Just a hunch...

4 posts ยท Jul 13 1997 to Jul 15 1997

From: Alex Williams <thantos@d...>

Date: Sun, 13 Jul 1997 15:04:23 -0400

Subject: Re: Just a hunch...

> On Sun, 13 Jul 1997, Rodrick Campbell wrote:

> is true that computers anr faster at sequential logic processing, I

Actually, this isn't true... back in the hayday of linear logical rulesets,
one of the biggies was called GPS, a rule system that would make logical
proofs about geometric ideas when given sets of axioms and rules for
interaction. GPS occasionally came up with unexpected paths of logic that
differed from how it had traditionally been proven that certain geometric
properties worked.

Does this not qualify for what you're asking for above? Given that a
machine won't /say/ 'Eureka!' if they come up with something new,
they'll simply accept it as it comes. Asking for the 'Eureka!' isn't asking
for Intelligence, but human bias.

From: Rodrick Campbell <rcampbell@u...>

Date: Sun, 13 Jul 1997 17:20:21 -0400

Subject: Just a hunch...

Paul,

No, I'm not from Cambell, CA. I live in the little backwask town

of Oil City, PA, birthplace of the oil industry. Woo Hoo!

Alan,

I agree that this has been one of the more interesting discussions (in the
five days I've been reading them anyway). Although it is true that computers
anr faster at sequential logic processing, I haven't found one yet that can
play a hunch or gets a gut feeling about things. It will be a long time before
any AI can make intuitive leaps of

logic.

I think the ability of humans to make judgement calls without complete
information is one of their best assets in combat ( gaming anyway, as I've
never experienced real combat). We put one and one together and sometimes get
two, or one, or three, or one thousand...

On the other hand some great military blunders have been made due to playing
these hunches.

Can AI ever really replace humans in these situations?

From: Allan Goodall <agoodall@a...>

Date: Mon, 14 Jul 1997 22:27:56 -0400

Subject: Re: Just a hunch...

> At 02:20 PM 7/13/97 -0700, you wrote:

> Can AI ever really replace humans in these situations?

I think given enough time, yes. I personally believe that within 50 years, an
automated aircraft will win a Top Gun style tournament (it might even do
it faster than that). G-force loading and speed of target acquisition
will make up for anything intuition provides a fighter pilot. Given 250 to
1000 years, I think AIs will be capable of replacing humans in directing
battle and war strategy.

I don't think grunts will be replaced until a workable AI can fit into a
robotic frame (not necessarily human shaped) that can conduct effective ground
combat. Even then, any side forced into a sufficiently tight corner will
through the robots AND humans into battle. After all, every side except the US
put women into combat roles in WWII, and many nations utilized children and
the elderly.

From: Alan and Carmel Brain <aebrain@w...>

Date: Mon, 14 Jul 1997 23:43:19 -0400

Subject: Re: Just a hunch...

> Rodrick Campbell wrote:
. Although it
> is true that computers anr faster at sequential logic processing, I

Well, you can fake it fairly well. Just use a genetic algorithm using sexual
reproduction, feed in all the scenarios that are plausible or even possible,
wait for a zillion generations and pick the survivors.

They work pretty well at faking intuition, insasmuch as they will, for no
apparent reason, occasionally do exactly the right thing at the right
time.. often not the very-best-possible, but always pretty close. It's
not perfect optimization, but it has no Achilles Heel, which is what you want
in an automated system.