Spells and the AI - not a priority?

[Pocus]

Hi,

Not wanting to disparage this fabulous work, but I fail to understand why the AI is unable to use even the most basics spells routinely, which are of the utmost importance, like Spring or Enchant Blade. Well I do understand in fact, it is a question of priorities, and new features like Forts (and their positioning by the AI) are deemed more important compared to teaching it how to use the basics of cantrips. I question this priority*, sorry ...

The AI adepts are inepts


*: The AI is even weaker on this aspect compared to FFH versions I played one year ago!

 

[xienwolf]

Actually there is no effort to teach the AI forts either. Or anything.

Some of us players are trying to fiddle with things and improve the AI a little bit with the tools at hand, but for the team right now the main goal is to make sure that the game works for the Humans. When we manage to find ways to make the AI considerably smarter with some simple code then it tends to find a way into the next patch.

So, crack open that code and play around Teach the AI a trick or two and post your results

 

[Kael]

I actually spent a lot of time makign sure the Ai uses forts in a somewhat helpful manner. For example the ai will prefer moving through tiles that have some fort coverage (just as it prefers ending in tiles with forests and such). It also tends to move units into forts to guard.

With that being said there are a ton of things the AI considers and it isnt hard to come up with any specific situations where the Ai does A when B would be better. We have to program the AI for generalities and then hope for the best in gameplay, Civ4 and FfH is way to open to get more specific than that. As long as the "less efficient" choice isnt disbanding the units or abandoning cities or other unreasonable options we are doing okay. We will never see an AI that plays as well as a human, though we will hopefulyl continue to get better.

As for spells the entire spell system was rewritten to make spell castign easier for the Ai to understand. In my opinion it is way ahead of where it used to be (the AI actually summons effective monsters against you and casts damage spells like ring of flames, etc). As stated above it isnt always perfect the specific situation but it does a reasonable job. The other option is to cheat and allow the ai to enchant any unit in the empire or spring any tile in their lands without going to it.

We will need to work on some specific code for some spells like Spring and Enchanted Blade. Currently the Ai does cast them when its in a place to be able to use them. But it won't move to a place to be able to cast them (so a unit that could use an enchanted blade wont go to the caster, or the caster wont go to a desert tile to cast spring on it). Im hoping to be able to add some thing along those lines before we are done.

 

[Pocus]

Thanks for your answer. I would prefer to have an option which enable the AI to use non-offensive spells at distance (if 3 adepts have Spring, then 3 deserts can be converted or 3 flames can be quenched a turn), as it would offer more challenges, strategies and interests for the crowd of solo-players, but it is you to judge.

AI is AI. I don't expect it to play better than a poor player actually...

 

[TheJopa]

About the casting AI- main trouble is that AI simply rarely builds mages and conjurers, and I don't think I have ever seen Archmage or Summoner. Probably it's the level requirement what cripples them.

 

[xienwolf]

It was recently discovered that the Weighting Factor of the Upgrades were set to well under that of the basic Adept, and in playtesting it seems that this does indeed prevent the AI from upgrading their Adepts.

I just can't remember if we ever actually posted the results in the Bug Thread...

But increasing the AiWeight factor on Conjurers & Mages will result in you starting to see a fair number of them on the field from certain civilizations.

 

[Nimbus]

It was recently discovered that the Weighting Factor of the Upgrades were set to well under that of the basic Adept, and in playtesting it seems that this does indeed prevent the AI from upgrading their Adepts.

I just can't remember if we ever actually posted the results in the Bug Thread...

But increasing the AiWeight factor on Conjurers & Mages will result in you starting to see a fair number of them on the field from certain civilizations.

No, I don't believe any of us testing it actually posted about it in the bug thread, just assumed Kael would read the thread at his leisure and judge the results for himself. could probably use to mention the guild/elf promotion bug again too, before the next patch hits.

 

[sylvain5477]

I thougth the IA focus was to be in the last phase (Ice?) of the mod ?

 

[BeefontheBone]

That was partly true - in short (as you can see from the Wiser Orcs thread, which I recommend the OP checks out) it's a pain to program a specific set of behaviour for the AI only to then take out or alter the relevant content of the mod and waste that work. Once the features are all in, that ceases to be an issue.

 

[Bad Player]

Can you tell the AI to do specific things on specific tiles in a scenario that you make?

E.g. If you design a choke point can you tell the AI to try and camp there with defensive units?

E.g. If there is a desert square that acts as a soft spot (i.e. 25% defense penalty) for an invading army (e.g. at a choke point) can you tell the AI not to spring it if it is in their territory (which would go against their natural instinct to improve the land)?

N.B. Im only talking about a map that you make and tell the AI to act a certain way on certain tiles in your map.