The Good, the Bad and Ugly of the CiV AI

[apocalypse105]

BAD: AI denouncement hell destroys international relations - one they start flying its all war all-the time on every channel
UGLY: Inane demands (75% of your treasury please - from an ALLY?), illogical denouncements and negative reactions for helping others.


You just pointed the main problems of diplomacy in civ 5 good job!!!

The friendship and denouning system is a good system but it doens't work correctly because the AI denounce and declares war constantly as result its a suicide to get decleration of friendship

replies:

So the devs straight up lie to the buying public? That's disgraceful if it's true.

I have to agree with you but not in the way the AI cheats on prince like some users say above me :
you just get a lower happiness then he does (So basicly he doesn't cheat)

I do agree that they cheated about "diplomacy"
If you look at the reviews from civ 5 they say nothing about the dissapointed diplomacy they only give possitif reviews only a few comanies said that this was a major problem..


and iF YOU LOOK AT the gamebox And the trailers it says that you can conduct diploamcy but I never saw this in game The AI can't handle it So why say it?

 

[That nerdy kid]

what i think is stupid is that Helis have to embark to cross water tiles. they freaking FLY

 

[Babri]

AI's don't get bonus happiness on Prince, it is the human player that get LESS happiness. Go check out Handicap xml. AI's play on Chieftain so the happiness given for Chieftain player is 15. Human player on Prince according to Handicap file get only 9.

But the devs claimed that AI has no advantage over player in happiness which is wrong.

 

[GoodSarmatian]

GOOD: can exploit local and temporary weakness when attacking undefended colonies
BAD: ignores the big picture

Example:
I had good relations with Siam (as if that mans anything) and founded a city pretty far away from my continent after getting navigation. It was a very good spot and had access to three new luxury resources and good bonus resources: wheat, cows sheep and so on. I started to quickly developp it but had no military on the island safe for a token frigate and one garrisoned unit: a quite obsolete pikeman.
The city apparaently made such a juicy target that Siam decided to throw our relations out of the window (canceling resource trades and an ongoing RA) and executed a pretty good naval invasion with a lot of frigates, cannons, riflemen. It looked like they used half their army ton take the city in three turns after declaring war.
The bad thing ? Siam and me were direct neighbors and while they had just shipped half their army to the other end of the world to take the one weakly defnded city in my empire my army was still on our continent.
Three guesses how it turned out for them.

 

[headcase]

Love this write-up, will probably read entire thread

My $0.02 on each:

1) Seige
- In response to your suggestion, they should make other versions of the units that don't require iron, require the same techs, and are slightly weaker or take slightly more hammers. And improve AI.

2) Defense
- For sure, this is their strong point. I am disappointed it is so close to being good but doesn't understand the importance of weakinening powerful units. The problem may be that it doesn't understand positioning at all, and is made to assume that any free hex is infinitely valuable and takes any opportunity to free up hexes.

3) Navy hahahahaahah
- AI seems to use it (ineffectively) for recon, but doesn't understand the concept of attacking or keeping away from stuff that can shoot at it if it can help it.
- Much worse is embarked units. It has nooooo idea whether other civ's boats will be able to painlessly capture the unit in their turn. It should really "mark" the hexes that rival boats can reach and avoid them if possible. Considering it shows your units' range whenever you hold right-click, this should be possible to program without slowing down the game.

4) Responsiveness
- Yeah I agree, another AI strength for the most part, but that's not saying a lot. It might help to give each city a "minimum defense threshold" and only break this threshold when it's totally desperate.

5) air
- Again, in defense, it's killer with planes. Not genius, but it at least understands "I can do damage with this plane, let's use it" which is pretty strong for this game's AI. So yeah, besides focusing too much on certain units like you say, it's pretty good here (and should be).

6) diplo
- Yup... too big a subject for me to say "they should do X" in this thread.

7)
- Yeah... the AI should not cheat so rampantly on happiness. All it does it let them expand, causing runaway AI. It doesn't help them fight, gain money, or do anything else except after it's a runaway AI. Fighting against a poor AI that is 3 times bigger than you is not fun. (For some games, it's very fun, but Civ5 isn't that kind of game).

Even if they can't fix the AI, they should give it more science, more gold, more hammers, more apples, hell, I would give it more military output and even more free promotions before giving so much happiness

 

[headcase]

No. It. Does. Not.

There have been plenty of players who have posted screenshots of them obtaining similar levels of happiness as the AI on prince.

Start a game on Prince. Immediately press F9, look at Approval and tell me what you see. Unless they changed it from last patch, you'll see your civ in last place and all the other civs with much higher initial approval.

The devs (more accurately, their PR) lie about it because less cheating = higher quality AI.

Most people aren't fanatical enough to actually check all the cool stuff that's been posted on this thread (or to think to check demographics so early).

But whatever, I don't really mind when devs lie about that stuff. They're usually so busy on their job that they just kinda say what sounds best for the game and hope that's the way the game really is. No one's going to stop playing a game because they got it wrong.

 

[nokmirt]

I don't get how catapults and Trebuchets need iron, but Cannons don't. Definitely I think we need need one of catapults or Trebs to be iron-free, and then slap on an iron restriction on cannons, and that should balance it out better.

And it's really bad at using siege. My game, Genghis was at war with 2 city states. Against 1, it literally had only 1 cannon bombarding it. Using the cannon - great. But it never captured it since it didn't send a melee unit.

And against the other, it did capture it, since it literally filled up every tile within 2 squares of the city with archers, keshiks, and the like. So eventually it set them all up, and attacked with all and took it. But it was definitely a lot slower than it could have been with even 1 cannon.

Iron became more common and more readily available as time went on. Early on it was more scarce and limited. I suppose that is their reasoning for needing it for catapults and trebuchets.

There is one thing though. Russia during the reign of Peter the Great had a large scarcity of iron, at that point the Russians had not mined the Urals to any great extent. So what he did was to take down all the church bells in the land and cast them into cannon. For ships as well, he wanted to build a modern Western style navy, to protect his new trade routes from St. Petersburg. So UWHabs, you may be correct for cannon needing iron, some areas were not as keen to mining iron as others. Although Russia was backwards, way behind other major nations at that point in history.

What I think firaxis should do is create some kind of engineering iron improvement that a great engineer can build in a tile near a city, say and iron, siege foundry or something. The building has to be built on an iron source within a citiy's radius, and of course adds production when worked. It would alleviate the need for any siege units needing iron. The benefit to this would be say you have only a iron source that is only worth two iron, building the foundry would give you the ability to build siege units without penalty. Of course to keep it fair, other units that need iron would still have to pay the cost of one iron per unit.

There are plenty of good ideas civ players have come up with. But how many actually get put into the game? Probably not nearly as many as we would like. Not that this is a true viable good idea, but I like it!

 

[headcase]

What I think firaxis should do is create some kind of engineering iron improvement that a great engineer can build in a tile near a city, say and iron, siege foundry or something. The building has to be built on an iron source within a citiy's radius, and of course adds production when worked. It would alleviate the need for any siege units needing iron. The benefit to this would be say you have only a iron source that is only worth two iron, building the foundry would give you the ability to build siege units without penalty. Of course to keep it fair, other units that need iron would still have to pay the cost of one iron per unit.

(sorry for the cluster of posts)

The GE's Manufactory might be able to double strategic resources on that tile, though this would make the GE even more powerful! Maybe the Trade Merchant's improvement can do it instead?

Another solution could be to have certain later techs reveal additional sources of iron on the map, if that doesn't put too much luck into the game.

 

[Bezal]

Being a new Civ player, I thought I'd search out a few forums to try and understand the logic behind the AI system. I was hoping to find a solution to something simple I was doing wrong, but it seems, from what I've seen here, its at worst: broken, and at best: fanatically and frustratingly hell bent on beating up the leading player?

Every time I play it feels like as soon as I start to push my nose out in front of the competition, the AI comes in with a whirlwind of denouncements and declarations of war.

Why oh why would Firaxis do this? The only reason I can come up with is its a designer's very clumsy attempt at making the game more challenging - but with an awful opportunity cost: it just feels entirely unrealistic and ridiculously illogical. What's the point in a cultural or diplomatic victory option if 90% of the time I'm forced into a very paranoid, aggressive and one dimensional play style?

 

[sleepysteve]

Why oh why would Firaxis do this? The only reason I can come up with is its a designer's very clumsy attempt at making the game more challenging - but with an awful opportunity cost: it just feels entirely unrealistic and ridiculously illogical. What's the point in a cultural or diplomatic victory option if 90% of the time I'm forced into a very paranoid, aggressive and one dimensional play style?

couldn't have said it better my friend

 

[Putmalk]

I wouldn't even take Jon Shafer's word for it anymore, he doesn't even work for Firaxis anymore, does he?

Can anyone propose a solution to make the AI more challenging without massive bonuses, however? I'm curious.

 

[headcase]

I wouldn't even take Jon Shafer's word for it anymore, he doesn't even work for Firaxis anymore, does he?

Can anyone propose a solution to make the AI more challenging without massive bonuses, however? I'm curious.

There are some complicated things that will take a long time to code in.

Then there's easy stuff, like identifying the value of Mints, shooting at things with boats, knowing when a move is pointless suicide, etc.

If and when Civ5 is opened up entirely, there will be modders attacking both sides of the problem.

 

[nokmirt]

(sorry for the cluster of posts)

The GE's Manufactory might be able to double strategic resources on that tile, though this would make the GE even more powerful! Maybe the Trade Merchant's improvement can do it instead?

Another solution could be to have certain later techs reveal additional sources of iron on the map, if that doesn't put too much luck into the game.

Those are good ideas as well, especially having it become more abundant as time goes on.

There are some complicated things that will take a long time to code in.

Then there's easy stuff, like identifying the value of Mints, shooting at things with boats, knowing when a move is pointless suicide, etc.

If and when Civ5 is opened up entirely, there will be modders attacking both sides of the problem.

Yes indeed! BTW has anyone heard anything about when the .dll will be released?

 

[Bezal]

couldn't have said it better my friend

Would love to see a losing AI player settle for second best, and cosy-up to a leading civ with similar social policies, instead of shooting for an 'all or nothing'/'beat the human player or die trying' approach'.

Counter-intuitive as it sounds, it feels too much like the AI is trying to win a game of Civ and not enough like they're attempting to run a civilization.

 

[That nerdy kid]

this is true. In Total war (at least in empire) the AI isn't trying to win, just stay alive. the USA doesn't have the most land area, nor the most people. However it does have highest GNP, most powerful military (Either USA or Russia, Russia has more nukes and weapons, but USA has more soldiers, and available soldiers). Yet if one were to say that China was the worlds #1 civ, the USA would not attack, send all of its nukes, send a whole bunch of planes, bombers, marines, etc. and take it over/ die as it lost all of its cheap manufacturing/ have every other country declared war on it. That would not be a good choice. a Civ doesn't need to be #1. they just have to be alive and happy.

 

[Glassmage]

this is true. In Total war (at least in empire) the AI isn't trying to win, just stay alive. the USA doesn't have the most land area, nor the most people. However it does have highest GNP, most powerful military (Either USA or Russia, Russia has more nukes and weapons, but USA has more soldiers, and available soldiers). Yet if one were to say that China was the worlds #1 civ, the USA would not attack, send all of its nukes, send a whole bunch of planes, bombers, marines, etc. and take it over/ die as it lost all of its cheap manufacturing/ have every other country declared war on it. That would not be a good choice. a Civ doesn't need to be #1. they just have to be alive and happy.

alive and happy doesn't win the game. o.O

 

[headcase]

I like that the AI goes for victory. I'd argue that the game should continue after non-dom victories so other players can get 2nd, 3rd, 4th, etc instead of DEFEAT. Then the AI could be programmed to be "well that civ's getting the rocket victory 1st, but if I keep pushing culture I might get 2nd" instead of desperate warring. Most human players would still go to war though.

If tech, spaceship parts, and SPs were hidden, well that would change the endgame quite a bit.

 

[Talim [own]]

After a new game post patch I decided to sum up a few AI observations

1) The AI builds siege units
GOOD: It builds artillery and Cannon

Rat

I'm not sure if '' making a unit '' should be considered as '' good ''. =p

But I understand OP, it is very hard to find good things to say about the AI

 

[sleepysteve]

Would love to see a losing AI player settle for second best, and cosy-up to a leading civ with similar social policies, instead of shooting for an 'all or nothing'/'beat the human player or die trying' approach'.

Counter-intuitive as it sounds, it feels too much like the AI is trying to win a game of Civ and not enough like they're attempting to run a civilization.

Agreed, it really hurts the enjoyablility of the game to know that even before you find the other civs they are conspiring against you to "win the game"

 

[qemist]

Agreed, it really hurts the enjoyablility of the game to know that even before you find the other civs they are conspiring against you to "win the game"

Playing to win is not the same thing as "consipiring against you".