Lets list AI stupid behaviors

I don't see anything in the score list that ties directly to victory progress, whether attached SS parts, number of civs you have cultural influence over, or number of capitals controlled (except to the extent conquering affects the cities, population and land elements of your score). I also don't see anything in GlobalDefines.xml.

Spoiler :


<!-- Civ 5 Score -->
-<Row Name="SCORE_CITY_MULTIPLIER">
<Value>8</Value>
</Row>
-<Row Name="SCORE_POPULATION_MULTIPLIER">
<Value>4</Value>
</Row>
-<Row Name="SCORE_LAND_MULTIPLIER">
<Value>1</Value>
</Row>
-<Row Name="SCORE_WONDER_MULTIPLIER">
<Value>25</Value>
</Row>
-<Row Name="SCORE_TECH_MULTIPLIER">
<Value>4</Value>
</Row>
-<Row Name="SCORE_FUTURE_TECH_MULTIPLIER">
<Value>10</Value>
</Row>
-<Row Name="SCORE_POLICY_MULTIPLIER">
<Value>4</Value>
</Row>
-<Row Name="SCORE_GREAT_WORK_MULTIPLIER">
<Value>4</Value>
</Row>
-<Row Name="SCORE_BELIEF_MULTIPLIER">
<Value>20</Value>
</Row>
-<Row Name="SCORE_RELIGION_CITIES_MULTIPLIER">
<Value>1</Value>
</Row>
 

Attachments

  • Score.jpg
    Score.jpg
    11.9 KB · Views: 986
As far as I have explored into the dll, code checks for overlap of selected victories, not for progress towards one. In other words, if AI's "guess" (literally, they have a guessing function built in to avoid blatant cheating) is that the player is pursuing the same victory condition as the AI's, then AI increases chances of changing stance to more aggressive ones.

That's it. AI is not aware of general progress towards a victory. Something for the future, if time permits.

AI plays to win in the sense that it selects a VC and adjusts general strategy to it (and VC's are dynamic, if game environment changes, AI may change VC), but it does not play to BLOCK OTHER PLAYERS from winning.
 
They don't build enough workers
What few they build spend most of their time walking because it doesn't consider the proximity of a worker (enough? at all?) when choosing which tiles to improve
 
Didn't get into how the AI manage victory progress, but yeah, seems it only gets upset if you are ahead on them in some winning strategies.

Yesterday I have gone into several little issues I've seen on the core dll:

The worst case is the great scientist directive: In order to build an academy:

- Need to be pursuing a science vic as his grand strategy.
- The turns passed in the game should be less than one cuarter of the game estimated turns (I believe is 500 on all standard settings).

As the AI don't even consider science vic until much later in the game, the result is AI will rarely ever build an academy.

In the cultural great personal there's a little issue of: I cant' don't have a work slot? lets consume the GP right now.

The great prophet by design will near always enhance first and go convert cities later.

On heal orders, seems is always an option if the unit didn't take part on an attack and health is under 100%.

There's an action of move to safest plot, with doesn't take into acount if unit will embark as a result.

Most of them are issued. I'll publish a dll beta once I make a little more testing, would be great if other people help me to test if the dll mod cause some performance issues/crashes/unexpected new behaviours.
 
AI plays to win in the sense that it selects a VC and adjusts general strategy to it (and VC's are dynamic, if game environment changes, AI may change VC), but it does not play to BLOCK OTHER PLAYERS from winning.

ah, I've been mistaken all these years then. I "learned" from someone that you should not hook up the spaceship parts until you had them all built and I've always been doing it that way. good to know that's a waste.
 
Can anyone confirm (or bust) the myth that the AI reacts when it sees the "Player 1 has completed SS Booster!" pop-up like it does when you build a wonder they coveted?
 
Random diplomacy pop ups telling you that you're vermin...:lol:
 
When an AI asks for a research agreement and doesn't want to pay the costs of getting a research agreement. AIs on the other hand, ask for research agreements and want the players to pay and contribute for the research agreement. There simply is no way to get the AI to pay for its research agreement making the AI really stupid since the AI doesn't want to make a research agreement in the first place.
 
This is not really a flaw, but it would be an easy thing to do to make the game harder: alert the AI when you build a spaceship part, instead of just when you hook it up to the spaceship. Right now everyone knows to just stock up all the pieces and hook them all up on your last turn. If the AI was alerted as each piece was built, maybe some of them would launch a last ditch war against you to prevent you from winning. Plus it's annoying to store all your pieces around the capital to prevent the AI from learning your plans, so this is a win-win.

You aren't alerted when the AI builds a spaceship part, only when they hook it up. So why should they be alerted when you build one?

The really weird thing is the AI valuing GPT over lump sums for some reason. I mean that the AI will offer you less total gold if it's a GPT deal instead of a lump sum, am I right? Now why is that? Shouldn't it be the other way around, if anything? I don't get it, so I must be missing something. :confused:
 
Nice to see some real AI improvements rather than just stacking more cheasy bonuses to make the game "harder" :goodjob:

One stupidity i think i havn't seen mentionned (didn't read all 5 pages but i read the list in opening post). AI doesn't seem to understand when a war is non-profitable. Just because they have an army 10x as strong as mine in their territory doesn't mean they can reach me past that 2 hills wide bottleneck with forts and sending waves after waves of troops to my "training field" won't help them. Actually, it will hurt them if they keep producing those troops. However, they won't discuss peace if i don't offer them 1/2 my cities :badcomp:

From what i understand AI does take casualities into account to weight out the "value" of a war but it seems like sometimes the wieght of the available troops is just so high that it won't understand it can't possibly win.

Of course, finding the correct balance might not be easy as you don't want to have AIs surrender after loosing a few troops.

This is probably part of a larger issue of understanding the strenght of various civs. It would also be nice if AIs could take the result of previous wars into account when determining the strenght of an army. If my much smaller army have already beaten the crap out of 2 civs armies, they should definitely take me seriously. And the coward that serves as my military advisor shouldn't beg me to make peace with my neighbour just because it has twice as many men as i have. Size doesn't always matter.
 
AI still lose too much settlers due to barbs.

It looks to me like the cause is the AI doesn't know how to escort embarked settlers with an embarked military unit escort in front of it and as a result the naked settler lands and is immediately captured by the barbs.

So I'd focus on logic dealing with crossing landmasses / while embarked
 
Not sure if this has been mentioned or would even count as stupid behavior: AI making peace one turn before taking a city.
 
AI valuing crappy cities way too highly.

This just happened: (granted I was being ******ed, but it was hotseat and I was much more interested in the Zulu I was playing up North)

- playing Indonesia, stopped to build hanging gardens before expanding
- Aztecs city-spammed 8 cities locking me in a corner by turn 90-100
- Aztecs attack my capital (jealous of wonders)
- My defense? Impossible against to massive hordes
- What do I do? Why produce a settler. Why not? Send it over to the desert and settle.
- Aztecs ignore it as they are busy smashing my capital
- I settle in the crappy desert then open diplomacy screen and offer them the worthless city
- Aztecs accept, and happily burn it to the ground
- I have 10 turns to rush swordsmen and walls to bolster my defense

...what idiots...
 
Another case of weird diplomacy from my recent game:

1. Pocatello declares war on Isabella (3 cities total).
2. Pocatello conquers Madrid.
3. Isabella recovers Madrid.
4. Just a few turns later both sign peace treaty.
5. IMMEDIATELY after that both make DoF.
 
I had a recent game where Venice had 3 whales that they left unimproved all game. They bought whales off of me instead. Egypt had a coastal city with whales and fish in range. They improved the whales but not the fish. This might be related to their general backwardness in improving tiles generally, and not having enough workers.
 
I was very very close friend of Maria. Then suddenly, she declared war against me with Hiawatha in same turn. I defended my land successfully and captured and razed a few cities of Maria. Then she offered me peace and also tribute. I accepted and she became friendly after we made peace and she never betrayed me again.

The stupidity I've never understand, why she declared war on me while we were very very close friends.
 
I was very very close friend of Maria. Then suddenly, she declared war against me with Hiawatha in same turn. I defended my land successfully and captured and razed a few cities of Maria. Then she offered me peace and also tribute. I accepted and she became friendly after we made peace and she never betrayed me again.

The stupidity I've never understand, why she declared war on me while we were very very close friends.

Hiawatha made an offer she couldn't refuse.
 
Top Bottom