Internal build 0.78 is the first build that the full strategic AI has been checked in. From then on, it’ll be balancing it based on play testing and listening to players discuss how they beat it. But feature wise, the AI can now fully play the game with one exception – we don’t let the AI win via the master quest. But it will try to win through other means if you have it selected.
The next public build will be where we begin the tactical AI implementation. We’ve been writing the APIs for tactical battles for months but we haven’t made use of them yet since we first have to decide how tactical battles are typically fought. That is, knowing the rules is only part of the challenge.
AI Wars: Let it begin. LET IT BEGIN!
To see how effective the AI is at decimating each other, we put a ton of them on a small map. Specifically, small map, 18 players.
I, on the other hand, will be going out of my way to kill them all quickly.
As you read this, please understand, I’m not trying to play the game. I am trying to win the game as efficiently and, truth be told, as cheaply as possible. Any scummy tactic I can find I’m going to use.
It’s a three way battle: FE’s AI vs. My cheese vs. Kael’s design.
So let’s the cheesing begin…
Since I am scum..
So the other players are out there happily exploring the world. Getting gear, probably training pioneers with dreams of expanding their empires.
What they’re really doing is starting my cities for me because while they’re out, I’m going to take their cities.
So…
What should we do about that? What should they do to stop me?
Here’s the data we have to work with:
What sort of evil should we inject into the function above?
Getting ganged up on
To answer some of the things above, we can create some interesting consequences that will help “mix things up”.
For example, I can take advantage of the ideological differences between Kingdoms and Empires. So if I’m a Kingdom and I attack another Kingdom, that’s a big deal. Or I can make it a big deal if I attack early depending on the faction traits. For example, a player with a Diplomatic trait might freak out a lot more if I go and attack someone early on.
You don't want the AI to go on the defence too much, because that would limit his growth too much.
What I would do as a player is that I would leave a little defence (the free defence of the city might do), and when the city gets captured, I would go back and re-capture it.
Because he's trying to write an AI that can deal with anything, or at least as much as possible.
If we were all to use your logic, why write a video game in the first place?
What should we do about that? What should they do to stop me? Here’s the data we have to work with: It knows I’m the aggressor It knows what turn it is It knows whether I’m a Kingdom or Empire It knows my power rating. What sort of evil should we inject into the function above?
If the power rating is accurate, then it should do one of a few things initially. If the threshold is even or less than aggressor, it should play defensively. Certainly put a lot more emphasis on building more units and probably what will ever give it the quickest power rating in the most efficient manner.
Ff the power rating is equal, if not the AI a little more powerful, it should probably have skirmishing parties of some sort to harass the player. Scouts just for resource harass, small raiding parties for what it perceived as weaker.
If the AI is 50% or stronger, it should probably go for the killing blow. Usually, if the player is on his last limb his power rating will also be smaller so the AI sov or his strongest power rating party should be hunting down the player soverign.
...obvioiusly other wars and hostile roaming units should also play an integral part in this formula. Since this maybe why the player thinks he has the upper hand afterall.
The way I see it is that you teach the AI every dirty trick in the book. Customize some AIs to prefer certain tactics, and add a little random magic. Then you can scale back the list of tricks based on the AI difficulty.
Level 1 militia is there to stop random weak encounters from defeating your city. Adding a single well designed garrison unit and/or city improvements can help a lot in defense without having to station a champion or sovereign. Spearmen are particularly effective. I might even say that they are overpowered when compared to units equipped with clubs or metal weapons.
On normal setting 0.77, the AI might not have enough resources to properly garrison their capital in the early game but on challenging and hard it sometimes does.
You should assume the AI does almost all of what you mention already.
It looks at the world in terms of threats.
But also remember, it only knows what it actually can see.
Excellent. This is one of the most insightful things I've read on the forums recently.
My role as an AI developer is partly to write a competitive AI but it is also partly about providing theater for the player.
That's one of the things I've been working on recently with the AI interacting with players more. Half the balance there is how often it should do so.
One thing I noticed in a game last night was an opponent popping up telling me they weren't quite strong enough to launch an all out attack on me, but they were still going to harass my country-side to the fullest extent (paraphrasing.)
Awesome!
One thing that has always bugged me in a few other video games is that when you become the strongest faction (by whatever ranking system), almost always everyone wants to fight you and even your ties with trusted allies seems strained.
Why doesn't any of the AI try to become your ally in the hope of avoiding an attack? In games where I am a weaker player against other humans or other AIs, I will often try to become their ally when I know I will be hard-pressed to win in a fight. You don't necessarily become a vassal but you see the strongest faction and try to jump on that bandwagon. I would really like to see AI behavior of this sort occasionally. Sure, you or they could always be back-stabbed but that is the risks that are accepted, just as if you went to war assuming that you would be able to win.
I don't want to make alliances or be invited into alliances with the sole purpose 'locking-up' a potential threat (and often back-stabbing that player or AI to take his stuff). I want to be able to build long lasting alliances with both weak and strong opponents, and would want them to build alliances with me, whether I'm weak and even if I'm very strong. To me it just seems much more realistic.
If the AI has two (or more) cities to defend, and you have stacked all your power into one stack, the AI can't defend all its cities. So, what options does it have? Counter-attack. While you are getting his cities, he can get yours. Of course, the AI needs to know where your cities are to do that, and this is a MAD (mutual assured destruction) tactic. At least in the earlier betas it was a very effective tactic to left your cities nearly undefended. So, the AI should exploit that.
The AIs might "spread the word" that you are a scummy player, and not to be trusted. The other AIs could begin using scummy tactics (like building a stack-o-doom and picking your weak cities) if you use such tactics. Of course with the appropriate "diplomatic" messages.
So, basically there are a couple of things I would suggest to be added in the function:
Maybe this is about attack-defense balance? Make early militia get some bonuses. But balance here is crucial, as if you can just spam cities without any need for defenders, that isn't going to be fun either. Conquering cities could be made economically expensive, so you need to pay X amount of money for Y turns to "convert" the city. This way, you could not afford conquering multiple cities in the early game.
I love a villain or an ally with personality. I'm not sure if you've discussed this one, but each faction leader should have some kind of theme song to play when you converse with them and some kind of jingle if they "comment".
When I say comment, I refer to something like a small dialogue box that pops up and says "Don't think I haven't noticed your pioneers on my border, Froggy." I do believe I've seen this in 2x games before. It really makes the other civs feel more alive.
Hmmm a difficult problem.
-Yes diplomatically offering you a non-aggression pact should be the first thing they should do. If that fails they should seek to form alliances with other players, and make those alliances very public so you know what will happen if you attack them. Of course the problem with this is the tech req. for alliances.
-If they are already splitting their groups then maybe alternating champion groups as levelers and city defenders could work. This is what I would do now that XP is split.
-Build outposts near your borders so they can see you coming? They could also have scout units out patrolling safe areas, and roads in particular, outside their borders so you can't sneak up.
-Learn cloudwalk? This is the my personal cheese preference.
-Rather then depending on Faction Power which can be inaccurate how about checking to see how aggressive you have been, ie how many cities you have taken, and how many player units you have killed.
I think the problem here is more the effectiveness of the strategy, rather then the AI. Maybe have a chat with Derek about it.
Ive seen a few comments on the subject on other posts.
Sorry in advance for i am going to quote a "competitor's" game.
I play Civ4 alot and i think they did this personality part fairly well. Each civ has several diplo musics and a vast range of individual animated leaderhead expressions and movements. They did this even better in Civ5 i think ( too bad Civ 5 is not so good in most other aspects). While the diplo with leaders does not actually go that far in term of dialogue depth, each leader has some traits, habits and personality that makes him/her unique and interesting. Sometimes when im playing i really feel like the other leaders are different characters in a story and not just "another AI civ"
Ex: Montezuma is an insane warmonger who can DoW on you for no reasons and he will usually have huge stacks of not so advanced units.
Mansa likes to be protected by a more powerful civ and trades techs all the time.
Isabella of spain is a Religious zealot and will HUNT DOWN THE HEATHENS who are not of her faith.
Each player in FE has its own theme which comes up when you talk to them.
The problem is that a strategy is only good if it is formed by looking at one's resources and more importantly lack of resources. If the AI can do this, it can make good decisions about attack and defense. Often times the best strategy is to tech up because you have such a great starting position.
I would like to see the AI do some hit and run style raiding on the player's resources. Preferably, they would do it on opposite sides of the player's empire, or at least far away from the player's main force.
Maybe if the AI opened several fronts, people would stop complaining/bragging about how their super stack is unstoppable.
Boy was I confused when this topic popped up in my RSS Reader right next to my Arcen AI War RSS feed.
Talk about challenging AI, AI War will put the hurt on you.
One thing I like about AI War is that you are always on the brink of destruction. You always feel like you are a few minutes from being overrun, but each time you barley hung on until you could strike the final blow. Now, AI Wars does use an "cheating AI" in that the AI has unlimited resources, but its amazing balance and unique gameplay system keeps the player on the edge of their seat throughout the game.
For FE, I don't really care if the AI cheats (there should always be an option to turn it off)... but it shouldn't be a blanket multiplier that just gives the AI 2 time more resources like in other games. The cheating should be done in a manner that enhances the gameplay. For example, perhaps the AI knows you are researching advanced knights so it begins to focus on countering that with spear-men (assuming spear-men counter knights of course). This is something the players may never know (or care to know), but it could help the AI make decisions about future moves. If the AI didn't know this, it could be blind sighted and it take 20 turns to create a defense. Or another example: the player destroys the AIs stack of doom along with its most powerful hero. Maybe the AI gets a break on recruiting, leveling and arming a replacement. The cheating in this case is just minor bonuses here and there so that the AI can be competative and put up a fight.
I'm sure others will be passionate about this, so what do you think? I know we all want an amazing AI that is unpredictable and can play as well as a human, but is a cheating AI acceptable if it just enhances gameplay and the player's enjoyment?
Oh, and from a theme point of view, this could be considered a border skirmish type of affair. A slow build up before the city conquering all out war begins. The AI can contact the player, saying that the harassment will continue until a peace is made. Or, perhaps the AI will demand a tribute based on personality.
If three or four AI's were doing this, the player might think twice about going to war with everyone. Even if he is the strongest power around.
Seeing as guerrilla warfare IS the counter to a stack of doom in military strategy, if the AI can't do it, it's pretty near worthless.
You know, I tried to get this to happen to check (just in the event of a bug or something) and it occurs to me I have not engaged in a conversation with another faction yet. I'm sure this is partially because of them dying to the double kill bug, but that would be why their personality may also seem so... inexistent. I haven't even had a chance to interact with them.
Based on the fact that most games don't do it well. It appears to be a non-trivial AI programming challenge. I'm hopeful these guys try it, and pull it off.
Such an AI could also be very, very frustrating to play against. Or at least you would need better UI for guarding your caravans and region in general.
He did ask for evil.
An AI that could intelligently coordinate attacks on multiple fronts would be spectacular. This ought to be available on a higher-end (challenging +) AI setting only, if it were even possible.
Are you asking for us to detect, identify, and handle the cheese in this function? Or can I take it for granted that we know the type of cheese as your describing?
Tactical Battle Question - Why does an AI with Archers facing an opponent without archers or an offensive casting magic casting champion rush to close with the enemy? As a human player I would wait for the opponent so as to maximize the archer dealt damage before melee combat is entered. Can I assume that some Sovereign AI's are smart and will not do this?
AI Strategy Question - Does any of the AIs follow the conquer, plunder, raze strategy; Conquer a city, heal units, then raze before moving on. I would think that Empire AI should be more inclined to do this against Kingdom cities and smaller empire cities not of their race.
There are many great features available to you once you register, including:
Sign in or Create Account
