Still don't get AI reasoning

[roscoepfox]

I think the main problem is how the AI behavior is portrayed to the player especially because of the "denunciation" system.

In my last game I was condemned by practically everyone at one time or another. I had never fought ANY of them. As a matter of fact, the only factions I was next too were ARC and PA. I can kind of understand either of them condemning me , especially ARC because we were smack next to each other from the beginning. They literally landed right next to my explorer as he was doing an expedition.

Of course they then accused me of searching for salvage in THEIR lands. My telling them to piss off probably is why we never really got along throughout the game. But that's the thing, huh? There's no way I would agree to not look for slavage and I think most of us would ignore the AI's request for that anyway. They say "Do I have to escalate this?" yet I've NEVER seen them do anything about it.

Every single value that is meant to determine the likeliness of an AI to declare war on someone else is shown as something that they feel like a direct offense against themselves.

Because of that the player often reacts surprised or baffled asking himself "why is this AI pissed at me?"

The human player expects a valid, rational reason for someone to be angry at them, and the AI appears to be completely illogical in that.

But if the AI appeared to be not angry but rather simply greedy "I have nothing against you, but I really like your wonders and your land, sorry, nothing personal", that would be something that most people would understand as a logical reason to wage war.
It would be hypocritical not to.

Yep. Out of the blue the Asian lady contacts me to condemn me? Why? Affinity? Maybe. I dunno. (she was harmony I was Supremacy). But we were far apart and literally had only seen each others units once in 300 turns. It doesn't 'feel' right. Like earlier civ games where the AI would hate the human player just for being the human player. I think a little dialogue tweaking would help. If the AI had just said what the reason(s) were it'd feel sincere. Instead she's going to tell the world "my sins". What sins? I didn't attack ANYONE the entire game.

"You're an easy pray and a perfect target for my expansionist ambitions." is something that we all can understand, I think.
"Your military is weak! How dare you! I'm going to denounce you and tell the whole world that you must be eliminated!" Not so much...

I would like that. But then they'd have to attack me first. None of the AIs has attacked me yet. I'm surprised at that because there have been times when I had a pretty shabby military and they could have if not take me out completely, get some bases from me. I'm usually outgunned until midgame.

The other reason Civ V/BE AI is disliked is because of its inherent hypocritical nature. They expect you to don't settle near them without any previous agreement but they get angry at you simply for asking the same. They hate you if you conquer cities but they wage wars themselves. It is okay if a few leaders acted like that as part of their irrational personality, but not all of them.

Its the war thing that bothers me the most.. I rarely attack another civ for no reason....if I start a war its for land/resources, and in BE it's easy to get those things. I wish the 'warmonger' thing gave the player a small amount of leeway. I don't really know how at the moment but I'd like the ability to at least tell an AI that gets caught spying "This means war!" without a giant diplo hit. Apparently if I preemptively strike an obvious hostile civ I'm a big bad warlord.

 

[kirbdog]

It's not really random though. The opinion the AI has on other players is based on specific values and none of them are randomly assigned

Yeah, I know.

 

[Juanholio]

No surprise from CiV base. Just playing a game where Barre has scolded me twice and denounced me for killing aliens- while he's sent a sizable force against an alien nest adjacent to my capital. Amusingly he's been decimated by the 3 seige worms I've been containing and tediously shooting at. Second complaint about not being in harmony (he's harmony 1, supremacy 2) and a denouncement later and he's still attacking the seige worms.

Feels like the animosity is from poor logic, especially since harmony is not even his highest affinity.

 

[Ryoga]

Yeah it really seems that AI decides their official attitude toward aliens with a dice roll and then they kill them all anyway.

 

[Mount Suribachi]

Of course part of the problem is that even if you go Harmony, you still have to kill all the aliens....

 

[Juanholio]

Of course part of the problem is that even if you go Harmony, you still have to kill all the aliens....

True but if the logic for being hostile was based on the AI's current aggression stance it would feel less robotic.

In this case he'd committed 6 rovers, 3 soldiers and 3 gunners to the eliminiation of this nest. The aliens attacked him immediately so obviously he'd angered them already. It was faintly amusing that while commiting this force he was telling me off for taking potshots and then getting all but annihilated.

 

[joncnunn]

Moderator Action: "Hypocritical AI" & "Still don't get AI reasoning" merged

 

[Haggbart]

I agree. I want more variety in the AIs. I want mostly peaceful AIs who will treat you how they want to be treated. I want AIs who will agressively settle near my lands and take my excavations, but understand if I do the same. I want warlike AIs who war because they want your land and think you are weak, and are open about it. The problem is that all AIs feel more-or-less the same as self-centered hypocrites.

The problem with that is how the AI in CiV:BNW will let you alone and win a space victory even if it could easily crush you, as long as you play the diplomatic game somewhat right and don't have the likes of Alexander or Monty as neighbours.

The AI must also play to win, not let you in peace simply because you let it in peace. To strike that balance, you will have that conflict between roleplaying and strategic playing. I agree that there should be SOME difference, but I don't want the AI that would never declare war if you don't gave it a roleplaying reason.

 

[Gort]

The problem with that is how the AI in CiV:BNW will let you alone and win a space victory even if it could easily crush you, as long as you play the diplomatic game somewhat right and don't have the likes of Alexander or Monty as neighbours.

Yeah, the AI literally has a "victory competitiveness" number that judges how much they'll hate you for trying to win. Some just don't care if they win or not.

Sometimes I kinda wish Civ didn't have a "one winner" mode, instead you just got points for doing various things. So you get 10,000 points for launching a spaceship, 10,000 for dominating in tourism, that kind of thing. It'd be nice if one guy doing well didn't mean everyone else loses - that'd make the low victory competitiveness leaders actually make sense at the same time.

Could do the same in BE - no reason why Promised Land and Contact victories can't live together, after all.

 

[Ryoga]

The problem with that is how the AI in CiV:BNW will let you alone and win a space victory even if it could easily crush you, as long as you play the diplomatic game somewhat right and don't have the likes of Alexander or Monty as neighbours.

The AI must also play to win, not let you in peace simply because you let it in peace. To strike that balance, you will have that conflict between roleplaying and strategic playing. I agree that there should be SOME difference, but I don't want the AI that would never declare war if you don't gave it a roleplaying reason.

War is not the only way to win, this is a civilization game, not a war game.
You may think that a Civ AI doesn't stand much chance to win a scientific victory or cultural victory before the human player, so you aren't considering the option for the AI to avoid conflict in order to fully concentrate on a peaceful victory and beat everyone else at that. But do you really think that they have better chances with war? Considering how bad they are tactically I really think not.

If you really wanted the AI to play to win then they should rather attack a weak civ instead and avoid the human player as much as possible, but I think that what you actually want is simply for them to attack you, because that's more fun that way.

 

[Haggbart]

War is not the only way to win, this is a civilization game, not a war game.
You may think that a Civ AI doesn't stand much chance to win a scientific victory or cultural victory before the human player, so you aren't considering the option for the AI to avoid conflict in order to fully concentrate on a peaceful victory and beat everyone else at that. But do you really think that they have better chances with war? Considering how bad they are tactically I really think not.

If you really wanted the AI to play to win then they should rather attack a weak civ instead and avoid the human player as much as possible, but I think that what you actually want is simply for them to attack you, because that's more fun that way.

Are you denying that it is a valid and much used tactic in CiV:BNW on deity to turtle with 4-5 cities, avoiding any conflict like the plague while spending $$$ to bribe the AI into attacking each other and winning on science by spamming RAs?

If the roles were reversed, and I was one of the numberless AI civs sharing borders with me during all my playthroughs, and had lots and lots of more military units, I would 100% of the time attack them when they were approaching victory. What is happing is that they will attack you 0% of the times you are about to win the game.

I'm not saying the AI should always attack the human player or always make war (or indeed that it is effective in making wars), but it's a much known problem that it is too easy to sit with a small civ and almost no military and win the game without any of the AI care. I can tell you for sure a lot of players would have to change their tactics if there was a real risk of a DOW when approaching victories. Because the size of army you can get away with now is totally insane.

 

[roscoepfox]

I just finished a game. Playing Pan Asia going harmony. The hardest part was trying to get as much xenomass as I could. Even though Hutama wasn't going harmony he seemed to single out those spots and try to settle them before I could. Kavi was the same way but didn't colonize as aggressively. She only had three cities. ( and yes she had a lot of space to build more.)

I went for a contact victory and so did Brasilia. He was Harmony too and far away so we didn't butt heads. ARC hated me from the minute affinities came into play. (ARC was purity) So did the francos but they too were too far away until her expansion towards me by the end of the game. Now I had bottlenecked Hutama into a small piece of land, but he managed to build a few more bases by bypassing land going over water. I fully expected him to launch an invasion. I would have, it was obvious I was only playing nice with him while I built up a decent force. But he stayed friendly. He condemned me once and I did it back and he was still friendly. Heck, the very next turn he wanted to trade materials. (Huh? Yeah, that puzzled me too). But Kavi....she condemned me left or right and stayed hostile throughout the game. We had borders together and I expected HER to strike since I was building the beacon.

But no...she just called me names every other turn. When the beacon had only three turns left on it I got fed up and attacked her. With one city left she wouldn't agree to my terms of surrender....which I admit I wanted all of her money and resources....so just for the catharsis I destroyed her. (I had already won by that point.)

But the AI didn't EVEN try to stop me. If the human player was in the same position they would most likely try a suicide run to destroy the beacon. The beacon, the mind flower, etc...they are super easy to demolish. ....and from where it was placed Kavi did have a slight chance of getting to it. Hutama might have been able to kill enough o my troops to get to it. They didn't even try. Even the hostile ones. (Elodie, ARC, etc....and ARC had a massive army for the AI and dominated most of the land where she was.)

I don't get it. Victory isn't as fun if they don't even TRY to win or stop your "I WIN" project.

 

[Ryoga]

Are you denying that it is a valid and much used tactic in CiV:BNW on deity to turtle with 4-5 cities, avoiding any conflict like the plague while spending $$$ to bribe the AI into attacking each other and winning on science by spamming RAs?

Isn't that exactly my point? If you agree that to turtling up and avoiding conflict like plague is one of the most effective way to win, why isn't exactly this what you are asking the AI to do if what you really want is for them to be competitive?

 

[Haggbart]

Isn't that exactly my point? If you agree that to turtling up and avoiding conflict like plague is one of the most effective way to win, why isn't exactly this what you are asking the AI to do if what you really want is for them to be competitive?

Oh it does, and - surprise - it's not very good at it.

A simple increased chance of DOW when the AI has a lot more military and the human player is approaching victory would significantly change the way the human player must play in order to win.

 

[traius]

Isn't that exactly my point? If you agree that to turtling up and avoiding conflict like plague is one of the most effective way to win, why isn't exactly this what you are asking the AI to do if what you really want is for them to be competitive?

It's an effective way to win because the AI is already doing it, but poorly. Turtling up on a few cities and going for an endgame victory condition is easy because the AI makes no attempt to stop you. Modifying them such that they are "going for it too" is unnecessary because they already are being super passive, and suck at getting the VC.

 

[roscoepfox]

It's an effective way to win because the AI is already doing it, but poorly. Turtling up on a few cities and going for an endgame victory condition is easy because the AI makes no attempt to stop you. Modifying them such that they are "going for it too" is unnecessary because they already are being super passive, and suck at getting the VC.

It'd be different if the AI would do something to stop the player while turtling their way up. I think they may focus on wonders too much. Even early in the game I see messages that whoever built whatever when most of us have three, maybe four cities and are fighting for health. The wonders don't usually give that much of a bonus. I'd rather have some improvements, units and maybe another base (thus gathering more resources.) Maybe that's why I can easily get ahead of the AIs.

I built the promised land gate and began shuttling earthlings. At the same time I was in a war with Kavi again. (this time Brasilia asked me to join them in the war and I said yes because the AI never does anything interesting and I wanted some excitement while basically winning.)

Africa and France were on my borders and both had the means to pprobablly rush the gate and stop me. Neither did. Kavi's cities fell like dominoes and AGAIN she only had built THREE while having the room for more. Brasilia didn't actually do much during the war but afterward they rolled up to my newly acquired bases with a MASSIVE army. (we had open borders.) If they had this llarge army why didn't THEY try to stop me? Even though I shared affinities with them and Africa I'd have attacked them in a hot minute if they were about to win.

I know the AI can't "think like a human" but its only slightly better than the Civ % AI when it launched...and it really sucked.

 

[Gort]

Isn't that exactly my point? If you agree that to turtling up and avoiding conflict like plague is one of the most effective way to win, why isn't exactly this what you are asking the AI to do if what you really want is for them to be competitive?

There are two ways to look at AI in an unbalanced game. You could say, "What a stupid AI! Doing X is obviously the best move - look at this dumb AI doing Y and failing!"

Or you could say, "Man, I wish X wasn't always the best move - then when the AI did Y it wouldn't fail!"

I'm all for improving AI, but often it plays the game the designers thought they were making rather than the game they actually made. Of course, when it comes to patches Firaxis often leave a lot to be desired when it comes to both AI and balance.

 

[Ryoga]

Oh it does, and - surprise - it's not very good at it.

And I think we can all agree that it's not really very good at war either

A simple increased chance of DOW when the AI has a lot more military and the human player is approaching victory would significantly change the way the human player must play in order to win.

What I'm trying to get to is that this has nothing to do with competitiveness.
Avoiding conflict isn't a sign of a lack of competitiveness at all, because that's actually a perfectly valid strategy to win.

The fact that the AI fails just means that you are better than them, not that they made a poor decision, because chances are that their military attack is even more likely to fail.

It's an effective way to win because the AI is already doing it, but poorly. Turtling up on a few cities and going for an endgame victory condition is easy because the AI makes no attempt to stop you.

Actually it's even more effective when the AI focus on military, because then it's even less likely that they can beat you to a peaceful victory since they have expended several turn in building units rather than scientific or cultural building\wonders.

When they attack you, yes you'll be forced to build military too, but the defender has always the advantage, so you need considerably less units.

 

[Haggbart]

And I think we can all agree that it's not really very good at war either



What I'm trying to get to is that this has nothing to do with competitiveness.
Avoiding conflict isn't a sign of a lack of competitiveness at all, because that's actually a perfectly valid strategy to win.

The fact that the AI fails just means that you are better than them, not that they made a poor decision, because chances are that their military attack is even more likely to fail.



Actually it's even more effective when the AI focus on military, because then it's even less likely that they can beat you to a peaceful victory since they have expended several turn in building units rather than scientific or cultural building\wonders.

When they attack you, yes you'll be forced to build military too, but the defender has always the advantage, so you need considerably less units.

"Better" is a relative term here. The AI is competitive on higher level because of huge bonuses. Several on these bonuses are directly related to military. IIRC it has 50% unit cost, 50% unit maintenance and 50% upgrade cost. It doesn't need to "focus" on military to have these bonuses. And it tends to build huge armies late game without using them, or at best attacking other AI that aren't even close to winning the game

If it almost never declares war it won't have any advantage over the human player with these bonuses. And it doesn't matter how inefficiently it moves the units around, you'll be forced to have some military units that will slow down your pursuit for victory somewhat.

But right now the AI is roleplaying too much, and playing to win too little.

 

[Gort]

I always thought the AI's armies would be more threatening if it got free promotions instead of production bonuses - after a point, the number of units you have doesn't matter.

If you've got ten thousand bad units attacking on a five-hex-wide front, all you're going to get is a meatgrinder.