Status of Civ 6 AI

[dexters]

Did they do something to the AI?

Getting Civs running at 100 bpt fairly early in game. I just lost my 2nd immortal game in a row to a science runaway on Sumeria (granted I'm not great with Sumeria and I had an island start)

Post summer patch, AI also seems to value lux highly when they are going for science win, trading away culture items they no longer need, negating the rebellions that would torpedo them in the past.

 

[Archon_Wing]

But the point isn't that the actual functioning of the AI is coded to improve at higher difficulties, it's that it reaches a certain tipping point where the AI feels better being about aggressive and is more successful at doing so.

I'm already well aware that the AI doesn't get smarter; it just gets bonuses to compensate for its lack of ability, as I already said:

Just because they get bonuses to mask said stupidity, doesn't change that.

That's why I stick to examples that are largely unrelated to tactics or difficulty.

Point is, it's something that really should change.

 

[Acken]

Bonuses in themselves are normal and expected but you'd still expect the base AI to not be too poor. If your base AI is good enough then bonuses will make the difficulty more natural. Otherwise you have to go crazy on the bonuses and it becomes ridiculous like the 2 free settlers of Deity Civ6 iirc compared to 1 in Civ5. Or unwanted effect of bonuses like the inaccessibility of early wonders in Civ5. These difficulty bonuses are not really "fun" because they lock out a part of the game and break the idea that the bonuses are simply there to level the playing field toward better players.

 

[AntoineS]

Civ 6 AI is not good as is and needs drastic improvement, I think we all agree. I am worried that they never fix it enough for Civ 6 to really be worth it by the end of its life cycle, though.

I´m not too worried about content, I dare to hope that they will do a good job on that. However, by the end of Civ 6´s life cycle, I am concerned that the game still won´t be as good as it could be because of this bad, not fixed, AI.

My point and ultimate concern is that I don´t want Firaxis to come out with a selling point for Civ 7 that goes like: "now with the experience we got from from Civ 6 running on the same engine, Civ 7´s AI is gonna be good", whereas we would have to wait 4 years to get all the content Civ 7 would need to feel complete, and while we would have dumped a lot of money on Civ 6 that would not really be worth it because of its AI.

I don´t know what you guys think but that´s my main concern at the moment.

And I mean it´s not like we´re asking for the moon... I think if they brought the pace of the game and the AI up to Civ 4´s level, it´d be enough to be a fun game at least (I know I had fun with Civ 4, the AI part of the game was solid enough for the experience to be immersive) even though it would not reach its maximum potential.

It´s something they´ve done in the past... over 10 years ago! I think it´s sad that they carelessly and certainly to some degree purposefully allow for the game to be broken like that (driven by "marketing and financial strategies"). Modders do better... they really have no excuse

 

[oedali]

So given that:

1. It has been proven already that AI CAN be made good at tactical combat and empire management (eg Civ 5 VP mod)

2. Root cause is that very few resources are assigned to Civ 6 AI and not a priority as it caters to us 1% competitive players

3. It is still TBD when the game DLL files will be opened up to modders (could be years from today)

Why don't we start a petition to pool some $ and give it to Firaxis to hire an extra dev for x months focusing exclusively on the AI, and take input from gaming community? I would be willing to pitch in for something like this because it is worth so much to me. This could work keeping in mind so many Civ veterans like me are in their 30s or 40s and have the financial means. Has something like this ever been done?

 

[calad]

So given that:

1. It has been proven already that AI CAN be made good at tactical combat and empire management (eg Civ 5 VP mod)

2. Root cause is that very few resources are assigned to Civ 6 AI and not a priority as it caters to us 1% competitive players

3. It is still TBD when the game DLL files will be opened up to modders (could be years from today)

Why don't we start a petition to pool some $ and give it to Firaxis to hire an extra dev for x months focusing exclusively on the AI, and take input from gaming community? I would be willing to pitch in for something like this because it is worth so much to me. This could work keeping in mind so many Civ veterans like me are in their 30s or 40s and have the financial means. Has something like this ever been done?

Because Firaxis doesnt have to do that. Civ V sold well, Civ VI sold also well. As long as it is marketed proberly (graphics! who does not love dramatic Cleopatra?) sales are guaranteed. Meanwhile they can cut other expences, like AI development.

You can complain here, but +20k people are playing Civ VI every day according to steam charts. Anything other it is just an unnecessary expence. Only way to break this cycle is to invent a cheap way to implement AI, but likely some yet non existing company wil do this in the future because Firaxis is doing just well on their own mind.

 

[oedali]

You are right they have no incentive to do that. I think the most realistic and likely resolution will be for them to release the SDK have the modders fix it, just as they did with Civ 5. I just hope they do it sooner vs later, ie months vs years..

 

[calad]

You are right they have no incentive to do that. I think the most realistic and likely resolution will be for them to release the SDK have the modders fix it, just as they did with Civ 5. I just hope they do it sooner vs later, ie months vs years..

I dont want to be a jerk, but making code open and modable is another expence. Most of the players will never use any mods, nor will demand them. Even MP is neglected. In Civ V it never worked out well, dont know about Civ VI but MP section is just dead.

Also I dont have experience on graphic design, but I can imagine making new leader animations in Civ VI is significantly more difficult than in Civ V or Civ IV. So even the most marketed element is geared to be as little modfriendly as it can be. This is the reason why Civ IV has tons of new civs, but Civ V barely few.

 

[Gehennas]

Don't know if AI became better or worse, but it definitely became weirder.

1. AI is now trying to trade great works. However, it does it the strange way:
- Nubia and Germany constantly want to buy my great writings for luxuries and some gold. This deal is bad and actually becomes annoying after some time.
- AI sometimes suggests to exchange great writings. This deal is ok, I think, but totally pointless. I detect a lazily-written AI here.
2. "Nuclear weaponry is the future" on 40th turn as already reported multiple times.
3. This time two AIs declared joint war on my while being on the other side of the map. I haven't seen this before. Usually one of AIs was already at my border. I've signed peace with both of them without actually seeing any of their units.
4. Combat AI is still awful - nothing new here. Had a very long war agains Australia when AI kept creating cavalry and charging it into my encampment ignoring my unprotected cannons. Actually, with huge tech disadvantage (I think I built only one campus this game) I couldn't capture more than 1 AI city. Nevertheless, AI gave me 4 cities in peace deal.
5. AI has stolen a great work from me once. I Don't think I've seen this before.
6. AI became more friendly overall. However, it usually loves or hates me for no reason. It's quite weird to see a nearby happy Alexander declaring friendship with me while John Curtin DOWs me three times in a row. I understand, that this is usually related to hidden agenda, but this makes no sense.

Off-topic: I've lost my faith into Firaxis. I was going to pre-order the expansion for XCom-2, but now I think I'm going to wait for trustworthy reviews.

 

[Zuizgond]

Because Firaxis doesnt have to do that. Civ V sold well, Civ VI sold also well. As long as it is marketed proberly (graphics! who does not love dramatic Cleopatra?) sales are guaranteed. Meanwhile they can cut other expences, like AI development.

You can complain here, but +20k people are playing Civ VI every day according to steam charts. Anything other it is just an unnecessary expence. Only way to break this cycle is to invent a cheap way to implement AI, but likely some yet non existing company wil do this in the future because Firaxis is doing just well on their own mind.

Yep, since civ 5 all budget goes into graphics, animation and voice acting. Oh and also creating new civs that can be sold as 5$ DLCs. This game is not for me. I no longer belong to the targeted playerbase obviously.

Hope a new company will start creating 4x games with a focus on gameplay and AI rather than shiny stuff.

 

[Ramkhamhaeng]

I also assume that the next good Civilization-like game will not be created by Firaxis.

 

[DrJambo]

In my King game, Germany had conquered two capitals and was looking strong. Monty has also conquered a civ so things were looking promising. However, they didn't have much of a navy or airforce to speak of, and so when the push came to the shove, my invasion eventually took them out. Yeah it was fun and the military attempt was a lot better than previous iterations, but the airforce that they had built was still piss poor. If there is a function to the patrol action, the AI needs to know this... Battle of Britain this will never be.

Also like others have said the diplomacy has taken a few steps in a variety of directions and mostly for the worse. It has the feel of something that hasn't been tested thoroughly enough:

1) I now get offered deals by the AI that don't have an accept button, which really means the AI doesn't want the deal - wtf!?
2) I can't get any form of Alliance with anybody under any circumstances, and given I was playing as Gilgamesh that was a real downer for his civ ability...
3) They want artwork now, which is good, only the deals are very perplexing at times - like swapping a Great Writing piece with a Great Writing piece...
4) I took Brussels as part of a peace deal with Rome. It remained in an 'Occupied state' for the rest of the game with no way to get it 'unoccupied'. I imagine it is because it was a City State.

The previous patch seemed to have a better, more functioning, diplomacy.

 

[Spudsie74]

Designing a good AI for a complex game is always difficult; given the changes from V to VI, as well as the way DLC's and expansions are released, it is tough to provide a decent computer opponent. Multiplayer is a whole lot easier (Multiplayer Civ still has a dumb AI, the issue is not as critical). Granted, this AI is exceptionally stupid, but I don't expect Firaxis to have a good AI anytime soon...too many other priorities (DLC, Exansion, bug fixes, updates......). I'm not of fan of the system, but since Firaxis is in business to make money, they are going to budget their design resources accordingly. From what I have read, most people who own a copy of Civ. don't play at the higher levels, nor do they amass 1000+ hours playing it (I have 1600). Firaxis is just following their business instincts. That being said, it is frustrating for a dedicated player.

Civilization games are an evolving system, moddable; a work in progress. Since experienced players (who else posts stuff here?) tweek the base game (w/mods), there is hope.

Some of the player groups (Filthy Robot & No Quitters come to mind) have a mod/rule set that, while optimized for multiplayer, seems to work fairly well (or it did w/Civ 5). It will take a while, but the dedicated fan base will develop (possibly w/some help from Firaxis) evolving improvement, even with the AI. One tiny step at a time.

It would be nice if they had a good AI, but that's just not going to happen in an evolving, extremely complex game with so many different facets. (Most of us enjoy playing, no?)

 

[earlc]

I don't think "they focused on X instead of Y" is necessarily fair. The caricatures of the leaders was a fantastic move for the series - makes them more memorable and identifiable, and probably influenced by games like the recent Street Fighters. That sort of thing really sticks out to people, and isn't really something they could come back and fix later.

From what I know about the industry, budget was most likely the defining issue. I don't think that they don't care about good AI or reasoned that it wasn't worth their time, but work on it was probably limited by budget after they fleshed out the items they decided would be defining of this generation.

I still have hope it will be adequately improved this generation, but on the off chance it isn't, I see no reason to expect the formula won't be refined in an eventual Civ VII.

I'm already well aware that the AI doesn't get smarter; it just gets bonuses to compensate for its lack of ability, as I already said:

That's why I stick to examples that are largely unrelated to tactics or difficulty.

Point is, it's something that really should change.

Yes it should. But if the combat aspect of the game is greatly improved by something as simple as increasing the difficulty (even without the combat strength bonuses), it seems like that's something that should be easy to balance. If we had Immortal-level armies on Prince, the game would certainly be better even if the targeting is still dumb and the AI can't figure out how to effectively use air and navy. As a third example, Harry just invaded my little continent from across the sea with probably 6 or more knights and a handful of crossbowmen. The attack was actually kind of difficult to repel. If he has actually sent ships along with the army as AI civs should, I would have lost a city or two.

By contrast, Prince wars are usually accompanied by 0 units for some reason, or a handful of weaklings sent to die and maybe pillage a little while you sit comfortably within your borders and focus on more important things. Which is awful to the point that war is almost a cheat code at that level.

 

[HF22]

I am not persuaded resourcing is the problem with the AI, so much as overall game design.

I think it is significant one of the differences between Civs 1-4 & Civs 5-6 is that the lead designer stopped being responsible for AI.

You could double the Dev resources on AI, but if you're not designing the game around the weaknesses of your AI approach (& all approaches have weaknesses), the AI flaws are going to be ruthlessly exposed.

 

[PhilBowles]

I don't think "they focused on X instead of Y" is necessarily fair. The caricatures of the leaders was a fantastic move for the series - makes them more memorable and identifiable, and probably influenced by games like the recent Street Fighters. That sort of thing really sticks out to people, and isn't really something they could come back and fix later.

I really wish they would fix it, frankly. I much preferred Civ V's leaderscreens to Clash of Clans. If they want the AI civs to be memorable they should give them distinct AI personalities as Civ V and to an extent Civ IV did - the complete absence of individual AI personalities keeps me from coming back to Civ VI (yes, once again I got bored around 150 turns in, if that) more than the AI's general performance or the game's mechanical shortcomings, or even its poor interface.

 

[Gehennas]

Civilization games are an evolving system, moddable; a work in progress. Since experienced players (who else posts stuff here?) tweek the base game (w/mods), there is hope.

Unfortunately, moddability is a double-edged sword.

First of all, modding allows to release a raw, soulless game skeleton instead of the actual game. Fallout 4 is a good example of this policy. Dev and publisher will not bother making a polished and balanced game when there are modders who are ready to do this work for free.
Secondly, modding requires the usage of simple scripting languages. Lua for example. Better modability means more game logic will be incapsulated into scripts instead of native code. While Lua is quick enough (if everything was done right, of course) it is still slower than precompiled code.

By the way, Civ VI was announced as a game with "a new engine that is expected to be more friendly to modification".

 

[oedali]

By the way, Civ VI was announced as a game with "a new engine that is expected to be more friendly to modification".

Exactly. That's why I don't understand why it's taking them so long to release the SDK and let the modding community get their hands onto the AI code. I am dreaming of the days that the Civ 5 Vox Populi mod team takes this on and makes the AI a cunning strategic opponent that is opportunistic and neither too friendly nor too hostile yet maintaining distinct leader traits. Plus they are so responsive on bugs and community feedback and I used to always post things on GitHub an get responses, vs. our bug reports here on this forum which seem to fall on deaf ears (do the devs even read them?)

Someone said earlier that releasing the SDK is also an extra "cost" to the devs, but it shouldn't be if your point is correct, right?

 

[empresskiova]

The AI is getting better (not including the wonky/buggy diplomacy stuff). Not a whole lot, but they're trying. Got a game where France settled 2 cities, and one actually had a Campus in it. Her Capital (pop 7-8) also had a Campus (and a Holy Site and Encampment). So the AI is at least attempting to build districts now. Granted, Cathy's cities became Polish after she captured Kandy, but that's neither here nor there. They still don't do a good job with adjacency bonuses, and they're still settling away from water (when there's no clear reason to), and of course tactics are a crapshoot, but at least the AI is capable of playing the game now. I've got Scythia, India, and Japan on my Continent too. No clue how their cities look, because they don't like me much.

Also, didn't the AI in Civ 5 cheat? I didn't play it, but from what I understand a good portion of the rules didn't apply to them...

 

[Gehennas]

Someone said earlier that releasing the SDK is also an extra "cost" to the devs, but it shouldn't be if your point is correct, right?

SDK is required for devs themselves, so I am pretty sure that it already exists. It may be not production-ready but it won't require much effort to be finished. This gives some hope that devs are going to polish the game first but don't have resources to do this now. While "XCom-2. War of the chosen" release is soon, I am sure that QA is pretty busy, for example. Of course, without any feedback from devs we can only make assumptions.