Can we get a better AI Please?

Users who are viewing this thread




Also what did happen here before? The Aseri tries to get a City back but, the Southern kingdom comes to defend. The aseri cancel the siege and attack the southern army until there start to run.
Now, the southern Army get some "help" and the Aseri change his mind a trys to run away in the first pic (i would attack it, because of the easy win).
But then comes the second Pic. BOTH Armies decided to do something else, instead of fighting each other and here i lost my mind a little.
 
Thats a bad excuse to let a heavily wounded enemy army unharmed.

That heavily wounded enemy army was:
  1. Still roughly equal in strength.
  2. Rapidly recovering.
Even if the Aserai won the battle, their army still wouldn't be able to deliver the siege, so it makes some kind of sense that they just walked away from it. The Imperials' objective is relieving the siege of their town, so it also makes sense that they didn't pursue battle: they'd already done what they came to do.
 
Last edited:
The problem isn't that the AI isn't smart enough, but that on occasions it's a little too 'smart' in assessing its chances and following determined objectives - and not entirely realistically. Armies in a war should display a little arrogance and bravado and take a few more risks, act more impulsively and react to immediate circumstances, not prioritise a siege halfway across the map over an opposing army right in front of them. But that would also produce odd behaviour in particular circumstances and criticism of the AI...
 
Idk it seems to me both sides want a conflict but are wavering because they're so evenly matched. I'm guessing the SE is waiting for troops to recover or get reinforcements and Aseari maybe waiting on getting reinforcements too so biding time is actually pretty smart.
 
Looks to me like the AI needs to be improved, but ok. Let the enemy faction leader just go away, so she can gather new fresh forces to take the next city of yours.
 
Let the enemy faction leader just go away, so she can gather new fresh forces to take the next city of yours.

That point is nothing that the AI considers. The AI calculates if a battle will be a clear victory, if not it will just avoid it (if it can). And that's the major problem with the AI: It constantly knows the exact party, army or garrison composition of every faction and if its own party/army can attack and surely beat it in autocalc.

You could change this behavior if the AI (and the player of course) never knows the exact strength of the enemy party but only a rough estimate. Depending on the scouting skill and distance this estimate could be more or less precise or even completely unknown.

The decision if the AI engages an enemy party should also consider if it can risk loosing its men. For example:
If AI is in war with a superior enemy it should be more defensive and not risk his forces if it isn't necessary to protect a town/castle.
If the AI faction is much stronger than his enemy it should act more aggressive and risk combat even if it could lose the battle, because it can replenish its losses better than the enemy.

There could be other factors that make sense like traits of the army leader (Valor, Honor) that should contribute to this decision. A leader with a high valor value could risk more and a high honor leader could even spare an inferior enemy party if it isn't a direct threat. Or vice versa a low valor leader would avoid unnecessary battles where a low honor leader would attack every outnumbered party it can catch.
 
Looks to me like the AI needs to be improved, but ok. Let the enemy faction leader just go away, so she can gather new fresh forces to take the next city of yours.

It doesn't matter; anyone else in her clan can gather fresh forces equally well. Nabbing her army as a decent consolation prize (they already broke the siege just by showing up) would be fine if it was guaranteed but going on a coin flip isn't worth it. Because by existing as army, they can stop SE armies from besieging their stuff, just by showing up.

That point is nothing that the AI considers. The AI calculates if a battle will be a clear victory, if not it will just avoid it (if it can). And that's the major problem with the AI: It constantly knows the exact party, army or garrison composition of every faction and if its own party/army can attack and surely beat it in autocalc.

That's because the autocalc is overly deterministic. If the autocalc had more variability in outcomes, this wouldn't be an issue. But since it's hundreds of dice rolls, the bigger number is almost always going to win.
 
Guys, i know that stuff, ok? :grin:
My point is, the world could be a little more risky for the AI. There is no point in let the enemy leader just walk away, because his numbers a little higher than yours. There is no point in ignoring each other, because some numbers change.
what happens next after the picture where made? Well, the Aseri decided to ignore the previously attacked city, with no militia in it and go for a siege, a castle, far away. the southern army go for some villages to refill whatever there needed.
That's just bad AI design.
 
That's just bad AI design.
It's not "bad AI Design". Write a better one yourself if you're really claiming this.

The AI is actually doing what it should. It's more a matter of tweaking and sorting out some bugs. And this is done with data and valuable feedback. "Bad AI design" does not help - i can tell you that much.
 
Guys, i know that stuff, ok? :grin:
My point is, the world could be a little more risky for the AI. There is no point in let the enemy leader just walk away, because his numbers a little higher than yours. There is no point in ignoring each other, because some numbers change.
what happens next after the picture where made? Well, the Aseri decided to ignore the previously attacked city, with no militia in it and go for a siege, a castle, far away. the southern army go for some villages to refill whatever there needed.
That's just bad AI design.
Why should the ai take risks? A lot of players would disagree and say that playing conservative is the right move. It doesn't sound like bad ai, more like you just don't approve of what decisions it's making and there is a big difference between the two.
 
Guys, i know that stuff, ok? :grin:
My point is, the world could be a little more risky for the AI. There is no point in let the enemy leader just walk away, because his numbers a little higher than yours. There is no point in ignoring each other, because some numbers change.
what happens next after the picture where made? Well, the Aseri decided to ignore the previously attacked city, with no militia in it and go for a siege, a castle, far away. the southern army go for some villages to refill whatever there needed.
That's just bad AI design.
It isn't bad design, it is because of how AI calculates numbers. If there is even 1 troop increase on that city AI may not attack it. Same for armies.In your situation SE wasn't going to attack Aserai army, they would just wait until Aserai attacks their city so they can have a number advantage. Aserai army knew they can't win because a nearly equal army is coming towards them, they lifted the siege and tried to attack them. But SE escaped, (again numbers) and Aserai probably didn't attack the city because troop number, maybe militia number increased while they were following SE army.
 
Last edited:
I wouldn't call it smart with how the player can exploit this behavior when the AI try to seize them. They're too binary in term of committing the force, to the point of feeling too zergy. Of course, the opposite of AI always commit to a fight when it's not possible to win isn't smart either, but neither one where they only fight when victory is certain is a good system, despite how it sounds. Again, the reason why the player can cheese and exploit the AI because it's 100% predictable.

They should have it work based on a threshold. Like a 10% disadvantage is considered an acceptable risk, then have the AI roll a dice based on this probability to see whether it will accept the risk or not. The battlesimulation also need some kind of probability dependency as well, not just straight up number crunch with all hard constant. There is a reason in pretty much all combat system have a "chance" or "roll" mechanism behind it, otherwise things become predictable, and predictable = begging for exploit.
 
Back
Top Bottom