. Someone was musing about
them.
> > You have to figure, 150 years the AI is going to be pretty good,
good
> > enough that you won't need someone in the tank.
>
> I don't know, the pattern recognition will be much better, possibly
the
> threat ranking will be better, but as far as I can tell there isn't
> that much progress being made with respect to AI.
Speaking as probably the only person on this list who's actually
designed and built a military-grade AI (for naval anti-missle defence),
I'm not so sure.
You don't have to have something much smarter than an ant, or at best a
frog, to be really useful.
Some large percentage of the time - 50%, 90%, 99% ,depending on the
exact context, a simple "reflex" system will do the job. Forex if a tank
defence system detects an object incoming at 500-1500 km/h (or whatever)
and it's heading directly towards the tank being defended, it doesn't
have to know very much in order to take appropriate action. Similarly an
object heading in at 3000+ km/h will be treated as a hostile long-rod
penetrator. Yes, it might be a piece of shrapnel. It might even be
friendly fire. But if the projected path intersects the defended tank,
it will be treated as a Baddie, and dealt with.
In the case of a tank, a lot of the routine decisions, such as selecting
ammo options and programming the FC computers accordingly, can be done
automatically with no more smarts than a Z-80, and done faster. The easy
bits should be automated, as very often the requirement is a really
simple decision to be taken and acted on in 1/10 of a second. But to try
to automate the harder bits requires vastly more intelligent systems: it
just isn't worth it in the general case. But if 90%+ of the load can be
removed from the manpower, they can concentrate on the hard 10%, and
deal with that either more effectively, or with fewer heads.
--
http://www2.dynamite.com.au/aebrain
aebrain@dynamite.com.au <> <> How doth the little Crocodile
| Alan & Carmel Brain| xxxxx Improve his shining tail?
| Canberra Australia | xxxxxHxHxxxxxx _MMMMMMMMM_MMMMMMMMM
abrain@cs.adfa.edu.au o O*OO^^^^OO*O o oo oo oo oo
By pulling MAERKLIN Wagons, in 1/220 Scale