[GS] Domination, air combat, AI and the new era

[glider1]

I cannot give Firaxis the "benefit of the doubt" it has been too long. Even with the game on a huge sale right now, it is still very expensive when the price of the next expansion is included. I would have bought the game but without a competitive AI, there is no point paying big money for a puzzle game that becomes an empty experience so quickly. A good AI makes the game magic.

 

[AriochIV]

If Fireaxis offered a DLC for $40 that just focuses solely on AI improvements, would you buy it?

No, for two reasons.

First, it's ridiculous to charge extra for functionality that should be in the base product. It shouldn't require an addon for the AI to be able to use units that came with the base game.

Second, there is no way to quantify AI improvements. They could hand you an empty box and say, "this improves the AI," and there is no way you could prove them wrong.

What's really frustrating about the air combat AI is that, as far as I can tell, it has gotten worse over the course of the patches and expansion. Early on, the AI did use aircraft... not well, of course, but it did build fighters (and more rarely bombers) and use them for ground strikes. Then for a while it would build only a few, and just fly them from place to place without attacking anything. Now it doesn't seem to even build them anymore; it still occasionally builds aerodromes, but I don't think I've seen a single AI aircraft in the last 5 or 6 games I've played.

For that matter, I've noticed a pattern over the same span where many AI civilizations reach the industrial or modern era and have no army. There was a case in my last game where Poundmaker declared war on me and had literally zero land units. There may be a number of factors at play here, but I think the Corps and Army system may be largely to blame. I think it was intended to ease the pressure on the AI by allowing it to manage fewer units, but I think it has backfired... because the AI is poor at keeping its units alive, putting all its eggs into fewer baskets has made it easier for the AI lose them all.

But I digress...

 

[Cagarustus]

If it is real, and the improvements are very noticeable and provide a real challenge, probably... but my point was this: AI improvements should have been part of at least every major expansion, and they were not... they should be part of the 40$ new xpack, and everything points to it being ignored. Why should I work for free to fill in the huge emptiness regarding AI when they are charging for it and not delivering?

No you’re absolutely right. I see your point. I think they will make some changes. They indicated in the video and on the webpage that some more tweaks will be made. Hopefully the AI is one of them.

Apologies if I came across as being as a prick.

I think one of the main reasons why they haven’t focused on the AI is because they don’t want to discourage new/casual players from playing the game by rage quitting every time. Also, lots of civ players play multiplayer so the AI doesn’t really affect that.

 

[Cagarustus]

It seems like at first, we were all super excited about the expansion and all the new features it promised. Then after thinking about it for a couple days, we've realized that the AI isn't competent enough to handle the current game yet so why would they be able to handle all these new features? I think we'll probably see a dramatically improved AI maybe next summer or so, several months after the game has come out.

That’s a good point. They’ll probably focus on fixing the AI once all feature and mechanics have been implemented. Although I’m not a game developer I would assume that it would be a waste of time to get the AI great in vanilla only then have them not adjust to new mechanics etc.

Get all the mechanics down first, then focus on the AI.

 

[Cagarustus]

No, for two reasons.

First, it's ridiculous to charge extra for functionality that should be in the base product. It shouldn't require an addon for the AI to be able to use units that came with the base game.

Second, there is no way to quantify AI improvements. They could hand you an empty box and say, "this improves the AI," and there is no way you could prove them wrong.

What's really frustrating about the air combat AI is that, as far as I can tell, it has gotten worse over the course of the patches and expansion. Early on, the AI did use aircraft... not well, of course, but it did build fighters (and more rarely bombers) and use them for ground strikes. Then for a while it would build only a few, and just fly them from place to place without attacking anything. Now it doesn't seem to even build them anymore; it still occasionally builds aerodromes, but I don't think I've seen a single AI aircraft in the last 5 or 6 games I've played.

For that matter, I've noticed a pattern over the same span where many AI civilizations reach the industrial or modern era and have no army. There was a case in my last game where Poundmaker declared war on me and had literally zero land units. There may be a number of factors at play here, but I think the Corps and Army system may be largely to blame. I think it was intended to ease the pressure on the AI by allowing it to manage fewer units, but I think it has backfired... because the AI is poor at keeping its units alive, putting all its eggs into fewer baskets has made it easier for the AI lose them all.

But I digress...

Presumably the amount of effort that would go into the coding of the AI would warrant a new dlc. I read somewhere that content is cheaper to design that coding in all games, which is why many game developers focus on new content rather than on coding.

It would make sense to me that improving the AI would be done after all content is done, so that it can be the sole focus of the team post civ 6.

 

[Malachi256]

I'm ready for civ to get a reboot by a different developer. The same way simcity was usurped by cities: skylines (their awful DLC business model is a different discussion).

I want an AI that can play the game. I want the early game to be as compelling as it is now, and the end game to be actually compelling.

Firaxis will never be this developer. They are entrenched in the "iterate and beautify" model that milks a customer base, but with this the genre never evolves, it stagnates.

 

[Cagarustus]

I'm ready for civ to get a reboot by a different developer. The same way simcity was usurped by cities: skylines (their awful DLC business model is a different discussion).

I want an AI that can play the game. I want the early game to be as compelling as it is now, and the end game to be actually compelling.

Firaxis will never be this developer. They are entrenched in the "iterate and beautify" model that milks a customer base, but with this the genre never evolves, it stagnates.

It really comes to down to business decision. Most casual players are attracted to visually appealing games. I know I am. That’s why I won’t play Civ 5 or Civ 4 even if the AI is better.

And as some people here have said, making the AI better is subjective and may make them much harder thereby turning away many casual players. I have no doubt they’ll do some minor changes, particularly around late game units since the new dlc is focused around the late game. I expect to see these changes later on. Furthermore if they allow dll modding then modders can contribute to this.

 

[Tech Osen]

I'm reasonably satisfied with the Airpower mod. It does make the AI build a lot of planes. They still don't use the fighters very well offensively but at least you have to think twice before sending in your bombers. They do use bombers offensively and it has cost me some battleship armadas.

 

[ChocolateShake]

I currently use a mod to increase the time of each era, and another one to reduce unit costs. This allows the AI to make a ton of units, and it actually gets pretty challenging. Cleopatra took Athens like 2-3 times using her naval units, meanwhile I struggled to take a city from Gilgamesh (after he declared war on me, usually I play peacefully).

And this is on emperor. If I set it to immortal, the combat odds would get much tougher and I doubt I'd be able to have taken a city at all without a much more structured and coordinated effort. That's the thing though I think, eventually players will always be able to outsmart the AI, until we get an AI that can play almost exactly like a human would. That would be fun, but it'd also make the game a lot more warlike.

The AI needs help with its victory conditions in total - I've won a few science and cultural victories that I felt I should've lost if the AI was more focused on winning. I'm not sure if this is by design or not.

Almost all 4x games face this problem of having an AI that can't fully challenge the player. If there is a game out there that can reliably beat a player more than say 20% of the time, I'd be interested in trying that out. I'd love to see more AI investment in the future, and we do have to make it known that we want a better AI.

It seems like at first, we were all super excited about the expansion and all the new features it promised. Then after thinking about it for a couple days, we've realized that the AI isn't competent enough to handle the current game yet so why would they be able to handle all these new features? I think we'll probably see a dramatically improved AI maybe next summer or so, several months after the game has come out.

I agree. I'm still very excited about the expansion. I think in the future the AI will be improved or the DLL will be released, and we'll see a significant leap in the AI's abilities.

I'm reasonably satisfied with the Airpower mod. It does make the AI build a lot of planes. They still don't use the fighters very well offensively but at least you have to think twice before sending in your bombers. They do use bombers offensively and it has cost me some battleship armadas.

I'll have to give it a try

 

[Stilgar08]

I'm so glad someone opened this thread and that it wasn't me. I'm not alone after all and I wholeheartedly agree with OP!

If Firaxis offered a DLC for $40 that just focuses solely on AI improvements, would you buy it?

Yup!
Why? Because it's a key-thing and affects almost every aspect of the game! BlingBling sells, AI helps keeping u engaged and stick to the game.
I love CiVI. Much more than I ever loved CiV , but I have much less hours under my belt. Compared to the same duration I own the game and I have high hopes GS will bring me into the 1.000hour Club.
AI has been a major gripe from the start, not only fanwise, but in the reviews as well... It's either too hard to fix or just not sexy enough. But it is sexy. You'll see it if it's not working as intended...

It seems like at first, we were all super excited about the expansion and all the new features it promised. Then after thinking about it for a couple days, we've realized that the AI isn't competent enough to handle the current game yet so why would they be able to handle all these new features? I think we'll probably see a dramatically improved AI maybe next summer or so, several months after the game has come out.

Why? I don't see reason for that They either fixed it remarkably now with GS or not so much. I'm expecting AI improvements, but I doubt they'll be substancial enough. Fxs has a good record of updating and patching their products for a long run, that's granted. But since this AI bugged us for so long they should be able to get it to a decent state with this XP.
CiV was not a bit better in this regard but AFAIK it was the case, that BNW fixed alot of AI.
What happens after modding is finally unleashed is another thing but I don't want to wait another 2years for that. I'm a Star Citizen I'm fed up with waiting...

No you’re absolutely right. I see your point. I think they will make some changes. They indicated in the video and on the webpage that some more tweaks will be made. Hopefully the AI is one of them....
I think one of the main reasons why they haven’t focused on the AI is because they don’t want to discourage new/casual players from playing the game by rage quitting every time. Also, lots of civ players play multiplayer so the AI doesn’t really affect that.

MP in CiVI is still a very small minority I'd say. And the vast majority plays the long Single game instead of MP scenarios. Civ just isn't MP in its core.
My fear is that Fxs tries to get alway with its bad AI by deflecting. "Ok, AI still sucks but you play against the map now and puzzle your way though to victory..." ok, cool! But...

...
And as some people here have said, making the AI better is subjective and may make them much harder thereby turning away many casual players.

That's what difficulty is for not stupid AI...

 

[Tech Osen]

It seems like at first, we were all super excited about the expansion and all the new features it promised. Then after thinking about it for a couple days, we've realized that the AI isn't competent enough to handle the current game yet so why would they be able to handle all these new features?

I reached the point where I just kindda accepted the AI shortcomings, at least to the point that it's not ruining my enjoyment. So it's certainly not dampening my excitement for GS. That's not to say I wouldn't very much welcome improvements in that area.

 

[King of Prussia]

i find the late game armies/planes for the AI to be hit and miss. From what i seen, part of the issue is that the costs of units outpaces production to the point where you need the cards + a very high production city to get units out in a decent time. or you have to rely on money and faith to buy units. all of which the player will do far better than the AI. I modded my files to change how long it takes units to be made. in the late game, i am knocking upwards 100-150 production off some units. I find the pacing to be better. I began seeing more units from the AI, and even more air planes. There are still times where the AI seemingly has no troops, and idk why they aren't building any. or in other cases, you kill off their army, and they are too slow to get a new one. This tends to be made worse in the late game, when sieging is a lot easier, thanks to arty + balloons and planes so you can take their cities before they can get a new army.

and i will add that the AI doesn't use its planes well. sometimes it just flies them back and forth between different cities/bases without doing anything.

i would also add that i think tech/culture can really outpace production, making it hard to "keep up" with the latest stuff, as by the time you build it, there are 5 more new things you want to build.

 

[Trav'ling Canuck]

First, it's ridiculous to charge extra for functionality that should be in the base product. It shouldn't require an addon for the AI to be able to use units that came with the base game.

For someone like me who really enjoys the civ series and would get many hours of entertainment from a civ game with a better AI, though, the base game is underpriced. If the economics are such that Firaxis needs to extract more funds from people who value a better AI in order to invest in that, I'd be willing to do so.

And yes, it would be a leap of faith as to whether they deliver, for anyone who pre-ordered. They'd still have an incentive to deliver, though, as early reviews would be what drove whether they got later sales.

I think one of the main reasons why they haven’t focused on the AI is because they don’t want to discourage new/casual players from playing the game by rage quitting every time.

I remember when you started Civ on Settler and worked your way up to Prince. But then, I'm old and bad at computer games.

Anyway, to my mind, the best way to improve the experience for new players is to improve rules consistency from one system to another, to ease the learning curve, and improve the quality of the information presented to the player. Keep the AI aggression low on lower difficulty levels, but otherwise have them play well and new players might even be able learn a few good tactics from the AI, rather than bad ones.

I'm ready for civ to get a reboot by a different developer.

I'm ready for a reboot, mostly because I think a lot of unnecessary complexity has worked its way into the underlying mechanics. I don't care who reboots it.

 

[Roxlimn]

LOL. This again? Civ AI has always been bad. We have this discussion about every iteration since ever. That's why both Civ V and Civ IV have community-based AI mod enhancements. Heck, Civ4 AI was so bad on release, it wouldn't even improve its own tiles, right into the endgame. Basically, don't hope for good AI. Firaxis has never sported or intended for good AI for Civ. The game's development and priorities are clearly not in that area.

There are developers who prioritize AI. Galactic Civilizations has decent AI, not that many people really want that, apparently.

 

[Aristos]

They could hand you an empty box and say, "this improves the AI,"

They are doing that already.

 

[AriochIV]

They are doing that already.

They haven't even mentioned AI in the new releases, that I've seen.

 

[Aristos]

They haven't even mentioned AI in the new releases, that I've seen.

Exactly. Empty box @ 40.

 

[Xmonger]

I love these discussions about AI ... by day I'm an engineer working in AI - a competent one hopefully. Machine learning specifically. I keep saying that Civ isn't a game that you can use machine learning for. All the news you hear about AI is in ML/DL. So that leaves an expert system, which is a type of AI we abandoned decades ago, because they're horrifically expensive to make and fragile. All that aside, it's a little naive to think that Fraxis wouldn't put more resources on AI if there was some low hanging fruit on it. If there is some low hanging fruit, then I'd believe it's some corporate culture issue that is preventing them from doing much here (such as fear of touching the existing system or something). There certainly isn't some complicated conspiracy reasoning about how they can get 'away with' not doing what customers want.

Just two bits ...

 

[glider1]

I love these discussions about AI ... by day I'm an engineer working in AI - a competent one hopefully. Machine learning specifically. I keep saying that Civ isn't a game that you can use machine learning for. All the news you hear about AI is in ML/DL. So that leaves an expert system, which is a type of AI we abandoned decades ago, because they're horrifically expensive to make and fragile. All that aside, it's a little naive to think that Fraxis wouldn't put more resources on AI if there was some low hanging fruit on it. If there is some low hanging fruit, then I'd believe it's some corporate culture issue that is preventing them from doing much here (such as fear of touching the existing system or something). There certainly isn't some complicated conspiracy reasoning about how they can get 'away with' not doing what customers want.

Just two bits ...

There certainly is low hanging fruit. The AI is reasonably easy to improve now that the basic AI mechanisms are in place and has already been demonstrated as greatly improvable for 1UPT in the VP mod with hand coding. The enthusiasts worked on it in their spare time and not necessarily with full computer science skills and with no pay. The Civ6 AI could be improved significantly by just hiring one final year undergrad from a computer science degree for six months that could make improvements as part of a final year project. The computer science aspect of the AI is not that hard for an undergrad and it would not cost much and the technology is already two decades old at least.

Problem is that Firaxis never get around to it because they are always adding features to increase sales and you cannot improve the AI properly until it is feature complete. In reality, it could be improved if they wanted too but I think they believe hiring an AI coder is too much of an impediment to work flow (profit).

 

[Roxlimn]

Eh. No. Firstly, I really don't think AI is that much of a thrust for Civ since its release. Even the Civ 1 AI was dumb as bricks, and that was an extremely simple game. But the most important factor is the one already mentioned. You can't hand-craft an AI algorithm for an incomplete ruleset. The rules have to be complete and unchangeable before you can make anything like a credible half-decent AI of that nature. This means it is necessarily work done after the last expansion is released, and for no profit. Firaxis has always been a for-profit company, and it doesn't sponsor games or ongoing sales models like Blizzard. If they were to charge monthly for Civ, then there would be a reasonable incentive for them to improve the AI after the game was considered finished and feature-complete.

It does bear mentioning that some games sport better AI even with incomplete rulesets. The GalCiv ones, of course, but Endless to some extent. Both games are not modeled after human-centric board games like Civ is. They're specifically created to be playable by a pre-conceptualized AI. The AI is built first, then the game is built around it. To some extent, Civ's success has always been about how much like a board game it is, and how fun it is to push toy soldiers around. I still play the game that way, and demand no more of it. Some argument can be made that since online capability is now enabled, MP balance should be the driving force behind Civ design, and then AI can be made for it.