My first Deity game and why the AI needs work

[KayAU]

So, I decided to go for the Deity achievement, while also trying out one of the last leaders I have yet to play with, Tomyris. I don't like to play as a warmonger, but was after a quick game, so I went for a Duel map, aiming for early conquest.

I restarted the game a couple of times until I got horses near my capital (just didn't feel right to be horseless Scythia), and then immediately went west. This turned out to be a lucky choice, as I very soon ran into Alexander. That's potentially interesting, I thought, another war-focused civ. I soon found myself in a position to capture a worker with two charges left, so I declared war. Playing on Epic speed, I was just getting close to finishing my first slinger. Now, against a competent, or even a fairly weak player, this should have been suicide on Deity. Alexander already had 3 cities, at least 5 warriors, and at least one or two slingers, as well as a general combat bonus for the difficulty setting.

However, when it comes to combat, the AI is not merely weak, it is completely helpless. True, Alexander rushed me immediately, and even managed to land a few shots on my capital, but once that first wave was defeated, he had nothing. I went straight for archery and then horseback riding. My archers and single warrior easily ran through his empire, and the game ended on turn 64, the year 2425 BC.

To be fair, this game had some special constraints, with the very small map size and my knowledge that Alexander was the only thing I would ever need to deal with. Still, given the massive advantages that the AI has on Deity, they should be able to put up some kind of fight. Their initial attack was poorly executed, and easily fended off with only two units. During my archer rush afterwards, I met only a few combat units. A couple of slingers, a couple of heavy chariots, one at a time. Apart from that, Alexander had prioritized holy sites and an unescorted settler.

It is fortunate that I generally don't like to play as a warmonger, as I think I would struggle to find much satisfaction with the current state of the AI.

 

[oedali]

Very true

 

[unpossible251]

I always win, but then again I am pretty bad for "Edge Of Tomorrow" -ing it with lots of autosave spammage for the dream combat run. Massacre everyone without taking a real loss is simple, as long as save spam and rage quit are part of your toolbag (they definitely are mine).

Would anyone else be ready to admit they play like that?

It might make an interesting challenge (or mod) - we all know that the AI is bad, but i got to wondering last night how maybe my play style makes the AI dumberer than i think.

 

[KayAU]

I will admit to saving and reloading at times, but during this specific playthrough, I am not sure I reloaded even once. If I did, it was to change a tech choice which I did by mistake. Not sure if that was this game or the one after it.

 

[cazaderonus]

I think i stopped saving/reloading with civ v. As for combat AI, yes it s very weak when they have units you can match. But last game i played, the attack poland launched pn me with a gazillion knights made some real damage. Tbf she should have been able to just take half my empire if she had played better, but i was pleasantly surprised in how she executed her offensive.

Biggest problem atm is ai s trade. You can get insane deals from them for a few gold coins

 

[TheMeInTeam]

I will not save/reload in this game unless the UI lies in a way I didn't know/forgot. It's already so freaking slow, reloading to grind out the same turn yet again when it's completely unnecessary to win? No thanks.

 

[bitula]

Actually I had the opposite experience in several post summer patch Deity games. They came in two waves all together like 15 units, archers, swordsman, chariot vs. my 5 units surrounding my town. I could not fend them off because they didn’t blindly attack my city but decimated my army first, starting with the weakest unit. This was the first time when I actually calculated with their attack, was prepared, and lost nontheless.(no save/relaod ofcourse)

 

[mckinney156]

I'm not exactly sure why you think the AI needs a reboot here? You backstabbed him, with a knowledge advantage, defeated his garbage starter units, then ran through him with Scythian horses, after re-rolling several times to guarantee you could build said broken horses.

What did you expect the outcome to be?

 

[KayAU]

You got a few things wrong there. Yes, I did re-roll a couple of times to get horses, but that was the extent of my "cheating". As I said, I just felt it appropriate that Scythia should have horses (although their UU actually doesn't require it, that's how UUs generally work). When I try out a civ for the first time, I always make sure I have a start where their abilities come into play. I would also have re-rolled if I was playing as Kongo or Brazil for the first time and didn't get any jungle, or Nubia or Egypt without desert, for example.

Anyway, it doesn't matter. Mounted units didn't have time to come into play in the attack on Macedon, that was all done by a few archers and the single warrior unit I got at the start. Also, I did not "backstab" the AI. I stabbed him squarely in the front, with basically no preparation. At the point when I declared war, I had: one city, a warrior and a slinger. The warrior was up near the Macedonian border, the slinger was at my capital. What I would have expected to happen in that situation, on Deity difficulty, was to be crushed by the Macedonian army. At this point, Alexander had at least 5 warriors, a few slingers, and 3 cities. He had massive advantages in production, and his cities had been improved due to the builders he got at the start. He had large bonuses to science, culture and gold, as well as direct combat bonuses due to difficulty. To repeat, I had: 1 warrior, 1 slinger, 1 small city, no bonuses, units not even in position. No "broken horses" involved, nor to be available in the near future. I managed to produce a double Saka Horse Archer (which again, doesn't require the horse resource to produce) one time towards the end of the attack on his last city. I used these units to chase some barbarians, as I would not have had time to get them over to his city anyway before it fell.

If you honestly don't think Alexander should have been able to put up a fight in this situation, you have very low expectations for the AI. Firaxis aims to create challenge for the player by giving the AI bonuses. The bonuses on Deity are huge. The fact that a decent player (which is how I'd rate myself) can defeat them with such ease with no significant preparation, shows that bonuses alone are not an adequate way to ensure that the player gets a good challenge. The AI needs to be able to play the game at a reasonable level.

 

[Eliminator_Sr]

Rerolling for starting horses makes a pretty huge difference. You are essentially stacking the odds in your favor. Also Scythia is arguably the best civ in the game for domination. Yes the combat AI is atrocious so you aren't going to get a challenge at all on a dual map. Try it with 7 civs where you can't completely isolate one and take them out early.

 

[KayAU]

I don't agree that the horses made a huge difference in this case. For one thing, it is a fairly mundane thing to get near your capital, and my starting position was overall pretty average. With regards to what I got out of those horses, it was just a nice workable tile, 3 food, 1 production. It hardly compares to the kind of yields Alexander had.

You are of course right that it would have been far more difficult to win on a larger map with more players. I did mention that in my original post (see the paragraph starting with "To be fair"). The point of this thread was to highlight how bad the combat AI is.

 

[bitula]

Do you have mods installed, like AI+ which may conflict with summer patch? In my unmodded game the AI came in large groups, clustered, with insane flanking bonuses and archers at the rear, which is impossible to stop with two units. They always attacked, not just wandered aimlessly, like in Civ 5. Maybe you experienced an AI bug specific to Alexander, just start a new game and try again. I played mostly on duel pangae and had no major issues with tactical AI.

 

[KayAU]

I don't have any mods installed, just the base game and all the DLC. What you describe sounds a lot better than what Alexander did, but the poor performance did not entirely surprise me based on my previous experience with the AI on lower difficulties. It seems like Alexander just threw what he had at me at the moment, and while they arrived at approximately the same time, the movement and attacking around my city was not at all well coordinated. They did not wander aimlessly, but when they just managed to get one or two units in position for attacking in each turn, it was not hard to defend by utilizing the terrain and the cover provided by the city.

Perhaps the difference in our experiences is due to how early I started the war. While Alexander did outnumber me by far, he did not have archers or mounted units available at the time. Taking a city with slingers and warriors is very doable, but it does require you to understand when to strike, and when and how to position yourself. Alexander may have thought his army was large enough (which it should have been), and was not willing to wait until he had produced more troops.

 

[bitula]

Probably then yes, it’s due to very early attack and some AI glitch. Did he at least group all warriors and slingers together, like 5 warriors + 2 slingers? That would be tough IMO. Note, I played on online speed so there were always archers and chariots in the early mix and all were grouped closely together. I just recommend you start another game, you probably won’t be that lucky again. A sample of one game is not a good sample. Comparison with lower difficulties doesn’t make too much sense either.

 

[KayAU]

I think they were grouped together fairly closely, yes. Positioning is lot more difficult when you have slingers as opposed to archers, as you have to get the slingers to the front in order to attack. A human player will be able to utilize cover and rivers to protect the slingers somewhat, and to more accurately determine when it makes sense to spring out from cover and attack. I have a suspicion that the AI is not even thinking ahead one turn, to evaluate if the moves it is doing is leaving it in a vulnerable position.

You are right that I don't have a good sample size here, but isn't the AI more or less the same on all difficulties, with only things like bonuses and level of aggressiveness being changed to increase difficulty?

I do think that online speed would be more difficult, as that would give the AI more time to develop, and exploit its advantages. I play on Epic speed, as I enjoy having a lower tech development to unit movement ratio.

Perhaps I'll try it again to see if I can do it equally easily.

 

[bitula]

I think you should try another game. Probably you had some good terrain conditions which prevented AI units to surround you. Because given even just 5 warriors vs. 1 warrior and a slinger. Given the +4 difficulty modifier, your warrior is killed in 2-3 strikes. Slinger 2 strikes. So how did you survive this without a single unit lost given the AI is not just wandering and is tightly grouped, so can easily surround your units? City with no walls is down with 3-4 strikes.

 

[KayAU]

The terrain was okay, I had a river and some forests to help me. It wouldn't have been too hard to surround me, though, but that would have required the AI to actually position their troops, wait, and then do a coordinated assault. My units were never in that much danger. The slinger was inside the city most of the time. The warrior was positioned behind the city (east) most of the time, where it was flanked by the river running along the south side of the city. I avoided landing final strikes with it when such strikes would move it to a disadvantageous position, choosing rather to let wounded units run away if necessary, or taking care to land final blows with the slinger, which could do so while staying safe inside the city. And of course, the city itself was useful in absorbing some of the attacks from the enemy warriors.

 

[PYITE]

The random hidden agenda of major civs can significantly alter how many military units they build.
Alex frequently gets cultured and Wonder lover. This causes him to spend his hammers building wonders and theater districts, instead of building his military.
He is clearly the most challenging when he gets the standing army hidden agenda which pushes him to build more military.

 

[bitula]

The terrain was okay, I had a river and some forests to help me. It wouldn't have been too hard to surround me, though, but that would have required the AI to actually position their troops, wait, and then do a coordinated assault. My units were never in that much danger. The slinger was inside the city most of the time. The warrior was positioned behind the city (east) most of the time, where it was flanked by the river running along the south side of the city. I avoided landing final strikes with it when such strikes would move it to a disadvantageous position, choosing rather to let wounded units run away if necessary, or taking care to land final blows with the slinger, which could do so while staying safe inside the city. And of course, the city itself was useful in absorbing some of the attacks from the enemy warriors.

We’ll I’ve seen this pre-patch. But after summer patch, they came in tight formation aiming for rather units than attacking the city, so sooner or later they flank your unit, but maybe it is another hidden agenda for Alexander „pacifist army”. Would be interesting if you play another game on Deity and report your findings.

 

[Xoon]

Sounds like you people might need to try more than one opponent. While I'm working on one, two more and a thousand barbarians hit me from behind. Doesn't make the AI any smarter, but you don't have to be too smart to whip a person in a fight if you have 3 buddies.