Revolutionary Thoughts

[Thresh]

How much AI is enough?

More than is present.

Todd

[NeverMan]

ORIGINAL: Thresh

How much AI is enough?

More than is present.

Todd

This reminds me of a funny poster I saw yesterday, the poster read:

Free Beer......Tomorrow!

[Marshall Ellis]

ORIGINAL: Mardonius
Exactly Marshall. There are so many dynamics in this game that go way beyond the brute force calculation that I too doubt that an AI will be even moderately capable when compared to a mid capability human.

And why do I keep asking about the AI and how much is enough? Glad you aked. Until now we have put the AI (and bug working out... which is another issue) as a higher priority than the OPTIONS or scenario editor. I think I know why... it probably has a bit to do with that survey you put out a few weeks back. Great customer solicitation but I am concerned that this pursuit of the deft AI will mean that you will never turn to the scenario editor and the OPTIONs similar to the ones I have been advocating so quietly.

At some point, you and Mssr Rutins (proud Vermonter) et alias should, perhaps, consider, puting one through the field goal posts and getting some more points on the board. A 1807 scenario should be a piece of cake ("slam dunk", in my hero, George Tenet's parlace)... other OPTIONs -- from both the SCRIPTURE and perhaps from apocryphal sources like those above -- could follow at regular intervals in your bug stomping campaign.

Vive La Revolution!
Mardonius

Mardonius:

I tend to agree with some of your obersavtions and we did pull back a bit on the scheduled AI improvements that we wanted to add but that was to help speed up and introduce some critical fixes. I hate to say this and some could argue but these critical fixes were game killers and had to take a P1 over the AI. I made that decision because I think it was the right one. I can and will keep improving the AI over the next several releases but would love the chance to get some fun stuff (Editor) out too!

Our AI improvement focus was a result of the overwhelming messages about what it was lacking and how it was acting. We had to react to this and we are.

[Mardonius]

Hi Marshall:

Man who works 7 days a week 'til the wee hours. [:)]

No, you are spot on. I think the decisions to date regarding the AI and Bugs have been the proper ones and needed. One must have some basis of an AI... and I, too, don't think we have quite yet reached that finish line. I'd just like to canvas our community to figure where we should strive to have that line.

My own druthers for an AI would be sort of like one of those jousting tilt-a-whirls that knock you off your horse if you don't duck in time. Great to practice on but if you want a real challenge, best take on the Black Knight. I reckon that is, at least an obtainable and useful goal.


All the best,
Mardonius

[delatbabel]

Programming AI is very hard. I have tried it a few times on various projects, and it's not fun. That is all from me on this topic.

[NeverMan]

ORIGINAL: eske
I certainly will. Just hoped you could give me a pointer to which programs you think of. You obviusly know more of recent succesfull chessprograms than I do. I sort of drifted to other fields when Deep Blue won against Kasparov. That was - as you probably know - based on the brute force approach.

I would be especially interested in programs that use both genetic algorithms and neural netwroks.

/eske

Deep Blue had other aspects of "intelligence" besides brute force. You are correct, it was "based" on the brute force approach, but you imply that is it's only aspect of "intelligence", when in fact this is not true.

Here are two OLD projects (~10 years ago) out of UC schools:

http://satirist.org/learn-game/projects/morph.html

http://satirist.org/learn-game/systems/sal.html

Of course if you are really interested in the newer techniques, I would STRONGLY suggest ACM and/or IEEE portals for paper searches.

There are certainly more interesting topics for AI than Chess; however, it's a classic everyone has some knowledge of, this is why I brought it up.

[gwheelock]

ORIGINAL: HanBarca

mmmhh....no program has ever passed the turing test. I'd be surprised if EiA AI could make it

This is not completly true. There have been a few cases where people
have mistaken an Eliza program (a rather simple program that mimics
a freudian psychologist) as another person making fun of them.

What the program would do is take in a general statement; perform
some basic English manipulations to transform it into a question & then
throw it back at them ... or sometimes just throw out a generic question.

Sort of like this : (note the simple, basic English manipulation - without a
shred of real understanding of anything said - it just picks apart the
sentence ["don't like cheese"]; makes a few pronoun changes [I -> you]
and occasonally throws in a canned phrase ["because you hate your mother"])

Me : I hate pizza.
E: Why do you hate pizza .
Me: I hate pizza because I don't like cheese
E: It it that you don't like cheese because you hate your mother

You can probably see why people who thought that they were talking to another
person might get p-oed & think the other guy was making fun of them.

[bresh]

My big concern is the game balance.

As far as i can tell if we dont want hardcoded morale for minors, corps sizes etc.
If soldiers can improve so can the leaders. Ranging 1-6.
Why stay there, all leader stats can change to ?? Win a battle +1 in a stat.
Loose a battle -1 in a stat..........

Soon napy will run around with 6.6.6 rating, and all the rest have 1.1.1

This is realistic, so lets play that way.. since we are not held up by EIA-rules.

Sorry yeah sarcastic comment

Regards
Bresh

[RayKinStL]

This is such a silly debate sometimes.  Mardonius the side you debate is completely different from my (or anyone elses) expectations.  I don't not expect Matrix to put out an AI that will mimic how a human plays.  I understand that some of the decisions that are more arbitrary and based on emotion would be hard to duplicate in AI thinking.  That said, I do not think it an unreasonable expectation to take the diplomacy AI existent now, and pair it with some sort of intelligence with regard to movement, attack, defense, and supply.  I believe an AI to strive for would be one that knows where to commit its forces, and in what numbers.  And in the same vein, when to pull back and defend or even throw in the towel.  Plus, while we have seen some of it, a better understanding of how and when to attack supply lines would be advantageous for the AI.  I don't know WHEN the AI will get there, and I don't know what the final version should, or would, look like.  But programming is not running.  And to take an end justifies the means approach is not applicable, in my opinion.  I think with each iteration of the game, Matrix makes strides towards getting the game AI to an acceptable level.  But if you need an idea of where to go, I think the AI needs to get to the point where land and naval phases/combat are conducted in similar fashion to how a beginning/intermediate human would use the forces.  Until the AI stops throwing single suicidal corps at me, stops sitting in one region with an unmovable massive corps stack, and stops wasting whole calvary corps to try to assault a minor woth 1/1, there are improvements that can and should be made.

[Mardonius]

Hi Ray:

I think you make some valid points. What you have written, perhaps, we should consider as the AI goal. Then, at least, we know when we are finished.

best,
Mardonius

[Marshall Ellis]

ORIGINAL: Mardonius

Hi Marshall:

Man who works 7 days a week 'til the wee hours. [:)]

No, you are spot on. I think the decisions to date regarding the AI and Bugs have been the proper ones and needed. One must have some basis of an AI... and I, too, don't think we have quite yet reached that finish line. I'd just like to canvas our community to figure where we should strive to have that line.

My own druthers for an AI would be sort of like one of those jousting tilt-a-whirls that knock you off your horse if you don't duck in time. Great to practice on but if you want a real challenge, best take on the Black Knight. I reckon that is, at least an obtainable and useful goal.


All the best,
Mardonius

I'll take this as a goal ... to knock you off your horse

[AresMars]

ORIGINAL: Marshall Ellis

I'll take this as a goal ... to knock you off your horse

Just shoot the horse Marshall!

So much easier then fighting some armoured Knight!



ON THE AI:

We can all agree that this will always be a work in progress, that should last forever.... [:-]

[fvianello]

ORIGINAL: gwheelock

ORIGINAL: HanBarca

mmmhh....no program has ever passed the turing test. I'd be surprised if EiA AI could make it

This is not completly true.

There's a competition every year, and no one was ever able to get the first prize (unrestricted turing test) nor the second (restricted turing test). the third prize goes to the "best" program among the competitors and of course is always assigned.

[NeverMan]

From wikipedia:

While neither ELIZA nor PARRY were able to pass a strict Turing Test, they - and software like them - suggested that software might be written that was able to do so. More importantly, they suggested that such software might involve little more than databases and the application of simple rules. This led to John Searle's 1980 paper, "Minds, Brains, and Programs", in which he proposed an argument against the Turing Test. Searle described a thought experiment known as the Chinese room that highlighted what he saw as a fundamental misinterpretation of what the Turing Test could and could not prove: while software such as ELIZA might be able to pass the Turing Test, they might do so by simply manipulating symbols of which they have no understanding. And without understanding, they could not be described as "thinking" in the same sense people do. Searle concludes that the Turing Test can not prove that a machine can think, contrary to Turing's original proposal.[21]

I'm inclined to believe that some real people can't even pass the "Turing Test" (as the assumed definition, although Searle makes very excellent points).

[yammahoper]

I had a first happen to me the other night. 

I fired up EiA with the latest beta patch and started a game as Turkey.  I placed a three corp army in the holy land to invade Egypt and my fleets to bring in a diminshed Nazud corp and full Janissary with transport.  I put the other Janissary corp outside Constantinople and decided to fucus on getting North Africa.

Egypt held out till April and I moved West.  Spain had seized Morroco, and while I was in Cyrenica, Spain declared war on me and moved a corp with Castanos into Algeria.  I had declared war on Tripitania, so I moved my fleets into the blockade box with the transport feet off the coast to drop the full Janissary corp and supply the invasion.  To my delight and frustration, the Spanish navy moved in and captured my transport fleet and entire Janissary corp! 

On the negative side, after two losing battles in Algeria with the Algerian corp, I had the Spanish corp down to one factor while I had five factors left.  AT this point a bug occured and my Algerian corp became effectively invisible.  Even though I was besieging the city, control switched to spain and I lost the city.  Now I had the Spanish Heavy fleets sitting in the water off Tripitania so I could not move mine (though it did not blockade, which must have cost it a fortune), my Nazid corp trapped, AND I had made Syria a nuetral so my Anatolia corps could invade it and make it a conquered.  The result was a trip to instability and the loss of all my gains!

Then Austria declared war on me.

So now I am totally stapped for cash, my allies Russia and France keep refusing me any aid, even claiming my request for $5-10 are excessive, and so far the AI is being fairly responsive to my aggression.  These are nice improvements.  I have lost battles before, but never before has the AI handed me my hat like it did last night.

yamma

[fvianello]

ORIGINAL: neverman

From wikipedia:

While neither ELIZA nor PARRY were able to pass a strict Turing Test, they - and software like them - suggested that software might be written that was able to do so.

...and.. ?

That's the reason I entered this thread [:D]

[NeverMan]

ORIGINAL: HanBarca

ORIGINAL: neverman

From wikipedia:

While neither ELIZA nor PARRY were able to pass a strict Turing Test, they - and software like them - suggested that software might be written that was able to do so.

...and.. ?

That's the reason I entered this thread [:D]

The point of my posting that wiki paragraph was to show that possibly, the Turing Test itself is not a good measurement of AI and we, as a community, need to rethink what AI really is. That is all.

Oh yeah, and to make a joke about real people not having enough "I" to often know what they are doing or saying. [:)]

[fvianello]

...but... ?

[Marshall Ellis]

ORIGINAL: yammahoper@yahoo.com

I had a first happen to me the other night. 

I fired up EiA with the latest beta patch and started a game as Turkey.  I placed a three corp army in the holy land to invade Egypt and my fleets to bring in a diminshed Nazud corp and full Janissary with transport.  I put the other Janissary corp outside Constantinople and decided to fucus on getting North Africa.

Egypt held out till April and I moved West.  Spain had seized Morroco, and while I was in Cyrenica, Spain declared war on me and moved a corp with Castanos into Algeria.  I had declared war on Tripitania, so I moved my fleets into the blockade box with the transport feet off the coast to drop the full Janissary corp and supply the invasion.  To my delight and frustration, the Spanish navy moved in and captured my transport fleet and entire Janissary corp! 

On the negative side, after two losing battles in Algeria with the Algerian corp, I had the Spanish corp down to one factor while I had five factors left.  AT this point a bug occured and my Algerian corp became effectively invisible.  Even though I was besieging the city, control switched to spain and I lost the city.  Now I had the Spanish Heavy fleets sitting in the water off Tripitania so I could not move mine (though it did not blockade, which must have cost it a fortune), my Nazid corp trapped, AND I had made Syria a nuetral so my Anatolia corps could invade it and make it a conquered.  The result was a trip to instability and the loss of all my gains!

Then Austria declared war on me.

So now I am totally stapped for cash, my allies Russia and France keep refusing me any aid, even claiming my request for $5-10 are excessive, and so far the AI is being fairly responsive to my aggression.  These are nice improvements.  I have lost battles before, but never before has the AI handed me my hat like it did last night.

yamma

Yamma:

Do you have this saved right before the control switched?

[Marshall Ellis]

ORIGINAL: NeverMan
Oh yeah, and to make a joke about real people not having enough "I" to often know what they are doing or saying. [:)]

Hey! Watch it! I do have plenty of "A" though LOL!