[R&F] When will we get a AI overhaul mod?

[Aristos]

Through the Ages, and the AI doesn't cheat.

Also, Civ 4 & 5 at Deity, although with those the AI gets big bonuses to compete. Still, even unmodded those two have much more efficient AI at winning than Civ 6 does (so far).

Indeed, and that's without counting K-Mod for civ 4 and the CP for civ 5, both of which take their respective AI's to a whole new, better level that obviously the guys at FXS cannot even imagine... and for free, and in the modders' spare time.

 

[Trav'ling Canuck]

Indeed, and that's without counting K-Mod for civ 4 and the CP for civ 5, both of which take their respective AI's to a whole new, better level that obviously the guys at FXS cannot even imagine... and for free, and in the modders' spare time.

Yes, the modded AI is at a whole different level. To be fair, though, the modded AI takes advantage of a longer period of player knowledge.

One of the reasons the AI is so good for Through the Ages is that the board game had many years of player experience to draw on.

A second reason is that the developer drew on the best players and involved them in crafting the AI's logic.

A third reason is that the developer cares deeply about how challenging the AI can be for players.

Firaxis can't do much about the first reason under their current economic model (development pretty much ceases at the time when player knowledge of the final rule set is just ramping up). It also can't do much about the third reason if their economic model suggests that investing in an AI is not worth the additional revenue sales. It could, however, do something about the second reason for relatively little cost, i.e. bring some of those volunteers in house and tap some of the better Civ players' brains for the relatively small costs associated with short term contracts as the final release version is nearing completion.

 

[Ryika]

Civ AI is also not built purely for being competitive, but also for flavor. Some AIs will build way too many wonders, some other AIs will maintain way too many defensive troops, etc. That's not a problem in itself - I'd say it's a good thing - but the huge gaps in efficiency of different playstyles very much handicap certain AIs.

Still, I think when people say the AI in Civ is bad, then they're not so much talking about their ability to play the best "build orders", but rather fundamental stuff, like very obvious bad decisions in combat, the lack of goal-oriented gameplay, etc.

 

[Aristos]

Yes, the modded AI is at a whole different level. To be fair, though, the modded AI takes advantage of a longer period of player knowledge.

One of the reasons the AI is so good for Through the Ages is that the board game had many years of player experience to draw on.

A second reason is that the developer drew on the best players and involved them in crafting the AI's logic.

A third reason is that the developer cares deeply about how challenging the AI can be for players.

Firaxis can't do much about the first reason under their current economic model (development pretty much ceases at the time when player knowledge of the final rule set is just ramping up). It also can't do much about the third reason if their economic model suggests that investing in an AI is not worth the additional revenue sales. It could, however, do something about the second reason for relatively little cost, i.e. bring some of those volunteers in house and tap some of the better Civ players' brains for the relatively small costs associated with short term contracts as the final release version is nearing completion.

The third reason is deeply connected to the absolute lack of competition. In that sense, two of the best known (successful) franchises of the last 20 years, Total War and Civilization, suffer from the same 'monopoly" illness. We sorely need competition in both franchises. As an opposing example, look at what is happening in the space 4X market... it's very dynamic, with multiple different franchises new and old offering very good games, competitive AIs (especially compared to civ), good and above UIs, etcetcetc... all of it stemming from competition.

I am wishing for a civ competitor for the last 10 years at least, but it is not happening... yet. My guess: it will come from Paradox, not sure when, not sure how... but it will come. And it will be a fun ride when it happens, especially the scrambling we will see at FXS...

 

[Trav'ling Canuck]

Civ AI is also not built purely for being competitive, but also for flavor. Some AIs will build way too many wonders, some other AIs will maintain way too many defensive troops, etc. That's not a problem in itself - I'd say it's a good thing - but the huge gaps in efficiency of different playstyles very much handicap certain AIs.

Except that this isn't true. There's no meaningful difference to the way AI civs develop in Civ 6.

Still, I think when people say the AI in Civ is bad, then they're not so much talking about their ability to play the best "build orders", but rather fundamental stuff, like very obvious bad decisions in combat, the lack of goal-oriented gameplay, etc.

I can't speak for others, but when I say the AI in Civ 6 is lacking in it's current state, what I mean is:

(1) It doesn't provide variety in the interactions with different leaders. I don't need to approach the game differently when I start beside Cleopatra or start beside Genghis. One is not more likely than the other to attack me, more likely to backstab me, more likely to try to win a cultural victory, etc. Or if there is a difference, then it's so subtle as to escape me. This was not the case with past iterations of Civ, where AI leaders demonstrated distinct personalities that provided interest and re-playability to the game. ("Personality objective")

(2) It can't provide a meaningful challenge because even on Deity it can't complete a victory condition in a time frame that's even close to what a human can achieve. Not just a human playing ultra efficient, either. Ultra efficient and you can beat it with a 150 to 200 turns to spare. Mess around and pay little attention and victory still comes 50 to 100 turns earlier than the AI. (times based on standard speed, continents, deity) ("Pace car objective")

(3) It doesn't require you to pay any attention to the AI leaders because they won't interfere in your plans. Whether they like you or hate you, they're unlikely to attack you, and even when they do, they're inefficient at impacting you. Therefore, you don't need to devote any attention to diplomacy or devote very many resources to defence. ("Speed bump objective")​

 

[Spudsie74]

Through the Ages, and the AI doesn't cheat

Point well made, and acknowledged. With Civ 4 & 5....I think it is a bit more iffy. IMO Through The Ages is a great game, but adapted from a boardgame, and considerably less complex than Civilization. It has beeen a Looooong time since I played Civ 4 (because the original game won't run on my computer, due to the copy protection). As I recall, the game allowed stacking, and the AI (as well as avid players) would spam a mass of military units, and steamroll any civ that didn't have the production perks given to the AI at higher difficulties. Brute force, or, as I call it when I do it, "breeding like roaches." This is certainly a viable tactic, and I used it successfully to run through Civ 1-4. Civ 5 was something else entirely...no stacking, and significant additions to governments, religion, and other features made it a bit tougher nut to crack. No more using 2,000 spearman to overwhelm 1 panzer.

I still think that any AI opponent is bound to be stupid, relying on brute force, or perks not available to the gamer. Modders have done excellent work on the AI in 5 & 6 (as well as other games), but it seems, IMO, that making the AI stronger (or more numerous...as opposed to making it smarter) is the path Game Publishers usually take. Probably due to cost-benefit analysis.

 

[Trav'ling Canuck]

IMO Through The Ages is a great game, but adapted from a boardgame, and considerably less complex than Civilization.

Less complex rules, yes, as Civ seems to relish in complexity for complexity's sake. I'd argue that TtA's strategic depth is greater than Civ's, though, that there are far more meaningful decisions to be made (though far fewer actual decisions/mouse clicks), and that it's far more challenging in TtA to understand the best decision in any given circumstance than it is with Civ.

I'm no expert in AI, but I'd guess that it's far more challenging to teach the AI to play TtA well than it would be teach the AI to play Civ well. I'm also quite certain that TtA threw a lot more people time at the AI, and continue to do so as they polish the upcoming expansion (multiple years in the making even though it only deals with new Leaders and Wonders - TtA seems to be obsessed about game balance as well as making sure the AI can play the new options well right out of the gate).

Mind, the AI in TtA likely can't hold a candle to an expert player. But it's sure tough enough for even an experienced player. I'm only one person, so clearly this is anecdotal, but after the equivalent number of hours when I could beat the Civ 6 AI on Deity with 90%+ frequency, I'm only about equal to the non-cheating AI in TtA (I can win a 4 player game about 25% of the time).

 

[Ryika]

Except that this isn't true. There's no meaningful difference to the way AI civs develop in Civ 6.

That may be true. I was mostly going by my knowledge of Civ 5 there, because I have not really put much time into Civ VI as it didn't appeal to me.

 

[Trav'ling Canuck]

That may be true. I was mostly going by my knowledge of Civ 5 there, because I have not really put much time into Civ VI as it didn't appeal to me.

Your statement would have been 100% accurate with respect to Civ 5.

 

[Uncle_Joe]

(3) It doesn't require you to pay any attention to the AI leaders because they won't interfere in your plans. Whether they like you or hate you, they're unlikely to attack you, and even when they do, they're inefficient at impacting you. Therefore, you don't need to devote any attention to diplomacy or devote very many resources to defence. ("Speed bump objective")​

This is the key problem IMO. The only 'competition' in the game comes from massively inflated bonuses the AI gets. If you overcome those, you win. If you don't, you lose. But it doesn't feel like you are under 'threat' from other Civilizations.

 

[MisterBoomBoom]

Except that this isn't true. There's no meaningful difference to the way AI civs develop in Civ 6.

I can't speak for others, but when I say the AI in Civ 6 is lacking in it's current state, what I mean is:

(1) It doesn't provide variety in the interactions with different leaders. I don't need to approach the game differently when I start beside Cleopatra or start beside Genghis. One is not more likely than the other to attack me, more likely to backstab me, more likely to try to win a cultural victory, etc. Or if there is a difference, then it's so subtle as to escape me. This was not the case with past iterations of Civ, where AI leaders demonstrated distinct personalities that provided interest and re-playability to the game. ("Personality objective")

(2) It can't provide a meaningful challenge because even on Deity it can't complete a victory condition in a time frame that's even close to what a human can achieve. Not just a human playing ultra efficient, either. Ultra efficient and you can beat it with a 150 to 200 turns to spare. Mess around and pay little attention and victory still comes 50 to 100 turns earlier than the AI. (times based on standard speed, continents, deity) ("Pace car objective")

(3) It doesn't require you to pay any attention to the AI leaders because they won't interfere in your plans. Whether they like you or hate you, they're unlikely to attack you, and even when they do, they're inefficient at impacting you. Therefore, you don't need to devote any attention to diplomacy or devote very many resources to defence. ("Speed bump objective")​

One of the easiest games of Civ VI can be had by simply not engaging the other civs. Don't accept their emissaries, tokens, or requests at all the entire game. Very telling of how good the AI is in this game.

 

[Leathaface]

When will we get a AI overhaul mod?

If the DLL source isn't released, then never.

Plus I don't expect Firaxis to themselves improve Civ VI's AI to compete with the level of Civ V's Vox Populi.

 

[chazzycat]

I hadn't been playing for most of 2018, but recently got back into Civ 6. The AI is actually much better than I remember. They can pull off a pretty deadly knight rush for example, and the archers move & attack same turn...did civ 5 ever have that? IIRC, it did not.

I get killed fairly regularly on emperor level if I am not careful or don't abuse Magnus chops...

I guess from reading this thread it's the minority opinion, but honestly I think the AI is much better than it gets credit for.

If anyone here hasn't played for a long time, I'd suggest giving it another go before passing judgment.

 

[Leathaface]

I think the people who wish for better AI are skilled players who can beat Deity in Civ VI fairly easy, that they aren't really pushed. I've yet to move up to Deity in Civ VI, but from watching twitch streams playing on Deity it is not as difficult as in Civ IV or Civ V.

From my personal experience, i've close to 2,000 hours in Civ IV and didn't beat a single game on Monarch difficulty. Monarch difficulty is below Emperor, Immortal and Deity. Whereas in Civ VI i've less than 900 hours and am beating Immortal regularly, and TBH most of the time without much hassle. Just survive the early rush and build Pikemen to counter Knights in the Medieval era.

Granted my 2,000 hours in Civ VI was me learning how to play Civ for the first time. Plus a modder named Blake made an AI mod so good in vanilla Civ IV that Firaxis incorporated it in Civ IV's expansions, but still. I do doubt if I played vanilla Civ IV, before the AI mod, i'd be winning Immortal difficulty after 900 hours fairly easily when I couldn't get past Monarch in the finished Civ IV after 2000 hours.

 

[chazzycat]

I regularly beat immortal in civ 5 no problem. I could even get away with crazy stuff like going culture victory as pacifist Genghis Khan and win...

Maybe I just need to work on my early build orders, or accrue more hours like I had put into Civ 5. Not saying I'm a great player by any stretch. But not a scrub either

 

[Trav'ling Canuck]

I hadn't been playing for most of 2018, but recently got back into Civ 6. The AI is actually much better than I remember. They can pull off a pretty deadly knight rush for example, and the archers move & attack same turn...did civ 5 ever have that? IIRC, it did not.

I get killed fairly regularly on emperor level if I am not careful or don't abuse Magnus chops...

I guess from reading this thread it's the minority opinion, but honestly I think the AI is much better than it gets credit for.

Tactically, there are improvements, no doubt. Yes, AI ranged units will move & shoot, and no, the Firaxis-issued version of Civ 5 never accomplished that.

Also, the AI can be ruthless in assaulting undefended cities now, especially early with Warriors and Knights. Having Walls is no longer a guarantee that the AI can't take your City.

Despite that, at this point in Civ 6's development I'd suggest the AI does not offer as big a threat on attack as it does in Civ 5, and cannot defend as well as it could in Civ 5. That's based on a lot of time spent playing Civ 5 (Firaxis-issued, not modded) and a somewhat lesser amount spent on Civ 6, including up to the latest patch.

Civ 6's AI's issues are more strategic in nature. It builds the wrong things at the wrong times, and won't stop what it's building when under a serious attack. It probably builds a large enough military most of the time, but it commits those units to battle in such a half hearted matter that a force half its size or less can typically fight it off. This is facilitated by the AI's tendency to change objectives every couple of turns, resulting in it moving units back and forth between possible targets without ever pressing an attack on any of them. This is also part of the reason ware is more dangerous early on, when you only have a few cities, and the AI seems to find it easier to stick to attacking one city rather than wandering about.

Then there's how the AI handles naval units and air units, which are best left undiscussed. Rendered somewhat moot anyway by the AI's general unwillingness to attack you later in the game anyway, regardless of whether you've engaged it diplomatically or not. This was never the case in Civ 5. Perhaps the diplomacy game in Civ 5 was too easy, but at least you needed to engage in it if you were playing a typical 4 city empire, or else your larger neighbours could materially slow down your victory with attacks. The AI in Civ 6 won't do this unless you've been warmongering (and if you've been warmongering, you're likely tougher than them militarily anyway.)

All of this is fixable. Hopefully it will be.

 

[chazzycat]

This is facilitated by the AI's tendency to change objectives every couple of turns, resulting in it moving units back and forth between possible targets without ever pressing an attack on any of them.

I feel like the patches have improved this specific issue noticeably. I remember that happening a lot at launch for sure. But more recently they seem much better at not doing this. Or maybe I just got lucky (unlucky).

Otherwise generally I agree with your points when it comes to overall strategy. For example the non-religious civs spend way too much effort building holy sites. Building Petra with just 1 desert tile, stuff like that.

 

[Leathaface]

@Trav'ling Canuck Have you ever played Civ V with the Vox Populi mod? If so, what are your impressions? I hear it improve V's AI a lot.

 

[Trav'ling Canuck]

I feel like the patches have improved this specific issue noticeably. I remember that happening a lot at launch for sure. But more recently they seem much better at not doing this. Or maybe I just got lucky (unlucky).

All of my observations were post the second last patch (the one before the minor July 2018 tweak, which didn't change the AI).

@Trav'ling Canuck Have you ever played Civ V with the Vox Populi mod? If so, what are your impressions? I hear it improve V's AI a lot.

Only played Vox Populi once, a couple of years ago. I've heard it's come a long way since then, but it seems to be in (very) active development). I'm going to wait until it's finished before I try it again, as the community involved with it seems to very actively debating certain balance points.

In the meantime, I still find unmodded Civ 5 to be challenging and fun, so the extent I'm playing civ (not much now), it's Civ 5.

 

[Spudsie74]

Vox populi, Vox Dei! The did a great job on it...I have ~ 3000 on 5, at least 1,000 since Vox. Different sort of bird from six, but both are very tasty, although I still like to grumble.