Does the AI actually try to win?

Juanholio

Warlord
Joined
Jul 6, 2012
Messages
166
I can't quite figure out whether the AI are supposed to be obstacles to winning or are acutally 'playing' to win themselves?

I've never seen an AI get close to a cultural victory let alone complete one; AI diplomatic victory is more like a player-loss from bad planning in my eyes and I always manage to finish the spaceship even if a more advanced AI has long beaten me to Apollo Program and built a few parts. In the latter case I often see SS parts 'stuck' around the map or just sitting in an AI capital without being added.

I've moved up to playing far more immortal level games and I'm still adapting to having to change my build and expansion strategies since I had the luxury of meandering about on emperor - that said I'm still often winning science VC pretty late (turn 320-350).

I know I am skewing results on the basis of persuing games which are going well or I've made a good comeback from a difficult start so I'd be interested to hear others views on this.
 
I think the AI is more of a deadline you have to meet. If you don't win before a certain turn, then there is a chance you will lose. For Immortal, I think they can win around T270 if they have a good game. Turn 300 for normal games.
 
I think the AI is more of a deadline you have to meet. If you don't win before a certain turn, then there is a chance you will lose. For Immortal, I think they can win around T270 if they have a good game. Turn 300 for normal games.

That sounds about right, I tend to go after whoever is the leader with some aggression - maybe that buys me more time. I guess it's pretty hard to finish a space ship when it's raining nukes ;)
 
The AI does try to win--it's just not very good at it. I've heard the whole "Not finishing the spaceship" thing results from the AI having a worker idling in its capital, and simply being too stupid to move it out to add the ship piece; and 90% of the time the AI will ICS, destroying any real chance of a cultural victory.

Really, the only victory the AI can win consistently is a Diplomatic victory, and that results from the player building the UN and some random runaway buying out all the city-states, not the AI itself actually pursuing the victory.

The large number of warmonger civs also dampens most of the AI's chance of winning, as the ones that do stand chances of cultural/science usually get rolled by the warmongers, and the warmongers are completely incapable of actually winning a domination victory.
 
Not only do the AI try to win, but much like a real player, they decide far in advance how they would like to pursue victory. Their civ flavor influences how likely they are to try different strategies. Egypt and Siam, for example, are more likely to chill with a couple cities and try for a cultural victory.

They usually don't succeed in any victory condition because the player starts to outpace their tech gain very quickly, unless you are playing on immortal or deity. And the domination AI don't succeed because they'd have to go through you first. It goes without saying that if you want a better chance of an AI threat being real, step up the difficulty.
 
You won't ever see a domination victory by the AI unless you are the last capital it needs. I'm sure it is capable though. Smallfish has some impressive screenies of Hiawatha conquering his entire continent and getting a good start on the second.

Problem with culture is it requires a very specific way to play, and the AI generally sucks at working specialist slots and managing great people.

So yeah they try and win, but realistically they are more of roadblocks.

Unresolved, I'm assuming those numbers are for extreme runaways? T300 seems fast for Immortal. I'd put it closer to T330 range. If you focus on taking out the runaway, it is a long time before the second runner-up reaches the finish line.
 
Really, the only victory the AI can win consistently is a Diplomatic victory, and that results from the player building the UN and some random runaway buying out all the city-states, not the AI itself actually pursuing the victory.
On imm+ AI will launch every time if you let it. Any science/culture/expansion focused civ can do it.

You won't ever see a domination victory by the AI unless you are the last capital it needs.
242 dom victory has been reported recently! :D

Unresolved, I'm assuming those numbers are for extreme runaways? T300 seems fast for Immortal. I'd put it closer to T330 range. If you focus on taking out the runaway, it is a long time before the second runner-up reaches the finish line.
Not an extreme runaway, just a runaway. 300+ is an average danger zone from my experience.
 
242 dom victory has been reported recently! :D

Why no screens! :(


Not an extreme runaway, just a runaway. 300+ is an average danger zone from my experience.

After fall patch? I don't have enough fully completed standard Immortal games since then to confidently say one way or the other, but before that I was never too concerned about losing until 330+ (outside of extreme runaways, of course).

I could see fall patch changing that though. I think it was you that posted that Russia game a while back? My second attempt on that and Askia shows up on my doorstep with a billion (yes, I counted :D) mandekalu cavalry at turn 5 (ok maybe slight exaggeration :lol:).
 
300 is maybe a moderate runaway. I had a game where I finished the Utopia Project(and won) at turn 301 and I was sweating bullets the last 15 turns. Egypt(I think it was) only started pulling away the last 70 turns or so, but they were working on the last spaceship part around turn 300. They had all the research and stuff, so it was only maybe 5-10 more turns until they won. They only had around 1700 points, so they weren't a massive runaway.
 
it's definitely much better than earlier civ games when most of the AI's were just there and doing their own thing. now at least they pick a victory type and go for it. they're just not very good at most of them. at least now when you are building a spaceship, it's pretty common for them to make a hail mary DOW attempt to try to stop you - even though they never do it effectively, like with nukes. in the past games the AI's were happy just to exist and let you win.
 
Why no screens! :(
Indeed. Bummer. I've never seen anything like that.

After fall patch? I don't have enough fully completed standard Immortal games since then to confidently say one way or the other, but before that I was never too concerned about losing until 330+ (outside of extreme runaways, of course).
Yeah... I think. Sort of. :mischief: My experience is also a mix of imm/deity, so can't tell for sure as well, besides I try to finish well before 300. But based on few games that lasted long enough to make a prediction, I'd estimate it about 300. Buccaneer mentioned 270 launch a couple of weeks ago and, of course, you must remember this lovely thread. :D That Bismark would've launched at 270-280 with ease.

I could see fall patch changing that though. I think it was you that posted that Russia game a while back? My second attempt on that and Askia shows up on my doorstep with a billion (yes, I counted :D) mandekalu cavalry at turn 5 (ok maybe slight exaggeration :lol:).
DC1? :D Yeah, I was lucky enough to kill him early and take a healthy bite of Japan along the way and even more lucky with Maria Teresa not being a tech monster. I finished in 280's when she barely completed Apollo. But, AI got its payback in DC12 while everyone else but me cruised through. :crazyeye:
 
it's definitely much better than earlier civ games when most of the AI's were just there and doing their own thing. now at least they pick a victory type and go for it. they're just not very good at most of them. at least now when you are building a spaceship, it's pretty common for them to make a hail mary DOW attempt to try to stop you - even though they never do it effectively, like with nukes. in the past games the AI's were happy just to exist and let you win.
Still not good enough. Once I've been attacked by 40 (not exaggeration :p) Ethiopian planes like 3 turns before I launched. The idea is right, the execution... ehm... :)
 
Still not good enough. Once I've been attacked by 40 (not exaggeration :p) Ethiopian planes like 3 turns before I launched. The idea is right, the execution... ehm... :)

Was there a mountain range in between the borders too?
 
Yes and no.

The AI does try to win, but only in its own perspective. For example, when you are about to win by science, the AIs usually do nothing to stop you. However, when left by themselves, they will obviously win by science or otherwise eventually.

As far as cultural victory is concerned, the AI is simply not in a good position to win that way: because for CV they need to stay small, and this usually means they will get crushed by bigger neighbours.
 
Top Bottom