Going for victory?

[solanacea]

Does this AI do a better job of "going for" victories, i.e. maxing culture in 3 cities, teching to build the spaceship, etc? Or is it intended to role-play only?

 

[scu98rkr]

The AI can go for both Culture and Spaceship victories.

"Going for" diplomatic victories is pretty hit and miss even as a human but the AI got a diplomatic victory in my LAN game the other day mainly cos the guy I was playing with didnt want me to win.

I think some work needs to be done to make it go for conquest and domination.

 

[BobTheTerrible]

Jdog said in a test game of his, the Romans won a conquest victory in the 1900s due to vassalizing the world. Also, The default AI is BTS will refuse to stop wars at a certain point (option will be redded out) saying "We'd rather win the game, thanks." The AI doesn't specifically "go for" victories, but it will realize if it is close to one and try to take advantage of that (I think).

 

[solanacea]

To me it seems like AI can win cultural victory by accident in higher difficulty levels if the human player does poorly but keeps on playing. I don't know if the AI can "see" the possibility of cultural victory and retool GP, city prod, etc to achieve that goal. If it could, that would be cool.

 

[phungus420]

According to Blake the AI actually decides to go for a cultural victory around the time Jewdism/Confusionism is founded. If the human is playing below their difficulty level, they will usually control the religions and wonders to an extent where a cultural victory is unlikely, so the AI wol't go for it. AI cultural victory is not a case of the AI stumbling into it though, the AI directs it's actions for a cultural victory in a more coherent way then probably any other victory (At least in default BtS and the old Warlords Better AI, not sure about how jdog's BtS Better AI has changed things).

 

[Cybah]

A quote of jdog5000 after I told him that the AI is not declaring war with 2 times more military power than all possible opponents:

There are a lot of factors involved in whether and when the AI decides to declare war, including their relationships with other players and the state of their finances. In 0.50 you can now get some extra info about who they would consider declaring war on and with what odds by holding down ALT and mousing over different players in the scoreboard (have to have ALT down before putting the mouse on the player whose decisions you want to check out). You have to have the chipotle cheat code entered in CivilizationIV.ini to see this.

One thing I've run into and been thinking about is the issue of role-playing versus playing to win ... a human player with a huge power advantage over the rest of the world will pursue conquest/domination, but the AI never ignores its relations with other players and thus some leaders are unlikely to ever go that route even if they have a big military advantage.

On Aggressive AI this isn't much of a problem though ... the AI is much less tied to relationships and will push an advantage when it has one.

 

[scu98rkr]

As far as Im aware the power graph basically show the current strength of each play by adding up the strength of all current units then adding some things in for Walls, Castles etc.

If the AI is going to basing so many of its decisions on the power graph I think they need to be updated. Raw power is the only thing if the CIV has massive production and a tech lead it can quite easily build an army.

Therefore either the power graphs or the AI decision making should also take into account military tech + production capability.

The production capability is reasonably easily studied as there is already a graph for this but as far as I am aware something needs to be done about producing a Military Tech graph.

For instance a AI with a large army will often attack the human because their power is low, but often the human has a tech lead and the newest units.

It might be better attacking another AI with a larger power rate but more backwards military. If we could somehow "RATE" the military tech or maybe look at the best unit each civ has then the AI might be able to balance current military strength against possible military strength to choice to attack the correct target.

Also I think the AI personalities need to evole to allow them to go for the win. For instance if an AI wins a war takes on a vassel etc it is more likely to declare war in the future or more likely to backstab smaller civs.

Or if they lose a war less likely to declare war in future so they dont waster their remaining troops.

This would not be taking away the AI personalities but would just mean they would change sightly during the game and hopefully fit the game they are playing in better.

In some ways blake's cultural victory routine already does this and the same could be applied to tech etc.

 

[BobTheTerrible]

Another thing is to have the AI beeline to techs that get it its UUs and then look for an opportunity to utilize those UUs.

 

[jdog5000]

Phungus is right, the AIs pursuit of cultural victory is the most coherent and complete ... to some extent because it has to be since it takes so long to build up. Blake developed a system where the AI knows which cities are the closest and will build extra culture buildings there, prioritize those cities for cathedrals, and in the later phases it will shut down science spending completely and build culture in its top cities and defense in the others. It's pretty cool to watch (if you're in to that kind of thing ...).

The AI is decent at getting spaceship victories, it makes good decisions but doesn't really pursue it like with cultural victories. It does take a defensive stance once it launches the space ship though I believe.

There are a few things the AI does to go for domination, recognizing it's close and building health buildings. As someone mentioned they will also not make peace if they're getting close. Domination and conquest are things which will occur more as the AI gets better at war and managing a large empire, though right now they're both highly dependent on vassalizing opponents since the AI rarely destroys other civs. Oh, and AI civs who can build their UUs are more aggressive at least through the middle ages/renaissance (they use the Dagger AI strategy more).

 

[armand453]

Have you planned to change something in the way AI choose to vassalise or finally not jdog?

Armand.

 

[jdog5000]

Definitely plan to look at that, but haven't yet. The changes to transport AI are almost finalized, vassalization is up next.