Bad AI is not likely to be fixed

[MxM]

The wildly different results are good, in principle. I'd just like the AI to make the most of the (dis)advantages of a location! There was plentyyyy of space to settle in my case, btw.

My guess is that the AI picks from a set of subroutines. That means the allocation criteria for subroutines have to be finetuned (with a chainsaw) and the routines themselves need work as well.

My guess is that this game was intended to be won equally well by small empires with high happiness. The AI is tuned for that. Something happened somewhere in development and bonuses for happiness and negatives for unhappiness are not strong enough to make it true. Can we mod in larger influence of happiness?

[pdxsean]

TBH I see some really weird things from the AI. Occasionally it'll do something that appears very brilliant. I've had the greeks attack me from two directions, bringing weaker units into the open to draw out my defenders while a stronger force swings around the side. When I saw this I felt stupid for falling into the trap... until the flanking units kept moving northward (passing within one desert hex of my archer!) toward... um, nothing really. One walked across my only horse resource, and continued on into the grassland hex adjacent. So I guess the "brilliance" was merely the coincidence of a horrible general.

[Zoetstofzoetje]

The wildly different results are good, in principle. I'd just like the AI to make the most of the (dis)advantages of a location! There was plentyyyy of space to settle in my case, btw.

My guess is that the AI picks from a set of subroutines. That means the allocation criteria for subroutines have to be finetuned (with a chainsaw) and the routines themselves need work as well.

[Guynemer]

I can't remember the last time this happened to me in Civ4, but I actually lost a city--my capital, no less--to the AI in my last game. China sneak attacked me while my army was positioned on the Iroquois front, Hiawatha having waged a pair of wars on me already. By the time my forces returned to the Chinese border, they had taken Istanbul and were on the brink of taking Ankara. If I didn't have the Janissaries' special ability of auto-healing when killing an opponent (mostly mustketmen and cho-ko-nu), I'm not sure I could've repelled their invasion.

[pdxsean]

On my continents map however, the AI's never built more than one extra city. The city-state system, while novel and potentially very interesting, is simplistic at present. What's worse, the AI had armies of warriors at a point where I had a prosperous 12 city empire. This is on King, mind you. Every now and then they would spawn better units (pikemen, horsemen), but never more than one.

I've heard this complaint from at least one other person, and it really mystifies me as in my games the AI is so aggressive with their expansion I don't understand how they manage it. I'm also playing on King level, and I've never had a game where one AI doesn't take over their entire continent and the other AIs take over as much of my continent as possible until I take them.

Strange how it would cause such wildly different results. Lots of improvements to be seen, that's for sure.

[Adagio]

Did you see any correlation of what workers are doing with your happiness/gold income? Because if you have high happiness and good gold income it make sense to put farms. If you income is negative and/or your happiness is around 0 or less, then it is better to put trade posts

My happiness was around 0, but I had plenty of money... but the worst part is that they keep on changing, which means the land is not much useful most of the time
At least in Civ IV you could set the workers to never replace tile improvements

[Robert]

Civ4 AI got huge improvements and got pretty good in the end, of course including 'cheating'.
But indeed, it's hard to make a very good AI for such a complex game.

Let's hope that we can improve it ourselves as well when the SDK will be released!

[Proteus_MST]

In the game I am playing (Rome, naturally), that's definitely not true. Oba has a number of cities, as do the Germans. Both are uncomfortably close to me, and the Germans have been coming at me with Landsknechts. True, the AI needs work as far as tactical battles, but they HAVE spammed cities. (on Prince, no less)

Similar to my experience...
all AI players were rather busy founding cities...
and at the moment (early industrial era) I am nibbling through Alexanders Empire who,
through foundign of his own cities and through conquest from other AI players as well as 2 city states,
had an 14 cities empire (well, now 4 less, including his capital)
And I dare not to speak of Siam who, despite my having 19 cities, through conquest and founding of cities, are on 1st place in terms of population and land,
having more than double of my population, as well as 50% more land compared to myself

[Jvstin]

I just managed to tinker Civ5 into working on my computer yesterday. Not quite sure what the problem is, but my video card is not accepted. Through a trick with a USB video card I managed to get it up and running. The funny thing is that the USB video card (a displayport) actually uses the capabilities of my normal video card (ATI X1300) to run. Anyway.

I found the game to be quite fun in the early and mid-game. The new mechanics operate in the same way as I used to run my Civ4 games. I always turned off research trading, put the game on epic, and the loss of the slider means long-term planning is more important. At worst it decreases your flexibility.

On my continents map however, the AI's never built more than one extra city. The city-state system, while novel and potentially very interesting, is simplistic at present. What's worse, the AI had armies of warriors at a point where I had a prosperous 12 city empire. This is on King, mind you. Every now and then they would spawn better units (pikemen, horsemen), but never more than one.

Why did the AI never get around to building more than a second city? Why did they not upgrade or replace their warriors?
.

In the game I am playing (Rome, naturally), that's definitely not true. Oba has a number of cities, as do the Germans. Both are uncomfortably close to me, and the Germans have been coming at me with Landsknechts. True, the AI needs work as far as tactical battles, but they HAVE spammed cities. (on Prince, no less)

[Zoetstofzoetje]

I just managed to tinker Civ5 into working on my computer yesterday. Not quite sure what the problem is, but my video card is not accepted. Through a trick with a USB video card I managed to get it up and running. The funny thing is that the USB video card (a displayport) actually uses the capabilities of my normal video card (ATI X1300) to run. Anyway.

I found the game to be quite fun in the early and mid-game. The new mechanics operate in the same way as I used to run my Civ4 games. I always turned off research trading, put the game on epic, and the loss of the slider means long-term planning is more important. At worst it decreases your flexibility.

On my continents map however, the AI's never built more than one extra city. The city-state system, while novel and potentially very interesting, is simplistic at present. What's worse, the AI had armies of warriors at a point where I had a prosperous 12 city empire. This is on King, mind you. Every now and then they would spawn better units (pikemen, horsemen), but never more than one.

Why did the AI never get around to building more than a second city? Why did they not upgrade or replace their warriors?

The key issue with Civ5 is definitely AI. I love this new combat system, it allows for all kinds of ingenious strategies. A smart player can use terrain and circumstance to their advantage. It's fabulous. But it's a bridge too far for the AI. Or two or three bridges.

This game should not have been released yet.

[Tyrathect]

It wasn't perfect but most people agree it was decent as far as AI for a pretty complex game goes. Would you like to expand on what you think was wrong with it?

I agree it was decent in comparison to other 4X titles. Still quite dumb though, and I don't think the BTS AI was a major improvement upon the original, only an incremental one.
What's wrong about it? I can give you some examples. It still can't conduct a successful marine invasion. It can't properly prioritize different fronts in warfare. It still builds a ****load amount of unnecessary naval units. It still tries to conquer the same border city wave-by-wave with too few siege units, instead of massing a single stack and steamroll you in a single attack. All you have to do is just have enough archers/longbowmen/infantry with defensive bonuses in one city and you're fixed for a massive XP farming. After he finishes throwing his entire army at you, all you have to do is just retaliate and conquer his cities one by one with ease.

[ColdPhoenix]

Nope. Civ4 BTS AI with latest patch still sucks.

It wasn't perfect but most people agree it was decent as far as AI for a pretty complex game goes. Would you like to expand on what you think was wrong with it?

[Tyrathect]

You do realize that Civ4's AI was significantly improved via patches, since they included improvements from the better AI mod right?

Nope. Civ4 BTS AI with latest patch still sucks.

[dregor]

There's one simple thing they could do that would slow down early rushes by human players. Right now you can wreak havoc on the AI civs with three or four horsemen and a great general. It doesn't even matter if the city you're attacking has walls. It'll just take a turn longer to conquer the city. Why not make it virtually impossible to take a city without siege equipment? That would slow down rushes because the catapults or trebuchets wouldn't be able to keep up with the mounted units. It would also make city walls a meaningful defensive upgrade.

They would then have to make catapults and trebuchets not require Iron, which they currently do. And honestly, it's a band-aid fix and doesn't address the problem.

[Dactyl]

There's one simple thing they could do that would slow down early rushes by human players. Right now you can wreak havoc on the AI civs with three or four horsemen and a great general. It doesn't even matter if the city you're attacking has walls. It'll just take a turn longer to conquer the city. Why not make it virtually impossible to take a city without siege equipment? That would slow down rushes because the catapults or trebuchets wouldn't be able to keep up with the mounted units. It would also make city walls a meaningful defensive upgrade.