Does the AI care if it wins?

[bestsss]

^It's true, for instance AIs dont produces ICBM in always peace. (that's supposed to be like joke)
however the options are implemented in the code (taken into consideration while making 'decisions') and they do have effect.

 

[Degenx2]

nothing compares to the BS the AI pulls in every single EA Sports game ever made!!

imagine Charlton Heston in Planet of the Apes when I say DAAM YOU EA SPOOOOOORTS!!!

Oh my god this. FIFA games are just the absolute worst.

'We can't build an AI to save our lives, so why don't we just build an AI for the human's players that tells them to make a massive ing gap in their defense, then refuse to switch to those players when the human is mashing the 'switch player' button'

 

[bestbrian]

Oh my god this. FIFA games are just the absolute worst.

'We can't build an AI to save our lives, so why don't we just build an AI for the human's players that tells them to make a massive ing gap in their defense, then refuse to switch to those players when the human is mashing the 'switch player' button'

Dude, he's talking about Sports Video Games, not soccer.

So what seems to be the verdict? Does the AI know what the VCs are in the base game (I doubt it)? I use BetterAI, but I haven't tested whether the AI has awareness of the VCs there. This is actually a question I've asked in a few different places over the last couple of years, and never got an unambiguous straight answer. I should probably just bounce over to the mod forums and ask a code monkee.

 

[jUmpSt0p]

Oh my god this. FIFA games are just the absolute worst.

Dude, he's talking about Sports Video Games, not soccer.

heeee heee heee

actually, it IS FIFA 10 that I'm playing every day and MAKES ME WANT TO FLY TO REDWOOD STUDIOS AND PUNCH SOME FACE!!!

DAAM UUUUUU EAAAAAA SPOOOOORTTTTS!!!!


... I am actually crazy enough to play on Legendary.

 

[WuphonsReach]

I believe that to be false. Grand strategies like going for a culture win or diplomatic win do in fact rely on the victory conditions being enabled. If I'm mistaken, then perhaps it is only Better AI in which the AIs take the VCs into account.

Not sure, but I know that the Better AI developers over the years have spent time making sure that the AI can chase a cultural or space victory condition. It's still a work in progress, but I've not played a game without some version of Better AI running in many years (going all the way back to Blake's original work).

 

[MajorPOOP]

I had a game go all the way to 2050 tonight. I won by only 150 points over Caesar, who was my neighbor from the very start of the game. We were extremely tight friends, going to war together, long term trades and such. It did feel a bit like cheating that my half of our shared continent was lightly defended from his huge army, which he could have used to crush me and easily "win".

My games allways go to 2050 and I allways lose.......

Thing is im really good at not getting my asrse spanked.... Im just not quick enough...
I play mostly on Prince or Nobel but have had a crack at the harder levels........

I dont get dominated and I manage to stop civs winning a space race or culture....... I just ALLWAYS LOSE on a points decision...

I know its sad but I rarely win..... Im allways playing catchup

 

[jUmpSt0p]

^Major there are lots of strategy articles here that will help you..

this one is useful for the early game

and this thread is very useful for high-level strategies...

I am only on my first game (Prince), and I'm surely going to lose on points in the end I think. however, with all the knowledge I've gotten from these boards (and my first playthrough), I feel confident I will crush my next game!

 

[DrakenKin]

To me it has always been a case of Civilization being a very indepth game which an a.i can't really handle. Firaxis solution to this has been to dish out huge bonuses to the a.i so that even though it is inefficient it can simply brute force its way to success and for the most part, it works.

Yes exactly... this reminds me of DOOM3, if you played on a higher difficulty level the AI just does insane damage, aims perfectly (headshots across the map with a shotgun anyone?), and reacts instantly (to the point where it shoots you before even being able to see you, for instance walking on an exploding rocket just as you are preparing to make a turn and enter its visibility range)... at that point the game completely loses its interest and even credibility.

My point is that yes brute force is sometimes necessary to provide a challenge... but it should only be used when everything else fails. The sad reality is that today's game developers prefer to put their engineers on overkill graphic eyecandy instead of investing into quality artificial intelligence.

The only viable alternative right now is to play against other people in multiplayer... but again it is very hard to find a dozen people to play a game without interruption... But hey maybe they could skip the problem entirely... remove all the asian kids from nike's sweatshops and pay them to be our opponents in civ?

 

[frizzy]

The most sophisticated artificial intelligence ever invented has the intelligence of a dragonfly, and runs on a supercomputer. Technologically, we're a long way from a Civ AI that can play as well as the human player, and if it existed you wouldn't be able to afford it and the machine to run it on.

It's easier to program the AI to react to the human player to try to keep the player from winning. That's how Civ is programmed. Certain development targets are also easy to program: not much to making the AI spam out military units, settlers, spaceship parts, or cultural buildings.

The difficulty levels are managed very basically: the AI plays the same at each level. All that's different are the pace of developments and the starting units.

 

[jUmpSt0p]

^that being said... lots of game developers could still do a better job on the AI with more work.

looking at Civ, there are a many ways they could improve the AI (especially during war) that are completely doable and would drastically improve the game, or at least make it much more challenging..

the AI is rather stupid during war, in obvious ways that I think we all know are possible for Firaxis to improve.

 

[ezwip]

I've often wondered if the AI isn't working together at times. My current game I am next door to Isabella. Catherine and Washington are on a seperate continent right above me. She has my religion and they do not. We both played hungry hippo on China which was between us until he had two cities left and capitulated. China and Spain both have the religion that I founded. I go to war with her to reclaim some of China's cities in an attempt at pushing her back to a choke point. I am far ahead on techs and her muskets are no match for my infantry/artillery. Everything is going well and I only need one more city to accomplish my goal. At this point we are almost equal in land if I can take that city. Washington shows up and claims two cities on the coast. Boom, she capitulates. He does not have her religion. It's not her friend. In fact she still likes me at this point. If I had declared peace I'm sure that I could get open borders in a few turns. It appears to me that she capitulated only to give him a huge advantage. With her selling out he is now clearly in the lead in all categories. I can not catch them.

I also built the Apostlic palace and I'm the current leader. I've already nuked Catherine so Washington can just walk right over her when the dust settles. My only choice appears to be nuking his cities so that he can't possibly get over there. It's just a really bad deal for me. I was on top then boom... just like that. This seems to happen a lot. Another thing I notice is that if a Civ shows up late in the game on a newly discovered continent... the AI's are going to give him every tech possible. In a few turns this will be the new leader even though he's running around with axe's at the moment.

 

[kosiosko 1]

.

I also built the Apostlic palace and I'm the current leader. I've already nuked Catherine so Washington can just walk right over her when the dust settles. My only choice appears to be nuking his cities so that he can't possibly get over there. It's just a really bad deal for me. I was on top then boom... just like that. This seems to happen a lot. Another thing I notice is that if a Civ shows up late in the game on a newly discovered continent... the AI's are going to give him every tech possible. In a few turns this will be the new leader even though he's running around with axe's at the moment.

It could be that late game most if not all other CIVS know them Techs so there value is of little consquence It's happened to me but NOT as much as I've seen the AI get

 

[Manfred Belheim]

I'm quite glad that AI plays to be like real civilizations, rather than to win, otherwise the late game would always degenerate into a desperate all out world war all the time. Kind of precisely the opposite of the real world, with countries (in general at least) looking more towards co-operations and less to war as time goes on.

 

[ezwip]

Isabella's vassal only lasted a turn or two before her land size broke it off.

 

[Junuxx]

Great, another thread full of AI misconceptions!

I know the AI is...artificial. It doesn't feel anything, but I finished my latest game last night. In a loss....

It just seemed that the AI didn't care if it won, that it only cared if I did. Let me explain...

Yeah, your situation would lead me to believe that the AI's keep/raze/liberate decision could use some improvement, as this is particularly silly.

In general however, this game's AI has been pretty much designed to be NOT good on a strategic level. If it would actually search for optimal strategy, that would (1) take ages to compute and (2) leave humans in the dust. Instead it just operates on simple if/then rules, which has given rise to the common misunderstanding that AI is nothing more, can never be more, than inflexible, eternal, written in stone rules of behaviour.

Actually it's a balancing problem of game performance, programming difficulty and challenge for the player. While a more thoughtful, strategic AI would be promising, I think the current paradigm with just some revisions/expansions would do a pretty well job.

AI doesn't care one bit about anything. It throws a pseudo-dice then does something based on the result. Strategic planning like "gift this back and I lose" is not considered or known. The AI is very limited in terms of knowledge on anything outside current turn information.

Indeed, in a way. But you might ask yourself what it actually means to 'care' about something. In this context, isn't it just prioritizing a goal?

computers don't care about anything. Civ AI is all just numbers and probability dressed up to look human.

Wonderful. And we are all just nerves, veins, muscles and organs and bones dressed up to look human.

A list of constituents is not a proof of incapacity. So, your point?

To me it has always been a case of Civilization being a very indepth game which an a.i can't really handle. Firaxis solution to this has been to dish out huge bonuses to the a.i so that even though it is inefficient it can simply brute force its way to success and for the most part, it works.

It is in-depth. Humans can't really handle it either, unless they take like an hour per turn to finetune everything. Even then their planning often breaks down after a few dozen turns.

Again, bear in mind that the ideal computer opponent (for most people) in a game like Civ behaves quite differently from a skilled human player. If the AI would manage to pull off Hall of Fame worthy victories, would you be content about it? Of course the first few times it'd be amazing to see them managing to do that, but I think most people would probably want their placid dummies back.

The most sophisticated artificial intelligence ever invented has the intelligence of a dragonfly, and runs on a supercomputer.

Oh seriously. Where did you ever get this utter nonsense? How do you even quantify the intelligence of a dragonfly? How would you go on comparing it with a computer's? Why would an algorithm (Turing-complete, just calculations that can be done anywhere) that runs on a supercomputer not run on an ordinary desktop? Memory size? There's not that much data going on in a Civ game.

This statement deserves an award, really. Asking for a source is not even necessary.

Technologically, we're a long way from a Civ AI that can play as well as the human player, and if it existed you wouldn't be able to afford it and the machine to run it on.

Technologically you just made this all up.

^that being said... lots of game developers could still do a better job on the AI with more work.

looking at Civ, there are a many ways they could improve the AI (especially during war) that are completely doable and would drastically improve the game, or at least make it much more challenging..

the AI is rather stupid during war, in obvious ways that I think we all know are possible for Firaxis to improve.

Yes, but developing proper AI takes time that game publishers (devs aren't the problem, Sören Johnson seems to be quite knowledgeable about AI) would rather not spend on too unconventional features.

 

[dirtyparrot]

I believe that to be false. Grand strategies like going for a culture win or diplomatic win do in fact rely on the victory conditions being enabled. If I'm mistaken, then perhaps it is only Better AI in which the AIs take the VCs into account.

PoM, I believe that you're at least partially mistaken. A while back, I wanted to see how easy it was to win by culture, so I made it an always peace game. The AI still went about building its stack and upgrading it's stack even though obviously there was no danger in getting attacked (unless you want to make the argument that it was trying to avoid it's cities flipping to me).

 

[Manfred Belheim]

The only victory condition I regularly disable is the Space Victory, and I have noticed it definitely affects the AI. They do still build the Apollo Programme, and then start trickling out spaceship parts, but they do it much slower and definitely seem to prioritise it a lot less.

 

[frizzy]

Actually it's a balancing problem of game performance, programming difficulty and challenge for the player.

Umm, that's what I said. Sure, you could in theory run a much more complex AI on a home computer and not on a supercomputer, I know that. Everyone knows that. Most people wouldn't want to wait hours for each turn to execute. I thought it went without saying that the game has to run reasonably on an affordable computer, and that puts limits on program complexity.

A list of constituents is not a proof of incapacity. So, your point?

The point is that it's not practically possible to program an AI to outthink a human player in a game as complex as Civ.

Again, bear in mind that the ideal computer opponent (for most people) in a game like Civ behaves quite differently from a skilled human player. If the AI would manage to pull off Hall of Fame worthy victories, would you be content about it? Of course the first few times it'd be amazing to see them managing to do that, but I think most people would probably want their placid dummies back.

So you're claiming that the reason the AI isn't more complex is that they don't want to make the game too challenging for players? They're keeping the AI simple on purpose. Sorry, but that's absurd.

Oh seriously. Where did you ever get this utter nonsense? How do you even quantify the intelligence of a dragonfly? How would you go on comparing it with a computer's?

It's called an "analogy" and I didn't come up with it. Sorry your AI isn't complex enough to handle analogies. Didn't mean to bog down your processor.

Why would an algorithm (Turing-complete, just calculations that can be done anywhere) that runs on a supercomputer not run on an ordinary desktop? Memory size? There's not that much data going on in a Civ game.

See above. It has to execute in seconds, not hours, or it isn't marketable.

Also, I can't believe you're seriously suggesting that the game's AI could be programmed to the decisionmaking sophistication of a human mind, if only the developers wanted to.

Yes, but developing proper AI takes time that game publishers (devs aren't the problem, Sören Johnson seems to be quite knowledgeable about AI) would rather not spend on too unconventional features.

Right, because making games more challenging and realistic is a bad thing .

 

[Junuxx]

Umm, that's what I said. Sure, you could in theory run a much more complex AI on a home computer and not on a supercomputer, I know that. Everyone knows that. Most people wouldn't want to wait hours for each turn to execute. I thought it went without saying that the game has to run reasonably on an affordable computer, and that puts limits on program complexity.

There's simply no need for a supercomputer. Most state of the art AI research is performed on a few desktop computers, not supercomputers. At least at my university and at those whose research has reached me.

The point is that it's not practically possible to program an AI to outthink a human player in a game as complex as Civ.

You don't have to outthink the human to outperform him. Yes, humans have a heap of complicated thoughts and strategies and perceptions and intentions while playing Civ, but also many misinterpretations and they miss a lot as well. It is definitely possible to program such a thing. I don't think the speed of the program would be the bottleneck either, rather it would be the added months or years of development.

So you're claiming that the reason the AI isn't more complex is that they don't want to make the game too challenging for players? They're keeping the AI simple on purpose. Sorry, but that's absurd.

Yes, that's what I said. They don't want too advanced AI for a variety of reasons and it's hardly absurd.

It would be very easy for the AI to always check the players' power graph, follow the whereabouts of the player's forces and predict when they are going to backstab whom. They could be taught to warrior rush. Or to really leverage a tech advantage the moment they get it. But I think it's a design choice that they don't, as such features would ruin more fun than that they'd create. The AI plays okay but pretty naive. You learn strategy and tricks to outperform it.

It's called an "analogy" and I didn't come up with it. Sorry your AI isn't complex enough to handle analogies. Didn't mean to bog down your processor.

Analogies are still supposed to make some amount of sense.

Oh, 'analogy' is a synonym for 'false claim' where you live. My processor has recovered, thanks.

See above. It has to execute in seconds, not hours, or it isn't marketable.

I know. I mentioned the demands of the market reality myself. What about minutes?

Also, I can't believe you're seriously suggesting that the game's AI could be programmed to the decisionmaking sophistication of a human mind, if only the developers wanted to.

The criterium is not 'decisionmaking sophistication', which does sound neat and poetic, but performance at playing Civ .

Right, because making games more challenging and realistic is a bad thing .

Too much realism can definitely be a bad thing in a game for everyone, yes.

Too high a difficulty can make a game sell worse, although you could of course tie some AI features to higher difficulty levels only. But then you'd be doing a lot of work for few people.

 

[frizzy]

I guess we'll have to disagree. No, consumers wouldn't stand for a game that took minutes to execute each turn, or even a game that won't run on the computer they bought a few months ago. That's the main reason AI's aren't more challenging.

If almost every customer really wanted easy opponents, then there'd be no money in developing multiplayer capability, since few players would want to play someone that might beat them.

Game development keeps pace with home computer technology. If it were true that developers dumb down games to keep players happy, then we'd still be playing CivI, just revised to have better eye candy. Conversely, a home computer from 1990 won't run CivIV.