Civilization VI


Chiming in with my experience…

The AI did improve…barely. Not enough to matter. I’ve been playing on Immortal.

It can finally take a walled city, IF it outnumbers you 6 to 1. And I don’t mean a 6 to 1 ratio, I mean only 6 units against 1. If it’s any more units than that, it will run into a traffic jam and have no clue how to herd all those units to overwhelm you.

The AI’s one and only shot at beating you is still in the very early first 20-40 turn rush, before walls, before you have a chance to match its massive starting army advantage. That’s it. That’s the AI’s only hope. If it can’t beat you in the first 50 turns then it’s never going to beat you.

That means the entire game past turn 50 is pointless. You will win. It’s just a matter of time. It still can’t win via domination. It still can’t win with space race because the requirements are so ridiculous and the costly space port districts so easy to take out. It still has no idea how to take advantage of tourism bonuses to generate enough to win a culture victory. It can’t win a religious victory because you just declare war and wipe out its missionaries the instant they threaten to convert you.

I’m not even an expert at this game. I was never able to win any Civ 4 games beyond Monarch difficulty, yet I’ve never lost a Civ 6 game on Immortal (barring the super early rush). I’ve never been in the slightest danger of losing to any of the victory conditions. The AI just completely lacks the ability for the final push needed for culture or scientific victories, and a human player will never lose to a religious victory.


The Steam rating for the last month is sitting at Mixed with 54%. I think this is a first for the series, no?


Weird that it’s a lowpoint, considering it is better than V.


Honestly they should turn this game into a solo city-(country)-builder, with occasional random barbarian incursions or revolutions to motivate the armed forces. There’s just no point playing into the middle game in the current release.

For masochistic fun I too played a game recently. England (and some other random country on the far side of the world) declared war on me out of the blue on that idiotic “player is too far ahead” basis. They attacked my musketeers and cannons with a force entirely composed of archers. 100% archers, swear to god, a dozen of them at a time. In effect it was just a way for me to conquer them without incurring war costs, which are negligible anyway. Come on, even if I just hit end-of-turn a hundred times they still would have been physically unable to take a single city.


Just chiming in. I also played (or started to play) a game this weekend, and it was just an abysmal experience, from the ai to poorly implemented gameplay decisions.

What a mess.

Started a game of Civ IV, lost 5 hours in a blink.


In discussing GalCiv3 AI here: , Brad talks very briefly about why the OS/2 version of the game (from 1995) had the strongest AI:

[quote=Brad Wardell]
Re why was the OS/2 version better at AI?

Primarily because it was the only GalCiv game that I designed and programmed and thus the AI and the gameplay were in perfect sync.[/quote]

It seems to me that as games have gotten larger, you end up with a few people designing that game and one poor sod who’s unenviable task it is to try to make an AI to play that game. This results in very interesting game designs that would theoretically be fun to play if the AI opponent actually had a clue what it was doing.

Is a good single-player strategy game basically reliant on having one super-dev who can design all the basic gameplay mechanisms and the AI to deal with them? I used to think the AI dev needed to have a veto on all design ideas, but now I’m thinking even that’s not enough.


I would argue that the vastly more complex gameplay systems of modern 4X games puts rather a burden on AI developers as well, yeah.


It’s sort of a double-whammy. The gameplay is more complex, so one person can’t keep as firm a grasp on it, and it’s exponentially harder to write good AI code form complex gameplay.

Still, I think if the AI dev has more control over gameplay design, they can steer the gameplay mechanisms to forms where they know how to write a stronger AI. Like says the combat balance between units (or unit design) is based on a specific logical structure that the developer can encode and optimize for.


Do not forget that OS\2 was superior to Windows in every regard, almost up to todays standards (It has REXX for one thing). Only thing better would of course be AMiGA (Which has AREXX).

I wonder if there is such a thing as an “AI” pack you could buy and add to game engines, sort of like how HAVOK did physics and Speed-tree did shrubbery.


Game design is an incredibly important part in regards to AI performance, but talent and resources are the other. I think it’s the latter in particular that is the problem with Civ games these days. Firaxis simply does not care. The AI in Civ5 was abysmal but the game sold like hotcakes to rave reviews, so why invest money in that area of the game?

I remember Firaxis (I think it was Jon Shafer) saying during Civ5’s AI development that Firaxis’ policy was to have the lead designer also be the AI guy, to make sure that the AI knew how to play the game. So they are certainly aware of how important the design aspect of the game is. The problem is they just don’t see value in a competent AI, so devote no resources towards it. AI requires many, many, iterations to get right and it’s clear that they just can’t be bothered.

From a design aspect, 1UPT on a scale of Civ is going to be a problem. But on the other hand, there’s so much low-hanging fruit hanging around Civ5 and Civ6 that could be solved. Things like in Civ5, how for years the AI didn’t understand a Great General wasn’t a combat unit itself and would charge ahead of the army. Or the ridiculous “diplomacy” in Civ6, where Teddy Roosevelt declares war on me for being at war on his continent, despite the fact that it was the Aztecs that declared war on me. Or building tens of thousands of missionaries to carpet every single hex in the game. Or any number of other issues where the AI is so flawed it should probably be categorized as a bug.

I don’t get upset with AI not being great when we’re talking version 1.0. I understand that the AI can’t really be tuned until the game systems and balance are mostly in place, so it needs to be done at the end of the project. But I expect to see heavy work continue on this aspect of the game, especially when there is a business model that supports ongoing development (expansions, DLC for a new Civ, etc).

Firaxis lost me for quite a while after Civ5. I didn’t pick up either XCOM game, I skipped Beyond Earth. Given my 20 year love of the Civ franchise I thought I’d give them one more shot with Civ6, but I am done with the franchise and the company at this point. All the cool design ideas (and I thought there were several) in Civ6 don’t make a difference when the AI is this inept.

EDIT: I should point out that the modding community in Civ5 was able to make drastic improvements to the AI in that game, once they were given the tools to do so. This clearly demonstrated to me that it wasn’t technical or design limitations that kept the AI back, it was simply lack of effort/resources on Firaxis’ part.


The XCOM games are really good (especially the second one), but I’m probably off Civ games until they show me something. I played Civ5 a ton despite the AI…the Civ6 AI is too much.


100% agree. I love the XCOM games and the AI can play those with brutal precision.


I guess I am reaching for a systematic explanation because terrible AI seems almost universal these days. I suppose “players don’t care about AI, so neither do the devs” is a systematic explanation, but every player I’ve talk to says it’s important to them, and so do the devs. Though I guess actions speak louder than words, in both cases. I don’t spend much time getting kicked around by a chess AI, for instance.


Thirded. Even as someone who has lost all desire to ever touch a Civ game again, Xcom’s were fantastic games.


Jake Solomon for lead design Civ 7? :)

I liked Civ VI a fair bit, thought I’m still more fond of CiV (plus expansions) currently, I feel like a meaty expansion could really push me over the edge. There are a lot of gameplay mechanics in 6 I really, really like and the AI improvements at this point would probably have to come through the development of an expansion anyway.


And if you turn off the religious victory condition, the AI will still try and go for it. It does nothing to stem the flow of religious unit spam and the AI will still build religious districts and buildings like crazy.


I think that’s simply it. The majority of players don’t want a really challenging AI opponent. They may say otherwise, but the sales numbers don’t bear that out.


Of course there’s a difference between a challenging opponent and and an imbecile. Just pretend you’re playing the game, is all I ask. But Civ VI tactical AI can’t even pretend. It just sits there and drools. And the strategic/diplomatic AI isn’t AI at all, it’s random meaningless events meant to give the player something to do.


I think the AI in AOW3 is a pretty good example of what I’m looking for, it does a competent job on the strategic map (on only the 2nd hardest setting I’ve had it repeatedly take cities I either failed to defend or under defended) and it’s even better in the tactical battles.


This thread makes me sad. I’ve been planning to revisit Civ VI as I’ve been playing games from the backlog more. I liked the mechanics of VI a lot - definitely more than V, but if the AI is still as bad as everyone is saying then it probably isn’t worth playing. Maybe worth 1 or 2 reintroduction games.

Civ V was probably the version that brought Civ to more casual gamers, who perhaps didn’t care too much about the AI. Maybe now with more experience under their belt the AI in Civ VI won’t be enough.

Did the Civ V AI really improve that much to make it noticeably better than Civ VI?

I’ve been away from Civ IV for so long that maybe playing it will feel like a new game.