Battle Royale exposes the weakness of AI.

[Atwork]

" I haven't seen the AI in civ 5 capturing all capitals."

I'm finishing up a C5 game right now (just in time) in which Greece has conquered 4 capitals, France captured 1, and Rome captured 1. That's in one game!

For C6, I would anticipate a more aggressive AI and more war at higher difficulties, which I don't think we've seen yet.

But, that does not address the issue of whether or not the AI will be satisfactorily competent in the tactical sense, regardless of whether or not it is aggressive, maintains a large/updated military, etc. Due to the complexities of the game, I believe we are still a long way off from seeing an AI capable of challenging a 'good' human player.

But, I expect noticeable and substantial improvement in the AI from C5 or I too will be disappointed. But, even if the improvement is not as substantial as I'd like, it won't keep me from playing. Even if the AI is barely better than C5, the many other components of the game are much improved, and I'm still playing C5 fairly regularly.

Ultimately, if the AI still has a way to go, I'll just up the difficulty and expect a challenge due to the AI being more aggressive, having larger armies, and having advantages (especially early in the game). That's been true of every iteration of the series back to C1; eventually I'll be better than the AI no matter what; and then I just up the difficulty until I'm challenged.

The only way the game will be no fun at all is if there is zero challenge at any point of the game on any level of the game. From what I've seen so far, there's no way the game will be that bad.

And, I think we should have faith in the devs, modders, and the community, who together have never stopped contributing to this franchise.

 

[Wellfooled]

I think its the nature of games now that they get perfected over time. We all want a smart AI right out of the gate, but its likely something that's going to have to be worked on for awhile after release. Civ V improved after it was released, I'm sure Civ VI will do the same. I don't know what difficulty this recent AI battle was set at, but most of what we've seen so far has been on Prince. I imagine that many of the AI's shortcomings in war will be less obvious when they're playing with the boosts that higher difficulties provide. For example if there is an issue with the AI not upgrading units it won't matter as much if their production is boosted to the point they can pump out new units quickly.

 

[What.Do.You.Mean.Its.3AM]

Come on, these are all King AIs, so theoretically they have the same "fighting ability". If there is no obvious early game advantage, it's unrealistic to assume that one certain AI can achieve Domination Victory later on.

 

[Hajee]

If you wish to play the game where one AI takes half the world by the Middle Game then you can play only DOM VC and have that.

But I for like the have most of the civs at the start be there in the end. I want the AI to go after Sci and Culture VC, and understand these VC and plays to that. I want AI that can see other agendas and play the Dipo game, and not just declare war just for the sake of declaring war

 

[ExemplarVoss]

No reason to be a dick, nor a condensing ass hole.

" Just that having an opponent that can sometimes fight is a keystone in civilization games. "
The AI has through out this game fought many times.

So I ask you how often should the AI fight for you to feel comfortable?

Often enough to achieve objectives- particularly get resources. Maybe to upgrade units, perhaps?

But as we've seen in this, warmonger penalties forcing near zero wars are only part of the problem. Rome declared war on Japan twice. Once early on, once with about 15 legions hovering around Antium. The result? The legions stayed hovering around Antium and didn't take the undefended Japanese city. Only one or two even advanced on the city. So 10 turns later, peace was declared with neither AI doing anything.

Spain left tiles burning for 2000 years after they were pillaged, and kept a settler in their capital for most of that period. The number of idle settlers laying around (most of which have been there for 200 turns) is flat out nuts.

Russia's early game consisted of a settler bypassing a natural wonder with a nearby river location to settle at the other end of the continent in the Aztec's face. it delayed their first city by ~20 turns, and gave them a city they couldn't defend (not that that mattered until around turn 400).

All sorts of things are wrong with this AI, not just war.

 

[Hajee]

It's not about how often it declares war. It's about having a sizeable, up to date military; using that army when it furthers it's aims; capturing cities in war; taking advantage of the weakness of others; putting up a semi reasonable fight when attacked; and a myriad of other things. Every evidence we have on Civ6 so far suggests that the AI is 100% hopeless at almost all of these almost all the time.

I disagree with that. The AI civs in this game has had large armys to defend and defend well. This could be one reason why they have not gone for Dom VC. it shows signs are a smart AI, that this see other civs armies and will play smart wars

 

[Morningcalm]

Yeah, I guess you are referring to Japan's mighty army of warriors....

 

[brako]

• AI lead dev has NEVER seen an AI win a domination victory. Is that an okay thing??

I probably ask naïve question, but how often, if at all, hss anyone seen AI wining domination victory in any CIV game, especially Civ5? I usually don't play Deity in standard game (rather Emperor, Immortal) and only rarely Pangea, but maximum that happens is that AI might take 2-3 capitals, but it just doesn't plan and execute a commited campaign to go one civ after another till the end. At least not in time before it choses other victory types. It would be nice to have it at least as an option, aggressive AI that actually tries to conquer...

 

[Panixs]

I probably ask naïve question, but how often, if at all, hss anyone seen AI wining domination victory in any CIV game, especially Civ5? I usually don't play Deity in standard game (rather Emperor, Immortal) and only rarely Pangea, but maximum that happens is that AI might take 2-3 capitals, but it just doesn't plan and execute a commited campaign to go one civ after another till the end. At least not in time before it choses other victory types. It would be nice to have it at least as an option, aggressive AI that actually tries to conquer...

The problem here is we are talking about the lead AI developer here who has run thusands and thousands of AI only sims. Not only has he in all them thousands of sims never seen a dom victory he has never seen anyone take more than three capitals.

 

[smartcanuck1988]

To be fair we'd probably see the same thing if we did AI-only battles in all Civ editions. That said, I think it's good that the devs are doing these AI battles as it exposes some of the issues. But what I find funny is that in the early interviews the devs said they had an AI-only game running at the studio so they can watch how the AI behaves and make the necessary changes. When I heard that, I was very happy because it sounded like the devs were taking the AI very seriously.

But if what we saw today is what the devs believe is a good enough AI that they felt like showing it off in this stream (which I don't recall seeing many requests for before they announced it), then I'm a bit concerned with what the devs think is good regarding the AI. Maybe some of us just have a different idea of how the AI should play.

 

[ExemplarVoss]

To be fair we'd probably see the same thing if we did AI-only battles in all Civ editions. That said, I think it's good that the devs are doing these AI battles as it exposes some of the issues. But what I find funny is that in the early interviews the devs said they had an AI-only game running at the studio so they can watch how the AI behaves and make the necessary changes. When I heard that, I was very happy because it sounded like the devs were taking the AI very seriously.

But if what we saw today is what the devs believe is a good enough AI that they felt like showing it off in this stream (which I don't recall seeing many requests for before they announced it), then I'm a bit concerned with what the devs think is good regarding the AI. Maybe some of us just have a different idea of how the AI should play.

That is pretty much my take away from it. They DO know. They just find it acceptable to be awful at their jobs.

 

[Furycrab]

I'm finishing up a C5 game right now (just in time) in which Greece has conquered 4 capitals, France captured 1, and Rome captured 1. That's in one game!

How many players are you putting on one map?

Either way... Isn't that kinda a testament to bad AI? Sure one guy was successful at beating up a bunch of other people, but it means the other AIs were just as horsehockey at defending their own stuff.

 

[rpgalon]

I see some of you defending this really bad AI by saying Civ is not only about war, well, but the problem is, for example, Rome had a legion army that could conquer the whole continent in the early game, why did they build that if only to do nothing? this a huge issue.

The same could be said about the science victory where the civs would do nothing even if they could launch a module to mars 100 turns earlier.

 

[Cyon]

Maybe the AI bonuses should be changed /added; 50% discount on unit maintenance and upgrades. And add flavor for that. Also if built settler then settle a spot so long that it don't suck..

 

[Txurce]

smh

again if u wish to play a grand ARMY always fighting go play Total war games.

in the stream the AI has a large army is well defended.

The AI is choosing to win by SCI and CUL, thus does not need war.

ALL of the AI decided it didn't need war. Including designated warmongers like Montezuma and Gorgo. Right.

 

[King Jason]

They should just scrap strategic resources. They can't get them right in a meaningful way.

Turn them into bonus resources that - at best - give you cheaper upgrades and retain the tech boosts for being able to improve them. So having Iron gets you an army of swordmen quicker - that's it.

In a game like civ, that's a big enough advantage. But depriving it from half of the players is just stupid.

 

[pietro1990]

Maybe the AI bonuses should be changed /added; 50% discount on unit maintenance and upgrades. And add flavor for that. Also if built settler then settle a spot so long that it don't suck..

In civ 4 the AI got free upgrades for units. so yeah that could work

 

[ExemplarVoss]

Come on, these are all King AIs, so theoretically they have the same "fighting ability". If there is no obvious early game advantage, it's unrealistic to assume that one certain AI can achieve Domination Victory later on.

Kinda. But, statistically, if they're running thousands of simulations (which is what they made it sound like), even just fluking out, it should happen at least few times when a civ snowballs and just overruns people. That it NEVER happens is... concerning.

They should just scrap strategic resources. They can't get them right in a meaningful way.

Turn them into bonus resources that - at best - give you cheaper upgrades and retain the tech boosts for being able to improve them. So having Iron gets you an army of swordmen quicker - that's it.

In a game like civ, that's a big enough advantage. But depriving it from half of the players is just stupid.

I totally agree. This system is really terrible, and isn't tactical or strategic in any fashion. Pure RNG, and the fact that it utterly cripples the already bad AI makes it pretty intolerable.
That they realized its bad enough that they just dropped resource requirements from unique units is pretty telling. They should just be bonus resources like rice or copper.

 

[Furycrab]

I see some of you defending this really bad AI by saying Civ is not only about war, well, but the problem is, for example, Rome had a legion army that could conquer the whole continent in the early game, why did they build that if only to do nothing? this a huge issue.

The same could be said about the science victory where the civs would do nothing even if they could launch a module to mars 100 turns earlier.

I haven't had time to watch second half, had to tune out after turn 180, but it looked like Japan managed to get Samurai which would have defended quite nicely against the Legions. Looked like Rome missed it's timing window because it didn't manage to expand enough before hand.

 

[Txurce]

I totally agree. This system is really terrible, and isn't tactical or strategic in any fashion. Pure RNG, and the fact that it utterly cripples the already bad AI makes it pretty intolerable.
That they realized its bad enough that they just dropped resource requirements from unique units is pretty telling. They should just be bonus resources like rice or copper.

This won't happen. But it's easy enough to have enough resourceless units that lacking a certain resource just makes you have to scramble. If warriors upgraded to spearmen, and so on, all the way up, there'd be no problem.