How the AI could reliably win Domination and Religion

King Phaedron

Warlord
Joined
Oct 9, 2017
Messages
174
As it stands, the AI can only win with Science Score, or Diplomacy (Is it even possible for them to win with culture? I've gone over it a dozen times, and I just don't get the culture victory at all. I understood how it worked in Civ5, but for Civ6, I just don't get it. It seems impossible.)

The key is for them to mutually advance winning conditions. Lets say that Mongolia is living next to Georgia. Georgia converts all of Mongolian's cities to her Religion, but Mongolia captures her capital. AI Georgia puts up only the minimum effort to defend the capital, because what the Computer wants to do as a whole is advance victory conditions mutually against the players.

Now lets assume this is happening everywhere. One Civ will be chosen that has an impressive capital, such as one that is on land, but next to a lake, can garrison with both a ranged unit and a ranged naval unit, and is governed by Victor with his double city attack promotion. Also has the highest defense of any city in the game. That civilization will create a very large army to defend this final capital.

Here's the kicker: Once a player declares war on that Civilization at least one other Civ is standing by to retake their original capital, in case the player succeeds. The player should move their troops into position, and not declare the war until they are right on the borders, also lying about moving the troops if need be. The player might be able to take the capital before the AI can reclaim one of theirs, but probably not. The domination player really ought to just take every capital himself, and the AI cheat if you try to cheese this.

Civs who would like to win Domination, should occassionally build up some troops somewhere out of sight, but not far from players capitals. If there is only one capital left that hasn't been taken, at least one AI needs to be desperately attacking and sending units to your capital. The AI should get production bonuses for units when attempting this, as the AI plays with an expendible unit strategy whereas players tend to keep theirs.

For Religion victory, as the game progresses, the computer needs to chose a champion, and just give all of their new Apostles the Proselyzer promotion, so that they can begin to reliably convert civilizations with other religions. I want to have a game where someone has converted the world, and everything hinges on my ability to keep their religion out of my cities.

In the late game, AI should aggressively use Indie Rock Bands to turn a players capital, after creating conditions where the players capital and one other are the last two remaining. At that point, you either card music censorship or lose the game.

No matter how challenging or bad a game of Civ6 may seem, stay with it long enough and stop the AI from winning the science victory and you will surely be able to eventually win. In my experience, that's all you really need to worry about. the AI can only win with Science, Score, or Diplomatic victory.
 
The AI can also win Religious Victory on smaller maps especially on Pangaea. I have never seen the AI win a Cultural Victory, I agree.
 
I dont think religion is a problem for the AI... I however think the religious habilities should cost faith to use, and the charge system should be discarded... among other changes. But the AI does not do bad in religion, i just dislike the entire system.

Regarding war, the problem is a combination of factors:

The AI is not good enough managing units to take reliably cities. Also it gives up too easily on wars, and plans very badly where and when to atack. They should attack weak cities that are vulnerable and they can keep, and should continue expansion after taking one city to secure loyality.

Also late game agression is non existent, and I think that; because the devs though wars should not exist on modernity, they did not care to implement proper air, anti-air or naval operations.

The irony of turning the AI in social justice politicians in modern times while, at the same time, modernity is when they added more military units and military mechanics, would be hilarious if it was not a tragedy for the game.

Besides that, it is a matter of expanding the AI with more and better strategies for different cases, and improve some bad decissions, as I think the devs are already doing. They should just use some extra man hours to fix those problems, and that should improve most victory conditions. However, if they dont do some changes in diplomacy and AI agression, it will not improve much in the case of domination, which is a big stain in the current state of the game.

Is it funny to play domination against an enemy who is incapable of winning? Why the devs dont address this is beyond me.
 
Last edited:
I'd have them work on early military strategies before late ones. By the time aircraft show up, any human who's trying to expand militarily will be invincible. Although it would be nice to have a credible late AI threat against a human player who's winning some other way.
 
I'd have them work on early military strategies before late ones. By the time aircraft show up, any human who's trying to expand militarily will be invincible. Although it would be nice to have a credible late AI threat against a human player who's winning some other way.

Well I mean, you mean basic pathmaking, city sieging and unit management, cuting suply and reinforcement lines.... Those are not early tactics only. But i get what you mean.

While those can be improved, the game is not so lacking there. While at the moment the AI just is uncapable of handling air units, anti-air units, submatines, carriers, nukes... Which ruins the inmersion completely.

So to me, it is far more important that the game can handle late wargame, than make it slightly better at early war game.

That said, one should not detract from the other. However, the problem with the AI in early game is not that is bad with tactics. It is fairly good at defense. But it is uncapable of taking cities, and even worse, is uncapable of keeping cities when by some miracle it does because it goes for cities that cannot keep.

I think the priority would be in this order: 1) to make the AI more aggressive opportunistic, and constant 2) to provide support for the units and mechanics it does not handle, and finally 3) to improve coordination, pathmaking, cycling of units and advanced tactics.

However, none of this really matters. The game is 4 years old and around 15 to 20 patches, and we have only seen small incremental changes, with no intention of tuning agression or AI goals, or even real attempts of handling moddern war units and mechanics. And dont think we will. The only thing the devs have paid atention to are basic war and naval tactics, and some minor basic use of ir units.
 
A reflexion, maybe Fxs has taken away the replay movie at the end, cause we would see how bad and passive AI civs are if they hadn't.

Just imagime, you see the entire history of the world and you dont see a single succesful war in the entire world... ever... in any game...

This would be the complete account of 3000 years of military history.

Beginning of the military account of the world history:

- Year 700, Roma takes the city of Citystan.
- Year 705, Roma loses the city of Citystan due to loyality presure.

End of the military account of the world history.
 
Last edited:
AI is capable to win culture victory, most likely China in a game without Kongo, Rome, Greece, Russia (and alike). I had a deity game when I finished pretty late due to start position - completely surrounded by mountain range with 2 tiles to work (t300 or so) and China was sth like 10 turns to win when my spaceship secured my victory. And because it was my culture blocking Qin, I bet some civs are really able to win CV between t250-300
 
AI does well at religious victory ... well, no ... religious victory is the victory where the AI performs least badly, because it's basically just a question of spamming Apostles. In fact, the prime reason why the AI seems to fail at religious victory in most cases is because at some points Apostles becomes so expensive the AI changes to Missionaries who will then be killed and cause a lot of negative pressure, thus eliminating whatever progress the AI must have done earlier.

AI is effectively blocked from winning domination victory, not only by its poor pathing and tactical skills, but by the design of city walls and city defense scaling passively with unit and technology progress. The fact that any contemporary melee unit will lose a big part of its health and can easily be finished off after just one attack into a walled city obviously makes it a pretty monumental task to take a fully walled city even for a human player. Normal ranged units do next to no damage to walls, at least from midgame, and siege units suffer from being sitting ducks while they set up, again being decimated by the powerful ranged attack from the city. AI would probably perform much better if cities were much less deadly on their own and relied more on units for their support, as well as if AI learned to fortify melee units around a city while walls are taken down by range and siege units, instead of mindlessly suiciding its entire army against the walls in the first few turns. Also, as has been pointed out by others also, we really need a rule change that lets ranged units capture at city that is reduced to 0 health. The fact that you can have a city sit unclaimed at 0 health surrounded by many enemy ranged units because all the melee units have been killed is just downright stupid.
 
Just a response to a repeated comment - the AI is capable of a Culture Victory. The very first game I played I lost to a CV. I recently had to wipe my game so I couldn't tell you which Civ won it, but it pulled the rug from under my feet, not log before I was going to win a Science Victory.
 
In fact, the prime reason why the AI seems to fail at religious victory in most cases is because at some points Apostles becomes so expensive the AI changes to Missionaries who will then be killed and cause a lot of negative pressure, thus eliminating whatever progress the AI must have done earlier.

The fact that any contemporary melee unit will lose a big part of its health and can easily be finished off after just one attack into a walled city obviously makes it a pretty monumental task to take a fully walled city even for a human player

Exactly this
 
AI is effectively blocked from winning domination victory, not only by its poor pathing and tactical skills, but by the design of city walls and city defense scaling passively with unit and technology progress. The fact that any contemporary melee unit will lose a big part of its health and can easily be finished off after just one attack into a walled city obviously makes it a pretty monumental task to take a fully walled city even for a human player. Normal ranged units do next to no damage to walls, at least from midgame, and siege units suffer from being sitting ducks while they set up, again being decimated by the powerful ranged attack from the city. AI would probably perform much better if cities were much less deadly on their own and relied more on units for their support, as well as if AI learned to fortify melee units around a city while walls are taken down by range and siege units, instead of mindlessly suiciding its entire army against the walls in the first few turns. Also, as has been pointed out by others also, we really need a rule change that lets ranged units capture at city that is reduced to 0 health. The fact that you can have a city sit unclaimed at 0 health surrounded by many enemy ranged units because all the melee units have been killed is just downright stupid.

I agree heavily with this point, and the strength of city walls is something I believe Firaxis should consider changing.
The difference between an unwalled city and just ancient walls is immense, and it's not just the AI that struggles with conquering walled cities.
As a human I myself too struggle with walled cities once the ancient walls come up, up until Flight gets researched (for Balloons and later on, Bombers) in the Industrial Age.
If I'm on the defensive, investing in Encampments and/or forts is almost redundant, compared to the very cheap investment into ancient walls - the return on investment for Ancient Walls is just ridiculous in terms of survivability and the AI can frequently lose their entire army as long as I have said walls and a high tech unit for city combat strength (even if it's on the other side of the world).
This also makes domination kind of boring for me these days, as I feel pigeon-holed into conquering as much as I can in the ancient/classical Era, before the AI walls it's cities.
Then it's usually a veeeery slow advance (or even stalemate), until I absolutely decimate the AI again once I tech flight.

I think Firaxis should seriously consider going back to the Civ 5-model for city defenses - have them have a base ranged attack to fend off early rushes, but not let them be the massive fortresses they are once the first ancient walls come up and city strength scaling accordingly once the first Knight (or whatever else the highest current tech unit is) is produced.
In civ 5, runaway domination AIs were a LOT more frequent, which made things interesting regardless of if I was defending or trying to conquer them myself.
Otherwise I'd love to see some 3-range bombard-class units again (which you usually got in civ 5 through unit levelups), but mostly I'd prefer the strength of walls to be reduced.

(Case in point: On Emperor difficulty the AI usually conquers neighbouring city states very quickly since they don't have walls, while on Deity the same city states live on a lot longer since walls come up immediately.)
 
I don't know if the AI actively tries to build Encampments to have overlapping fields of fire with city centers and other Encampments (I suspect not, and it's just a byproduct of the AI favoring Encampments and tightly-packed city planning in general), but in practice, it does so nicely. It's not super easy for even a good human player to take a defended city while using and protecting the artillery that's necessary for the task (the AI is also rather superb at targeting siege units.)

But the AI needs to get at least a little bit good at taking cities on the offensive, or perhaps have alternative tactics when they don't have the right unit mix. I've seen lots of Civs on the offensive that squander decent horse units in endless replays of the Charge of the Light Brigade, for zero value. It'd be nice if the AI could recognize, e.g., that an army with 12 cavalry units, one musketman, and zero ranged units won't be able to take cities, but could pillage the hell out of the enemy while building a proper combined-arms force
 
I don't know if the AI actively tries to build Encampments to have overlapping fields of fire with city centers and other Encampments (I suspect not, and it's just a byproduct of the AI favoring Encampments and tightly-packed city planning in general), but in practice, it does so nicely. It's not super easy for even a good human player to take a defended city while using and protecting the artillery that's necessary for the task (the AI is also rather superb at targeting siege units.)

But the AI needs to get at least a little bit good at taking cities on the offensive, or perhaps have alternative tactics when they don't have the right unit mix. I've seen lots of Civs on the offensive that squander decent horse units in endless replays of the Charge of the Light Brigade, for zero value. It'd be nice if the AI could recognize, e.g., that an army with 12 cavalry units, one musketman, and zero ranged units won't be able to take cities, but could pillage the hell out of the enemy while building a proper combined-arms force
I agree, although the AI does seem reasonably capable at placing encampments on the side of cities facing towards other civilizations. This, combined with their general unconditional love for spamming encampments, is definitely a big help for them in staying alive.

Also, while we're on the topic of encampments, if you build an encampment facing towards the AI, the AI seems programmed to target the encampment BEFORE it takes on the city. This is itself perhaps not a bad strategy, if not again for the fact that the AI will mindlessly throw melee units against a walled encampment chipping away perhaps 2 % of its health, while the melee unit loses anywhere between 50 and 90 % of its health in said attack.
 
Is it possible to modify Deity AI damage against city walls? Similar to the increased combat strength but only against walls? Of course, it's not an optimal solution and it's bad design but I can't think of a more effective one. Nerfing walls in general would make war too easy for the player again. I don't think anyone wants to go back to vanilla horse men rushes.
 
Maybe it is just the way i play, but i find the AI to be competitive in CV. Smaller more compact maps favor the AI in RV. Can't really blame them either. Moving religious units across the map is very tedious.

Setting aside how aggressive the AI is at declaring wars, there are a number of factors that hurt the AI in the late game when it comes to war compared to earlier eras.

The vast majority of late game troops not only take resources to make, they require upkeep. Some of these resources are also needed to keep your factories going. Neither upkeep or resources are required in the early to mid game.

Troops are more experienced, and combined with a players ability to preserve them better throughout the game, will get an advantage here.

Production costs do not scale well with tech pacing making the above issue worse.

Players make more money and manage it better so they can upgrade their troops. Not to mention better at dealing with the RNG system for resource spawning.

There are policy cards that are helpful in numerous ways to warfare, and i doubt the AI does a good job with these.

Encampments will do a far better job at slowing down an Ai than a player. Ai often face plants into them.

Combat strength differences due to a number of abilities, tech, and formations/armies will hurt an AI more than a player.

Humans will do a better job dealing with loyalty issues than the AI.

The requirement to say make Air units in the late game is a tough one due to all the steps required, assuming you can even make the units in the first place.

edit removed rant about FX devs and making the AI use planes. At least now the AI might use their planes.


Now how do you lower the gap or fix some of these? Easy solution is more cheats, but those aren't that fun to deal with as a player. Giving cheats to upgrading troops and experienced gain rather than say combat boosts are preferable to me. Improving the logic on say policy cards will help. Fixing the pacing would help and improving catch up mechanics. Making encampments not as powerful for defense but giving them some other boosts would be nice. Reducing the requirements to even make air units would let them be used more often.

Changing core mechanics is another route, but more expensive in time and resources. Take strategic resources for example. I would prefer if the need x to build something was removed and replaced with x resource provides some buff to your city, troops, and civ wide. That way exploring and settling on or near important resources still matters while not making entire civs irrelevant for large portions of the game. Another option is have buildings provide some resources.
 
Maybe it is just the way i play, but i find the AI to be competitive in CV. Smaller more compact maps favor the AI in RV. Can't really blame them either. Moving religious units across the map is very tedious.

Setting aside how aggressive the AI is at declaring wars, there are a number of factors that hurt the AI in the late game when it comes to war compared to earlier eras.

The vast majority of late game troops not only take resources to make, they require upkeep. Some of these resources are also needed to keep your factories going. Neither upkeep or resources are required in the early to mid game.

The recent change that makes strategic resources much more rare is a related factor. Sources of oil and aluminum can be so limited that the AI essentially can't make any good modern units. And while scarcity causing wars for oil isn't a bad idea in theory, the human player is far more likely to be able to plan and win these wars than the AI. In a recent game, I fought someone who had a large army, but it was something like 20 Field Cannon units and nothing else.

Troops are more experienced, and combined with a players ability to preserve them better throughout the game, will get an advantage here.

I've actually found the opposite to be true. The AI fights a lot of pointless wars, and loses a lot of units... but the survivors tend to have a lot of promotions.
 
In my opinion there is a serious flaw in the diplomacy system that makes it very difficult for an AI to win a domination victory and that is that the human player has it very easy with befriending AIs. In earlier iterations of civ you could never be certain if an AI would declare war on you and that slowed down the play a bit for the human player. Meanwhile it also made room for other AIs to keep up with the human player, when the human player was in war.

I don't think civ ever had a good AI to actual complete a domination victory, but what's more important for me is that the AI is as at least capable of PREVENTING a human player to get a domination victory.

Of all the civs I played civ iv did the best attempt of really competing AIs.

In short, the AI needs to be way more aggressive throughout the whole game. And this is even coming from a player that usually likes the pacifist game but it's not really a fun play anymore since it's so easy to befriend everyone. The human player needs much more punishment when neglecting Army maintenance.

A reflexion, maybe Fxs has taken away the replay movie at the end, cause we would see how bad and passive AI civs are if they hadn't.

Just imagime, you see the entire history of the world and you dont see a single succesful war in the entire world... ever... in any game...

This would be the complete account of 3000 years of military history.

Beginning of the military account of the world history:

- Year 700, Roma takes the city of Citystan.
- Year 705, Roma loses the city of Citystan due to loyality presure.

End of the military account of the world history.
Haha that's a great point
 
No matter how much someone may wish it to be so, city walls are not a substitute for an army.

while I personally find wall resistance to melee (85%) to be way too high, Almost every encounter fighting an AI would be improved if they built more units and especially kept building more to replace losses in war.

I think one of the warmonger AI personalities is capable of doing it. But if they did that then you’d be more focused on fighting the defending army than taking the city. The fact that there is a token defensive force is what leads players to harp about city sieges- I’m every war humans push right up to the AI cities quickly.

More units. More men to the front! Not one step backwards! For the motherland!!!
 
I think siege weapons should be more resilient to city attacks. You know if there is a mod for that?. That would greatly make wars more dynamic.

That said I agree, on the AI needing to be capable of replace losses in war and use more units in city Attacks. Also the AI should chose cities that are easier to take and keep. And keep fighting after taking a city, trying to take surrounding ones if neccesary to secure loyality.

The AI needs to be more constant and more ambitious when conquering. More expansionistic and oportunistic in general.
 
Last edited:
I think siege weapons should be more resilient to city attacks. You know if there is a mod for that. That would greatly make wars more dynamic.
I don't know of one on steam, but this change is fairly trivial to make using the development kit. I feel like I should bite the bullet and produce my "advanced unit rebalance mod" which would include that.
That said I agree, on the AI needing to be capable of replace losses in war and do some other stuff etc
Pumping out units is easy. The other stuff is hard. I think it's AI shaka who will pump units at you during war but I cannot recall. The AI, esp on deity, have great production boosts, and if players were made to feel that heat they would think twice before going to war.
 
Top Bottom