Proof of stupid AI spellcasting
 

Here's a shot of a Deathmatch fight between an Inquisitor and a Golden Naga Pretender
Look at the Inq's spells cast... No comment ! http://forum.shrapnelgames.com/images/icons/icon10.gif http://forum.shrapnelgames.com/images/icons/icon10.gif http://forum.shrapnelgames.com/images/icons/icon10.gif http://forum.shrapnelgames.com/image...s/rolleyes.gif
http://pascal.difolco.free.fr/Dom2_AI_Crap.jpg

Re: Proof of stupid AI spellcasting
 

I just saw this Last night. I went against Pythium with 2 Theurgs. One cast (let's call him "Joe") Communion Master and the other ("Fred") cast Communion Slave. Then the next Messages were "Fred casts communion master" and "Joe casts communion slave"! It was ok for me since they didn't actually USE the power they gained, but still....

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">It's not that stupid. That's a witch hunter from the picture. That means that it has fire 2 and astral 1. The only spell I can see it using to damage the Naga is star fires.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">The stupid part is casting Slave and then Master, unless that combo actually works, which it isn't supposed to ...

Re: Proof of stupid AI spellcasting
 

I see it mostly as a "look at the stupid, unwinnable fight my god got me into... mebbe if I act crazy she'll go away" http://forum.shrapnelgames.com/image...s/rolleyes.gif

Actually, I suspect the AI puts a high value on casting CS but includes itself when "looking" to see if there are any Slaves before casting CM. One "fix" might be to require a minimum of two available slaves before the AI is willing to cast CM.

~Aldin

[ February 21, 2004, 18:45: Message edited by: aldin ]

Re: Proof of stupid AI spellcasting
 

Is this with the latest patch? By the way, has anyone noticed an improvement in AI spellcasting with 1.08?

Re: Proof of stupid AI spellcasting
 

Did you have star fires researched? If the mage dont have any usefull spell to cast it does not really matter which useless spell it casts.

[ February 21, 2004, 20:09: Message edited by: mivayan ]

Re: Proof of stupid AI spellcasting
 

Eh, cast Body Ethereal, Personal Luck, Astral Weapon and charge. The audience wants to see blood, not a mage sitting there trying to commune with himself. http://forum.shrapnelgames.com/images/icons/icon12.gif

Re: Proof of stupid AI spellcasting
 

Some precisions :
This was a battle between AIs, in 2.08, and I just witnessed the match as Pythium without participating !
After the masterful CM+CS casts, the Marignon mage didn't even try Star Fire and was eaten by the Naga ... http://forum.shrapnelgames.com/images/icons/icon10.gif
Funniest was that in the end the Naga was killed by a Prophet Vanjarl who get it exhausted fighting Phantasmal Warriors ! http://forum.shrapnelgames.com/images/icons/icon10.gif http://forum.shrapnelgames.com/images/icons/icon10.gif This spell was a really good choice this time (at first I found it silly as the PW just delayed the Naga for one or two rounds ...)

[ February 21, 2004, 22:43: Message edited by: PDF ]

Re: Proof of stupid AI spellcasting
 

Then you should rename the thread PDF, because your example proves nothing.

This was a mistake of the strategic AI sending to the arena a mage that knew no combat spells (ie, had a little thau research, but no evo or alt research). Mages default to cast, if the mage just knows crap he will cast crap regardless of how good or bad the spell AI be.

Despite the nay-sayers, the spell AI is one of the best parts of the game. You want some proof? get the same mage, research evo1 & thau1 and send him to the arena...he will cast star fires instead of com master/slave. But you cannot plant melons & expect apples to grow.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">Hmmm, I could go so far as to say the AI is ok, but that due to the nature of the spell complexity that this is a really complicated process that needs some improving. You really need special logic for a lot of special cases like this - it should prevent mages from casting Communion slave without having any mages to benefit from this, or communion master if there are no mages to be slaves possible.

And the mage would have been able to at least cast Twist Fate, so he did have one more good spell to cast. If nothing else, then I guess he should have entered melee after that. but let's face it, he was dead.

Besides specific spell quirks like the Communions, the AI does have other problems that cause some problems:
a) no concept of it's chance of successfully hitting a target. Like a Warlock trying to Nether Darts an lone final unit, while he should be Mind Burning.

b) no concept of it's chance of successfully beating MR. The AI loves to repeatedly attack a SC pretender with MR check spells in his own dominion with 20+ MR, while he could instead by taking out the calvary units riding down on him and destroying his line that are the more immediate threat. I'd like to factor in the closeness of the unit to the mage and to friendly troops, how long before it could be in range (ie flyers should have a higher rating here), and some threat level (HP, skills, etc) with some random factors so it doesn't just always strike a particular unit.

c) no concept of friendly fire losses. Same case as a, but this time your warlock nukes a good portion of your troops. Even worse if that warlock has an eye loss or blinded. There really should be some logic for checking whether area/multiple effect spells should be used or the 100 accuracy single target ones based on the size of the force facing you, and how likely your troops are to get hurt which takes into account the mages precision.

d) Does the AI still pick spells that are obviously lesser Versions of another? Like Blessing instead of Divine Blessing, or FirebLast instead of Fireball. Unless this was fixed recently, this was a concern in previous Versions.

e) Casting area effect protections spells in crowds. Breath of Winter, Soul Drain, Fire Shield, etc from mages that fry your nearby leaders/bodyguards is an annoying flaw. You learn very quickly to place mages who don't play nice away from others, but early on you just get pissed and later on you only occasionally get pissed as you will forget one. The one I forget about the most is a random magic 1 Fire mage who casts fireshield - have to remember to sort them out and name them properly so I don't forget!

Sure there are more too. I am not alone in basically scripting out all my battle mages first 5 spells, and I know I definitely restrict my research to prevent my mages from casting spells I think will cause more damage on my side than not. That's a sign that the AI could be improved.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">Fire shield only affects units that hit the mage. There is no battlefield spell that gives a particular unit heat radiation.

Re: Proof of stupid AI spellcasting
 

Geo:

Your points a,b and c are mistaken. the AI does a calculation based on chance to hit MR, damage, tries to some extent to avoid friendly fire etc etc. Perhaps you do not think it wieghts friendly fire enough but it do these things.

Re: Proof of stupid AI spellcasting
 

George,
I do not think the AI rates the spells as castable/non castable, I think it rather hands out a value number to different spell+target comboes, and selects the spell with the highest number from a sample of alternatives. Thus useless spells will be cast even if their value is low, because there's no other option of higher value.

I do not want my mages & priests charging into melee if they have no usable spells, I do not think you want yours to do it either. If I tell'em to cast, I want them to cast.

On the matter of the AI targetting rules, I am pretty sure hps, big clusters of units & distance are taken into account, both in the spell choice & the target choice. SCs are a curse magnet, clusters of troops attract area spells/multiple effect spells, and mages often switch targets to hit whatever just showed up next square (not the 1st time I fly a SC towards a bunch of theurgs and get them to drop whatever crap they were casting & hit with multiple Star fires). You will not always agree with the choices made, but I see the AI often making better choices than many human opponents, so it's not like any fight that goes past the 5 scripted spells is doomed to spell missusing.

I gave an example of the AI doing better than me here:
thread

I only consider the breath of winter/soul drain complaint valid (and to a lesser extent the related friendly fire one), and maybe the AI should be better also at identifying elemental/poison damage & acting accordingly in order to raise protections..but apart from that I think it's extremely good.

Besides, did you notice that Dom's II AI is far less prone to fall to gem draining tactics? I find myself saving a lot of gems when I catch scouts around.

[ February 22, 2004, 20:51: Message edited by: Wendigo ]

Re: Proof of stupid AI spellcasting
 

Saw the same success of a Vanheim prophet in the Arena. He won two tournaments, taking out at least two pretenders. Used the same strategy you mentioned with phantasmal warriors and the fatigue they caused. Finally succumbed to the Great Mother, in his third tournament, due to routing from his own spell-casting fatigue.

Aikamun

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">Maybe the AI Marignon nation had few spells, but it was on turn 12-15 IIRC, he should have had some..
And anyway the guy should rather run, or attack with fists than casting CM+CS when he's alone !! That he only knows totally useless spells (in these conditions) isn't a reason to cast them...

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">IIRC, you cannot run when in the arena, which is where the pic was from.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">It had researched at least conjuration 3 and thaumaturgy 1. I imagine it was going for conjuration 4 for summer lions, which is what Marignon tends to do as an AI.

<font size="2" face="sans-serif, arial, verdana">Do you really want your mages to run into a melee on the battlefield when they decide that they don't have a useful spell to cast?

Re: Proof of stupid AI spellcasting
 

Graeme --
In this case, the choice was between casting useless spells or melee, because in the arena it's victory or death (unless, perhaps, you can Return out. Hm, if it's a one-round tournament, can you summon Lammashastas and return out, and still win? Hmmm.) If he has no useful spells, the choice became melee or die.
Casting useless spells also drives up fatigue, which could be bad if there's an opportunity to cast an especially draining spell later, or if melee is a risk, or if communion is an issue -- useless spellcasting is good for butchering one's own communicants, for instance.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">IIRC, you cannot run when in the arena, which is where the pic was from. </font><hr /></blockquote><font size="2" face="sans-serif, arial, verdana">He should try to run in circles, to exhaust his opponent ! http://forum.shrapnelgames.com/images/icons/icon10.gif http://forum.shrapnelgames.com/images/icons/icon10.gif ... I recall Chaplin doing that on a boxing ring, against a big badass , in a movie (don't recall which ..) http://forum.shrapnelgames.com/images/icons/icon10.gif

Re: Proof of stupid AI spellcasting
 

The AI never closes voluntarily for melee unless specifically told so, and I think most players would be very unhappy if this were to change, and expecting the AI to take into account the few times when he is alone in combat and then act differently in this very rare and special situation is to, in my mind, put to unreasonable demands on it.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">Are you sure you didn't leave his orders as cast spells or with no orders instead of telling him to fire closest? If his final order was left blank, then he will default to casting spells as a starchild is a mage.

Re: Proof of stupid AI spellcasting
 

Not to inflame things, but there was a mention awhile back (by Kristoffer) about possibly changing Breath of Winter so it'll only be cast when specifically scripted. That would be sooo nice.

I was just playing R'lyeh and the Traitor King pulled his (now familiar) trick of casting Breath of Winter without being told to and then dancing all about the rear row, doing great harm to my mages, illithids, etc. I wasn't particularly unhappy, even though he is a great water mage, when he finally went down under Abyssian fire magic.

I think spells that can damage friendly troops around the mage should not be auto-cast. Even when I set the water mages apart, they tend to come back towards the main group sooner or later, and they are also Banned from having non-cold resistant bodyguards while apart.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">His orders was luck, fire x4, cast spells. I dont mind him casting usefull buffing spells before star fires, but communion master and slave are not exactly usefull during an assassination.

<font size="2" face="sans-serif, arial, verdana">I would like that very much. For communion master and slave too.

[ February 24, 2004, 23:57: Message edited by: mivayan ]

Re: Proof of stupid AI spellcasting
 

Here is what my star child assassin did during an assassination against an ulm prophet:

scripted actions: cast luck. fire: mindbLast, mindbLast, mindbLast, mindbLast.

cast spells: astral shield, body etheral, astral weapon, resist magic, twist fate, communion master, communion slave, star fires, star fires, star fires, star fires.

There was a brief message: "the armies of special monsters have been routed". Then he ran away. The ulm commander was still paralyzed(5).

I would prefer that the ai dont cast communion master or slave when alone. In fact, I would prefer if it was never cast at all unless scripted to.

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">about weights of AI parameters... The best AI so far, in commercial games, are were there weights are externalized in text files, so that the most zealout of the fans can tweak and propose alternative solutions. As IW is a very small company, it would benefits even more from that process.

I remember the example of some Age of Kings (AoE II) or Kohan fans made AI which were able to routinely beat developpers made AI at 1 against 2. Its just because fans can put far more hours in a specific topic than the devs. I know of some fans, for the Hearts of Iron game (Paradox big simulation of WW 2) who spent more than 1000 hours, just on AI. (SuperAi pack 1.3 is far more superior than the Paradox vanilla AI eg).

Re: Proof of stupid AI spellcasting
 

<font size="2" face="sans-serif, arial, verdana">HoI w/ CORE is by far superior to plain HoI. OTOH, Paradox is infamous for coding AIs that are notorious for doing rather stupid things, so saying that CORE is "much better" really isn't saying that much. http://forum.shrapnelgames.com/images/icons/icon12.gif

If IW were to expose their AI, then we fans could have a field day tweaking it so that it presents an even greater challenge than it already does. One of the things that's most needed is the ability to enable/disable the AI from being able to cast spells we don't want it to. The spellcasting list could be exposed via a config file that we can just tweak with true/false flags. (Breath of Winter=0) The game would still keep its internal list of what's castable, but it then checks the config file for what's permitted. It's a straightforward fix, that requires no GUI changes, and only adds the burden of an extra file I/O at the start of a gaming session and the overhead of maintaining the config file's contents in memory afterwards.

Re: Proof of stupid AI spellcasting
 

add to this the preferences about castles, temples, etc. in AI budget. 2 patches out, and we still have AI building mostly light indeps troops, because there is never new castles build by the AI.

That is to the point that fighting Ulm, Man or Pythium is often the same thing, with a mass of rabble in front of you, with the occasional national unit or mage thrown into the lot.

Re: Proof of stupid AI spellcasting
 

I seem to recall reading that external casting lists/scripting... will not be put in due to wanting the better player to win and not the better script writer.

Personally I would like something like an external list, too even though I am fairly clueless about writing programs or anything like it, but this shoulnd't be too difficult and would actually add even more depth to the game.

CharonJr

Re: Proof of stupid AI spellcasting
 

My suggestion for externalizing the spells-permitted list is just that, a list. Only one option per spell: 0 or 1, no-cast/castable. of course, if the devs want, the options per spell could be expanded to allow much more variety, but a simple yes/no is all that's really needed to "fix" the AI spellcasting silliness. Anything more is just icing on the cake.

I do not advocate external script lists, though I do think they'd be pretty cool.