Frustrated with the AI

[El_Condoro]

Warning: rant follows.

I am wondering when the 'intelligence' will be incorporated into the Artificial Intelligence of PzC. I have been working on a campaign and testing each scenario. When designing a scenario where the AI is attacking it is frustrating to have basically two choices to deal with the stupid things the AI will do - make the core and auxiliary units as weak as possible and/or make the AI ridiculously strong (many units, high experience, over-strength).

This is a list of things an attacking AI will do that make it weak, in no particular order:
- the order of attack is usually wrong. Of course situations change, but attacking an entrenched infantry in a city hex will be something like: ART, bomber, engineers, infantry, maybe finish off with a tank attack, occupy hex with infantry. Not the AI - go in with the tanks first!
- artillery fires last instead of (mostly) first
- (a real problem, mentioned above) AI insists on attacking fortified infantry with tanks! It may have artillery and infantry in great supply but it will send the tanks into that city first. Result, of course; tank slaughtered.
- will attack with infantry in half-tracks even if target is at full strength
- will leave infantry in transports adjacent to strong/entrenched units
- AI tanks will see ships as legitimate targets even if their objectives are elsewhere. In a test I am doing the AI objectives are south but the player navy is north - the AI has tied up 4 or more of its tank units taking pot shots at ships off shore!
- AI will move artillery and mobile AA units to the head of the attack.
- when the weather is bad and the spotting range is only 1 the AI will still act as though it is clear and will stumble into ambushes time after time. Even recons are doing this.
- when a flag is on offer to the AI it will take it even if it will mean certain death to the unit. Case in point, a strong SPAT cleared a city flag hex and was out of ammo, instead of retreating or getting an infantry unit to take the flag it went in and was destroyed in the player turn.

I am well aware that AIs are notoriously dumb (I played PG2 campaigns a lot) but these things really need to be addressed for campaigns. It's why I generally avoid AI games and focus on human vs human MP games. I think I read a while back that the AI was under review. Will thse things and others I have not mentioned with the AI be fixed?

[Razz1]

I posted all of this months ago. Have you been using the patch to see if any of this is fixed yet?

[El_Condoro]

Yep, designing and testing with 1.01.

[lordzimoa]

AI updating and testing will take a lot of time, to be honest the only reasonable big overhaul you can expect is with the Africa Expansion planned for later this year.

In patches we handle bugs and improvements, but the big overhauls with a lot of new content is reserved for the expansions.

We have planned ahead well into 2013 even, so be sure overtime the Panzer Corps series will expand and improve.

[EuroBoss]

AI updating and testing will take a lot of time, to be honest the only reasonable big overhaul you can expect is with the Africa Expansion planned for later this year.

In patches we handle bugs and improvements, but the big overhauls with a lot of new content is reserved for the expansions.

We have planned ahead well into 2013 even, so be sure overtime the Panzer Corps series will expand and improve.

Surely smaller AI updates maybe two in each patch rarther then a huge AI update would be a better option?

[VPaulus]

Surely smaller AI updates maybe two in each patch rarther then a huge AI update would be a better option?

That's the way BI does with ArmA series. Patch by patch, they improve the AI. That would be the ideal.
But even though, it takes a lot of time and Slitherine don't have BI's resources.
We've to be patient. The game in 2013 will be a better game, in all aspects, from what we've now.

[EuroBoss]

Surely smaller AI updates maybe two in each patch rarther then a huge AI update would be a better option?

That's the way BI does with ArmA series. Patch by patch, they improve the AI. That would be the ideal.
But even though, it takes a lot of time and Slitherine don't have BI's resources.
We've to be patient. The game in 2013 will be a better game, in all aspects, from what we've now.

I consider bugs and improvements to be highly relevant to the AI.I think it's slightly unfair to take the the tone of no AI update until we purchase a new add on way off in 2013
I feel at the very least there should be one AI update per patch

[VPaulus]

I consider bugs and improvements to be highly relevant to the AI.I think it's slightly unfair to take the the tone of no AI update until we purchase a new add on way off in 2013

All I said is that in 2013 the game will be better than what we've now. That doesn't mean that an AI patch (or more) isn't introduced in between.

But remember that with each tweaking the AI will eventually mean that the game must be balanced again.

[EuroBoss]

I consider bugs and improvements to be highly relevant to the AI.I think it's slightly unfair to take the the tone of no AI update until we purchase a new add on way off in 2013

All I said is that in 2013 the game will be better than what we've now. That doesn't mean that an AI patch (or more) isn't introduced in between.

Lets hope that's true maybe of the dev's can confirm this please?

[lordzimoa]

think it's slightly unfair to take the the tone of no AI update until we purchase a new add on way off in 2013

You misunderstand, I mean we will be constantly upgrading and improving the Panzer Corps series, engine and AI well into 2013. Sometimes with small steps like in a patch, sometimes with big overhauls like in expansions.

[EuroBoss]

think it's slightly unfair to take the the tone of no AI update until we purchase a new add on way off in 2013

You misunderstand, I mean we will be constantly upgrading and improving the Panzer Corps series, engine and AI well into 2013. Sometimes with small steps like in a patch, sometimes with big overhauls like in expansions.

Sorry about the misunderstanding.Anyway that's great you just made my day thank you.

[Rudankort]

Changing the AI in patches is possible, but there are two issues with this. First, I have seen a lot in the past that an "improvement" in the AI can generate new unforseen problems, and overall the AI can become worse rather than better. To avoid this, a lot of testing is needed which takes time and may not be feasible for every patch. Second, AI changes affect game balance. It can be argued that certain AI problems are obvious bugs and should be safe to address, but fixing them without changing the scenarios may render some existing scenarios (e. g. defensive ones like Bagration where the AI already has a big material advantage) completely unplayable. PGF is a good example of this, a lot of people could not beat Poland against the new AI, and this ruined the campaign and the game altogether for them. Once again, extensive testing is needed to avoid such issues.

These issues limit our options in providing quick AI patches.

[EuroBoss]

Changing the AI in patches is possible, but there are two issues with this. First, I have seen a lot in the past that an "improvement" in the AI can generate new unforseen problems, and overall the AI can become worse rather than better. To avoid this, a lot of testing is needed which takes time and may not be feasible for every patch. Second, AI changes affect game balance. It can be argued that certain AI problems are obvious bugs and should be safe to address, but fixing them without changing the scenarios may render some existing scenarios (e. g. defensive ones like Bagration where the AI already has a big material advantage) completely unplayable. PGF is a good example of this, a lot of people could not beat Poland against the new AI, and this ruined the campaign and the game altogether for them. Once again, extensive testing is needed to avoid such issues.

These issues limit our options in providing quick AI patches.

Whist i agree with 99% of what you just said there are some very basic things AI can't seem to do
for exmaple

Disembark from air transport

Know how to use air transports from land(embark)

Know how to use sea transports from land(embark when atacking

[El_Condoro]

Changing the AI in patches is possible, but there are two issues with this. First, I have seen a lot in the past that an "improvement" in the AI can generate new unforseen problems, and overall the AI can become worse rather than better. To avoid this, a lot of testing is needed which takes time and may not be feasible for every patch. Second, AI changes affect game balance. It can be argued that certain AI problems are obvious bugs and should be safe to address, but fixing them without changing the scenarios may render some existing scenarios (e. g. defensive ones like Bagration where the AI already has a big material advantage) completely unplayable. PGF is a good example of this, a lot of people could not beat Poland against the new AI, and this ruined the campaign and the game altogether for them. Once again, extensive testing is needed to avoid such issues.

These issues limit our options in providing quick AI patches.

While I understand that the campaigns and existing single-player scenarios would need to be modified for an improved AI, I see that as also an advantage of having an improved AI. This is my personal preference, of course, but I would sacrifice all the campaigns for a good AI that can be used to create more 'realistic' and challenging new campaigns than continue to see the things the current one does (and doesn't do, such as paratroops). If things continue to go the way I am already seeing, guys will be playing more and more custom stuff, anyway. Good challenging games are the bread and butter of this sort of game. I find it in MP, but there are a lot of guys who (only) play the AI and would like a fair fight.