Is AI still brain dead?

[ThunderLizard2]

It's been six months since I checked in with the board on the status of the Civ6 AI. Consensus then was that it still needed a lot of work, especially combat management and AI civ management. Has it improved much? I was eyeing the 50% off sale but would rather wait until there's a working AI before plunking down any money.

Have any major AI improvements been mentioned for R&F?

 

[darkace77450]

It's much better than it was but still nowhere near as good as it should be. At 50% off I'd recommend giving it a shot to see for yourself if it's acceptable in its current iteration, budget permitting.

 

[CPWimmer]

No. Don't bother yet. I love Civ 6 but if your enjoyment is going to be based on the AI possibly winning the game against a serious player don't jump in yet. Check back after this expansion pack coming out next month.

 

[ThunderLizard2]

budget permitting

Budget's not issue - it's more whether the AI can put up a decent fight at moment or whether it's still brain dead.

 

[liv]

It is on life support so you should wait

 

[BrendanVI]

good thread, haven't played civ6 for a while because of the "cousin eddy" AI... got back into civ 5 deity and keep getting spanked hard. however tonight may be the night that I spank the world with my four highly experienced frigates

 

[Cyberian]

AI that isn't trash has never existed in complex games.

Its basically fake news that it exists or is possible in gaming today.

I played almost all major strategy titles and I can't remember a single title that had decent polished AI. Maybe Starcraft 2 but that is a very simplistic game without enough complexity. If you look at Paradox games or more niche titles as long as a game is sufficiently complex the AI sucks. It wasn't as bad 10++ years ago because in general strategy games were more simplistic and had less variables/choices.
And with AA(A) games getting ever expensive I doubt we will find studios which invest a giant amount of developer time on this as its mainly a black whole so not a good business decision.
Its time for a major revolution but I don't believe the AI researchers that we will get that very soon they are like Nuclear Fusion or Neuroscience guys they are hyping up what they actually can do and how long they will take to make considerable advantages to suck up more funding.

That is of course no excuse for studios to not fix what is easily fixable. Peoples just shouldn't expect any drastic changes. Not in any expansion or Civ7/8. You might even see the AI getting worse when the game gets more complex.

 

[CPWimmer]

Have any major AI improvements been mentioned for R&F?

To answer this question specifically, no. Nothing of detail has been mentioned. We are hoping for improvements, but the only quotes I have seen that are AI related have been this interview with the new lead designer Anton Strenger:

The Good:
"We have put out a lot of patches over the last year to address AI behavior, and we are continuing to play and monitor how they make decisions and where they are weighing different options inappropriately."
"For example, it understands the new Loyalty mechanics and places its cities intelligently."

The Bad:
"I feel good about the trajectory, and where the AI is right now" - hopefully he means in the Dev version of R&F he has access to, and not that he is satisfied with where the AI is in the production version that we all have access to.
"our AI engineer" - as in singular - one guy who not only has to make sure the AI is up to speed on the new systems, but fix the existing bugs.

And this is why I think you should wait until after the expansion (Due out Feb 8) and then probably for a month after that for people to get a feel for where things stand. He doesn't mention the known AI issues (airforce usage, loosing track of units, certain AI/Diplo

 

[RohirrimElf]

AI that isn't trash has never existed in complex games.

Its basically fake news that it exists or is possible in gaming today.

I played almost all major strategy titles and I can't remember a single title that had decent polished AI. Maybe Starcraft 2 but that is a very simplistic game without enough complexity. If you look at Paradox games or more niche titles as long as a game is sufficiently complex the AI sucks. It wasn't as bad 10++ years ago because in general strategy games were more simplistic and had less variables/choices.
And with AA(A) games getting ever expensive I doubt we will find studios which invest a giant amount of developer time on this as its mainly a black whole so not a good business decision.
Its time for a major revolution but I don't believe the AI researchers that we will get that very soon they are like Nuclear Fusion or Neuroscience guys they are hyping up what they actually can do and how long they will take to make considerable advantages to suck up more funding.

That is of course no excuse for studios to not fix what is easily fixable. Peoples just shouldn't expect any drastic changes. Not in any expansion or Civ7/8. You might even see the AI getting worse when the game gets more complex.

Europa universalis 4 on hard difficulty does a pretty good job. (for medium sized countries)
So is Xcom 2 on third difficulty. (you can use mods to add to the difficulty)
If that is not enough try beating Deep Fritz on max ELO. Its to much for Magnus Carlson. Or wait for true selflearning AI in 3 decades or so for PC games.

 

[Ondolindë]

In the end, why do you guys think are the reasons why Firaxis and probably other gaming companies don't focus on creating better AI? Is it only because of the complexity of games now? I can't believe that is the only reason.

But, @ThunderLizard2, to answer your question, yes, you probably should wait if you really want to have a challenging experience with the AI. The expansion looks really fun, but I am not sure about how the AI will be able to manage more mechanics well.

 

[DWilson]

Each update has notably improved the AI performance, but it's still pretty terrible. (When I can win on deity, despite being fairly casual for a fanatic, that's a bad sign). If you're into mods, there is at least one AI mod which does make some improvement as well. That said, it'll be limited until the full file sets are released.

 

[RohirrimElf]

In the end, why do you guys think are the reasons why Firaxis and probably other gaming companies don't focus on creating better AI? Is it only because of the complexity of games now? I can't believe that is the only reason.

I think most of the fanbase of civilization are casual gamers. Who play an hour or two after work/school or in the weekends. Not min/maxing veterans. So adding more content is more rewarding then improving the AI.
You see this across the board with 4x games. Some games are better then others but none of them can match a dedicated human player. No matter how much manpower or money you trow at it.
I`m often surprised how easely modders can improve the game with changing some sliders around. I do have high hopes for civ 6. Still think it will do a much better job then civ 5 at the third expansion. The community should make it even better.

 

[RohirrimElf]

Each update has notably improved the AI performance, but it's still pretty terrible. (When I can win on deity, despite being fairly casual for a fanatic, that's a bad sign). If you're into mods, there is at least one AI mod which does make some improvement as well. That said, it'll be limited until the full file sets are released.

I guess you are hinting at AI+ mod. What this mod does in the beginning of the game is spawning captured settlers back at the nearest city. This alone should improve the AI intelligence by a good bit.

 

[Karpius]

Let us not forget who we are here. This is the "civ FANTAICS" forum. We are the people who spend hundreds of hours playing the game, looking under the hood, and crunching the numbers. Many here have have figured out how to beat the game, the next DLC, the next expansion before its even released. We already know pretty much every move the AI can possibly make then complain the game is too easy In a game such as Civ6, only another human could provide any real challenge.

I would guess the majority of the market is made up of the casual player and not the fanatic. The casual player is not looking for the same challenge. The casual player wants to get in, get out, have a little fun and move on to a bevy of other games. The investment necessary to create an AI capable of challenging the fanatics must be considerable as well as counter-productive in light of the market share I described above. I am simply not sure that an affordable AI is currently possible for our needs; not to mention one capable of running on the typical desktop.

The real failure of the current AI is its inability to learn and predict. In a game as complex as Civ6 and with such a broad scope, no AI can hope to match wits with a dedicated human player. The AI is simply a series of algorithms with mathematical formulas looking at hard values weighted against other hard values. Is it any surprise that the dedicated human mind can beat that? We are far more complex, nuanced and unpredictable. We can and do look under the AI's hood, but the AI cannot peer into the mind of its human opponent.

That said, the sheer complexity and scope of the game is, to me at least, the real enjoyment. Every game is different. Even if I hand pick all the same conditions from one game to the next, they are different. In my opinion, that is the real testament to Civ6 and the developers.

 

[bbbt]

In the end, why do you guys think are the reasons why Firaxis and probably other gaming companies don't focus on creating better AI? Is it only because of the complexity of games now? I can't believe that is the only reason.

But, @ThunderLizard2, to answer your question, yes, you probably should wait if you really want to have a challenging experience with the AI. The expansion looks really fun, but I am not sure about how the AI will be able to manage more mechanics well.

Essentially the significant majority of players play on prince or lower and don't want a serious challenge. They want the AI to at least appear to be challenging though, which it's not really doing unit-wise at the moment. But in general AI doesn't sell games.

And pretty much the majority of games you can find people complaining about the AI (that it sucks and/or that it must be cheating any time it beats them). It probably becomes white noise after a while.

To some degree because of some of the challenges in the Civ AI - complexity, performance if they try and calculate to better tactical/movement combos, making the AI flexible for modding - they might feel there's a bit of a cap in that sense and a case of diminishing returns anyways.

I don't think the new mechanics will be that challenging for the AI in a general sense (less so then the continued challenge of tactical unit movement!). Of course it certainly won't use them optimally, and the top players will, which might just increase the gap between the two.

 

[Stringer1313]

It depends on what you are looking for. If you are looking for an AI to pose the same challenge as a human player (something I personally believe is impossible to achieve in complex strategy games right now) then don't buy. If you are looking for an AI to pose a challenge on Emperor and up AND you are a casual player, then buy.

 

[ThunderLizard2]

Thanks for all the feedback- I'll check back again in a six-months.

In a game as complex as Civ6 and with such a broad scope, no AI can hope to match wits with a dedicated human player.

Maybe that's the problem - why make a game so complicated that the AI can't handle it? Although many despise the CivIV SOD at least I felt on-edge while playing on King or higher that the AI might strike at any moment.

"our AI engineer" - as in singular - one guy who not only has to make sure the AI is up to speed on the new systems

This is very sad that they only have one engineer on this.

Europa universalis 4 on hard difficulty does a pretty good job.

Agree - you can have a fun game and you don't feel the AI is doing totally stupid stuff. I think PDS takes AI development more seriously than 2K

I would guess the majority of the market is made up of the casual player and not the fanatic.

If Joseph de Maistre were still around he'd probably tell us that in a free market people get the AI that they deserve

 

[nzcamel]

Not in any expansion or Civ7/8. You might even see the AI getting worse when the game gets more complex.

Alas...

 

[CivLuvah]

Let us not forget who we are here. This is the "civ FANTAICS" forum. We are the people who spend hundreds of hours playing the game, looking under the hood, and crunching the numbers. Many here have have figured out how to beat the game, the next DLC, the next expansion before its even released. We already know pretty much every move the AI can possibly make then complain the game is too easy In a game such as Civ6, only another human could provide any real challenge.

I would guess the majority of the market is made up of the casual player and not the fanatic. The casual player is not looking for the same challenge. The casual player wants to get in, get out, have a little fun and move on to a bevy of other games. The investment necessary to create an AI capable of challenging the fanatics must be considerable as well as counter-productive in light of the market share I described above. I am simply not sure that an affordable AI is currently possible for our needs; not to mention one capable of running on the typical desktop.

The real failure of the current AI is its inability to learn and predict. In a game as complex as Civ6 and with such a broad scope, no AI can hope to match wits with a dedicated human player. The AI is simply a series of algorithms with mathematical formulas looking at hard values weighted against other hard values. Is it any surprise that the dedicated human mind can beat that? We are far more complex, nuanced and unpredictable. We can and do look under the AI's hood, but the AI cannot peer into the mind of its human opponent.

That said, the sheer complexity and scope of the game is, to me at least, the real enjoyment. Every game is different. Even if I hand pick all the same conditions from one game to the next, they are different. In my opinion, that is the real testament to Civ6 and the developers.

Honestly, as a Civ6 apologist (if that's a thing), this is the best critique and defense of Civ6 in relation to its AI.

 

[Victoria]

Honestly, as a Civ6 apologist (if that's a thing), this is the best critique and defense of Civ6 in relation to its AI.

Agreed although I in no way apologise, as far as I am concerned it has never been a chess game.