Is 2k satisfied with the current state of the AI?
I'd like to see an official statement on this.
Is the military Artificial Intelligence in Civilization 5 satisfactory with 2K?
Or are there plans to improve upon it?
Last edited by VonPummel; 09-30-2010 at 07:12 PM.
They improved Civ IV AI from pants to not bad so I don't think they will have a problem here.
Agreed. I have been trying to get an answer on this for a while but to no avail. Even when they ask for questions they just clam up when this comes up. Must be on their avoid topic list. Understandably on one hand they have to defend their AI and say it is fine, which would upset everyone that thinks it needs work or admit the game has a serious flaw. Both poor choices, at best I would expect a non answers such as "we will inform you of patch changes as we get information from devs" or something.
I remember some modder made a smarter AI mod for Civ IV that seemed to improve things...
yeah lets hope we will soon have a similiar "better bug ai" mod :-)
but since the modding is poorly integrated in its current state this might not be so much of a help anyway. ( poorly cause u cant tell which mods you used in your games, everyone has to activate same mods for an mp game and so on, instead of automating this you gotta take guesses )
Well but i wouldnt count on that in CivV i hardly doubt that the AI is included in the "easy-access"-department in modding, not to mention that modding an AI probably needs more work and knowledge of the game.
Originally Posted by Easy Money
So i wouldnt expect any good AI-Mods for at least half a year. But hey its not like i would be upset if someone proved me wrong ^^
The AI is in the moddable part of the game in XML format. This wasn't the case in CivIV - the AI was hardcoded. From what I've read, the AI in Civ5 has a variety of events in which to assign a certain 'priority' which is just a number (15, 200, etc).
Originally Posted by Sethos
So some of the AI is very accessible and able to be tweaked.
However, I'm not sure if there is a 'tactic' bracket which assigns priority based on social polices, territory, terrain, etc. I'm assuming *if* this isn't present, its because the computing time would be much bigger than it already is.
However I read an article about the programming in "Magic: The Gathering" on xbox. They had the problem of tackling decision making as well, and I'd imagine a card game like that would have a comparable amount of variables to calculate and probabilities to assign.
They tackled it by defining certain goals, and having the AI reach those goals using the -current- turn as the only reference. It didn't consider past moves, and it didn't predict future moves. It simply made the best moves at any given time to achieve the goal. The way the 'dominos fell' was that the AI was pretty dang good at playing its hand, but a human still had the possibility of outwitting the AI through strategy.
I'm not sure if anything like this could be included in Civ5 AI.
2k is just the publisher, they had nothing to do with making the game or its AI.
The question should be directed at Firaxis (the actual developers / programmers).
My dear sir, that... is an insult to Rocks!
Originally Posted by KingYosef
The disappointment is that during the build up to the release date, the AI was being lauded as one of the great improvements to the game.
Right.. 2K is the publisher.. which gives Firaxis the go ahead to make patches or not.
Originally Posted by Hathur
I think that the AI is one of the improvements to the game, just not the combat AI. I think the diplomatic AI and general strategic AI is good. Just everything that involves the military or actual placement/movement of units is not finished.
And yes, Firaxis/2K and everyone that ever worked on that game is perfectly aware of it, here's proof:
The AI never uses their general, at all, never. Whenever the AI gets a great general, they put it to sleep and leave it there. They don't use it for citadels, they don't use it for golden ages, here's how the code probably looks like:
for(i = 0; i < mWaiting; i++)
switch( units[waiting[i]].type )
..... //other IMPLEMENTED units
default: //this is where great general and all other UNIMPLEMENTED AI features go
2K is just the publisher but do not underestimate their involvement in this mess.
Originally Posted by runtheplacered
I bet anything that Firaxis knew the game still needed work but 2K wanted to meet the Sept 21 publish date.
In a way the publisher has as much power or more then the developer.
My biggest complaint is the AI. Can we please get some feedback on this issue from Firaxis?
There are multiple lists from myself and others about what aspects of the AI need to be improved. Civ is, was and always will be as long as it remains turn based a primarily single player game, that's all there is too it. It MUST have good AI.
I fully agree, it's usually the publisher that gets a big part in what the game is like, usually that part can be easily called 'fault'. It's like the developers are trying to make a good game and the publishers take sole pleasure in making that impossible for developers. Truth be told developers would not be able to make $$$ on their own. Developers try to make games, publishers try to make products. Me personally I don't believe that you'll ever be able to buy a great game; when a masterpiece will come along it will be free.
Originally Posted by Achilleslastand
Originally Posted by cristic
On immortal I have seen much better AI with larger forces. I think the reason is that movement is impaired, so they are actually attacking instead of the random movement.
I have also seen civ's not improve their Capitol city, instead their workers were scouting. So, I believe that they are looking for diplomatic entities in a desperate way.
My conclusion is that the AI just has a mixxed up priority lists. Workers should be focused on maximizing terrain, and military should be focused on decisive attacks and building a defensive wall with ranged.
Hold on, so have you seen enemy generals being used at all?
Originally Posted by Davetopia
Every time I see an enemy general, its sitting idle in the city that spawned it! I never saw an AI-built citadel and they don't even try to move the general away when I conquer the city. That's what I was showing off as proof of unfinished AI.
Great generals are never used. They also do not think before attacking, they just attack for reasons such as "our boarders are touching"
Heres an example, Emperor difficulty. I'm #1, but France isnt far behind at all. Egypt is between us. France starts conquering Egypt, and their capital has 4 wonders in it. So, I decide to take my chance, declare war on Egypt, and take Thebes and free a city state they had that becomes my ally.
France finnishes off Egypt, and now, our boarders are touching, so as I expected a few turns later France goes to war with me. However, they didn't even allow time for their units who had been attacking Egypt to heal, and thus attacked me with several wounded units, which were the same tech level as mine, but easy pickings at 1/2 - 3/4 health. Had they waited even 3 more turns to heal, they might have been a real threat.
I have seen wounded units retreat, so how hard can it be to program the AI to not declare war unit its units have healed and they have re positioned themselves?
The diplomacy is a little lopsided too, it seems like they are much quicker to give up and offer a huge sum (like all their other cities, and all their gold) than in civ IV.
I am just finishing up a game where I saw the AI stack a Great general with his military units and attack my units. They certainly don't seem to be able to build citadels though which is a shame.
In my games I quickly sell off the Great General for a golden age.
The AI is so poor that I just felt guilty using one whenever I got one.
I just administered a Stanford-Binet Adult IQ test to Civilization V's AI, but the computer idled throughout the whole session and refused to yield any answers, so I had no choice but to grade it a standard deviation -6.0 result - an IQ of 10.
lol, on the serious side, yes, the tactical AI does need work; but I felt that the AI management for diplomacy and economics is rather good. It is still far from perfect though.
Anyway, personally squeaking, I mean speaking, I still find the game very fun regardless of the broken AI or not. Well it's just me I suppose, but if there is a bright side to it, it's the fact that I don't have to bother reloading back again and again just to pour more misery into the game for beating me constantly. Yeah it's a hollow victory even if you win, but for old folks in the seventies like me, I did rather have a care-free game than suffer a stroke from frustration.
The question is though do they consider the current AI to be so bad they will patch to improve it in the near future or do they consider it to be adequet enough to make us wait for an expansion to give improved AI?
Originally Posted by Simkill
AI enough for me at King Level
If 2K Firaxis does improve the AI skills, I will just have to play at a lower level than now. King Level is my maximum and any improvements would drop me to Prince or Warlord.
I hope for a better AI because it currently just embarks its units and sends then to their death into the water patrolled by ships. It really is pretty pathetic. My ship can be sitting in a coastal tile and visible to the enemy, and he embarks right by it.
One benefit to an improved AI, even if you have to lower the difficulty is to decrease their bonuses. Once they have a lot of bonuses, it is much harder to actually build a wonder since they have a tech and production advantage.