Likelihood we'll get a decent AI

[AriochIV]

This is irrelevant. Aside from that, you and me as consumers and players do not want a good AI. Studies show we automatically think a good AI is unfair and/or cheating.

I think that an AI that plays the game optimally is not very fun to play against except to the most hardcore Deity players... but there is a mile of space between that and an AI which is jarringly inept, to the point where there are entire classes of units that it has no idea how to use. "Good" does not mean "perfect." And if an AI is too hard, you can always turn the difficulty down. If the AI is inept, the only way to turn the difficulty up to to have it cheat, which becomes very noticeable when it has to cheat a lot. What is important is how the game feels.

And it's exceedingly insulting to tell people what they do and do not want. And yes, I know that Firaxis has said more or less what you just said above, and I find that just as insulting. As a developer you do what you think is right, not what the player polls tell you to do... but don't insult your audience.

 

[Gorbles]

And it's exceedingly insulting to tell people what they do and do not want.

Let's be fair - folks do this all the time. Contentious topics tend to split opinion (I don't want to name them, but we all know what they are).

But I think I actually agree with Nikolai here. We don't want to be beaten. We want a challenge, and that challenge is both variable and heavily skewed towards high-skilled players which make up a very small percentage of the overall playerbase (similar to something like including MP).

Is there room for improvement? Yes. Are there at times easy wins? Absolutely. But until people can coherently and with a large enough consensus agree on what "good" actually means, these semantic arguments will keep going around and around.

I'm personally hoping for a more realistic AI. I'm not the best player so I don't care about minmaxing my turn strategies against a Deity-like AI. And the effort required to be invested to reliably challenge a Deity+ player is probably why we don't see it. Heck, the return on investment for AI development in general probably (for better or worse) skews towards implementing things that benefit players like me over players at the top of the game. This is different to more inherently competitive genres like RTS or MOBAs.

An AI that is more realistic and internally-consistent (as difficult as that may or may not be) will do more for the image of the game than being able to satisfy the challenge required for the high-level players, imo.

 

[Eagle Pursuit]

high-skilled players which make up a very small percentage of the overall playerbase

This is an insanely undervalued point. High-skill players make up a large portion of CFC, but a very small -and shrinking- portion of the overall playerbase. It just wouldn't be wise for Firaxis to cater to these players to the exclusion of low-skill players.

 

[GeneralZift]

Aside from that, you and me as consumers and players do not want a good AI. Studies show we automatically think a good AI is unfair and/or cheating.

Is this a joke that I don't get?

 

[Trav'ling Canuck]

This is an insanely undervalued point. High-skill players make up a large portion of CFC, but a very small -and shrinking- portion of the overall playerbase. It just wouldn't be wise for Firaxis to cater to these players to the exclusion of low-skill players.

Which is why so much comes back to the vision of the dev team. What type of game do they want to make? There's little economic incentive to layer a competitive quality AI on top of the game they want to make if their vision of what makes a fun Civ game is "choose how you win". Since "choose how you win" was one of the guiding design principles of Civ 6 and the same senior dev team is (as I understand it) in charge of Civ 7, I would expect more of the same. But we'll see.

 

[leif erikson]

This is an insanely undervalued point. High-skill players make up a large portion of CFC, but a very small -and shrinking- portion of the overall playerbase. It just wouldn't be wise for Firaxis to cater to these players to the exclusion of low-skill players.

Depends upon how they do difficulty levels. If AI competence scales with difficulty level instead of relying on massive bonuses, then it might work out?

 

[AriochIV]

Let's be fair - folks do this all the time. Contentious topics tend to split opinion (I don't want to name them, but we all know what they are).

That people are rude and insulting all the time is not, in my opinion, an excuse to be rude and insulting.

Is there room for improvement? Yes. Are there at times easy wins? Absolutely. But until people can coherently and with a large enough consensus agree on what "good" actually means, these semantic arguments will keep going around and around.

Surely it is not a controversial assumption that "good" means at the very least that the AI should be able to use all of the features of the game.

 

[Lazy sweeper]

Ai cheats is an assumptions for good Ai.
Stockfish Ai has levels that works with the Elo system, and they memorize 10, 100, 1000, 10PetaTrillion moves in advance, and that makes the difference.

I'm no Ai expert, but I can beat a <1000<2000 Elo Chess Ai. Anything above, it's impossible.
Recently Ai is making Ai bots of strong Grand Masters, that analyse their plays, and try replicate their plays.
I wouldn't mind a tiny bit of telemetry, if there was a goal behind it, to make custom Ai bots.
Or make a send save games system that would analyse willingful players that would like to improve Ai by submitting their games.
Or a Twitch bot that would analyse online multiplayers games and evolve that way....


On a side note, it would be interesting to see an 8x8 tile map, fill the two sides like a Chess game with two civs units that could equal Chess moves, and see how the Ai would do.
It would need to enable the old King regicide and Queen 'queenicide?' rules to be effective as win conditions, but as an experiment, If I was delegated to Ai, it is something I would
probably do, given my love for chess.

Ps: Chess uses squares... and cis units all moves as either a king or the Queen... maybe the spearmint could replace the pawns? Eat only diagonally? Catapults move only horizontally and vertically as Towers and Archers only diagonally?
Evan only as a mod, it would be quite staggering to see... add a Star Wars mod to that... Eureka!!!

 

[Gorbles]

That people are rude and insulting all the time is not, in my opinion, an excuse to be rude and insulting.

shrugs

I guess we disagree on "rude and insulting". Would rather not spend more time hashing this out though.

Surely it is not a controversial assumption that "good" means that the AI should be able to use all of the features of the game.

It's not, but you're not going to get consensus on that, at least from the posts I've seen here over the years.

Some would call that serviceable. Or the bare minimum. Others want more from "good".

 

[AriochIV]

shrugs

I guess we disagree on "rude and insulting". Would rather not spend more time hashing this out though.

It's not, but you're not going to get consensus on that, at least from the posts I've seen here over the years.

Some would call that serviceable. Or the bare minimum. Others want more from "good".

Maybe... you should consider saying what you think, rather than claiming to represent what "some" and "others" think.

 

[Gorbles]

Maybe... you should consider saying what you think, rather than claiming to represent what "some" and "others" think.

I already did, but you opted to not engage with that.

In any event, I don't claim to represent anybody elses' views. I'm just saying they exist. Regardless, this doesn't seem to be continuing in a constructive way, so I'll bow out of this tangent here. I've said everything I wanted to say.

 

[Dale]

I'd like to see a Civ AI that meets, or hopefully exceeds, Old World's AI quality benchmark. Anything less than the quality of Old World's AI would be disappointing for me.

 

[Krajzen]

I want AI which is moderately competent in all game mechanics (not literally unable to use some of them) (not breaking immersion and causing boredom by its stupidity), while using this competence not to annoyingly min-max meta play to destroy human player but roleplay its leader/civ as one another civilization trying to stand the test of time and be the greatest in at least some aspects (economy, culture, war etc) if not all of them.

 

[TreVor_ish]

Looking at global achievements stats on steam for CIV VI I can safely say that, majority prefer to play an easy game. Less than 40% even won the game at any difficulty, (this is probably because of the late game drag) ergo, I'm not holding my breath on this, but time will tell.

Old World AI is much more capable in playing the game and has more options to tinker with bonuses for AI, so yeah, anything less than that would be disappointing.

 

[Trav'ling Canuck]

I'd like to see a Civ AI that meets, or hopefully exceeds, Old World's AI quality benchmark. Anything less than the quality of Old World's AI would be disappointing for me.

Old World was developed with the intent of being a game the AI could play. Per the discussions above, based on past experiences with the current Civ dev team, it is unlikely that this will be an important consideration for them. If they come up with a mechanic that they think players will find fun, they're unlikely to exclude it from Civ 7 just because they can't teach the AI to use it properly.

 

[Leyrann]

Recently Ai is making Ai bots of strong Grand Masters, that analyse their plays, and try replicate their plays.
I wouldn't mind a tiny bit of telemetry, if there was a goal behind it, to make custom Ai bots.
Or make a send save games system that would analyse willingful players that would like to improve Ai by submitting their games.

AI for a turn-based strategy game like Civilization is orders of magnitude (and that is not hyperbole) more difficult than AI for chess.

I calculated some stuff for Age of Wonders when the same topic came up, in Age of Wonders IV, a single 6-stack of units standing on average terrain without roads can, in one turn, move in a number of ways equal to the total branches in five full (that means: white+black) chess moves, and with significantly worse ability to prune options (e.g. a chess engine can prune any line that starts with 'hang your queen for no reason' as terrible without having to consider the 1 billion iterations that develop in the 4.5 moves after). Obviously, the math for Civ is a bit different, but we're just talking about movement here. How are you going to decide which one of 6 units, 8 buildings, 6 wonders and several projects and districts you're going to build in City #1? And what about City #2? City #3? And building something in one city affects building it in another - if I'm already building an Archer in City #1 I might be better off building a Library in City #3. If you've got ten cities, that's 20^10 = an enormous number of permutations. And then there's diplomacy, long-term strategy (prepare for war? make friends? focus on science? culture? religion?), what to research (x2), what government to run, and so on and on.

Long story short, an AI that exhausts all options like a chess bot, no matter how smart it is at weeding out the stupid ones, isn't going to work for Civ. And neither is attempting to replicate playstyles. There are simply too many variables. I think replicating players in particular is something an AI can only do in any way competently once the base AI is stronger than those players.

 

[AntSou]

This is an insanely undervalued point. High-skill players make up a large portion of CFC, but a very small -and shrinking- portion of the overall playerbase.

Wait, where is this coming from?

 

[The_J]

Well, you speak of two things:
- Having AI that can (somewhat) converse with us, more than the usual pre-recorded message we had so far would be nice, if the AI stay in the correct caracter (a Pierre II that with an aristocratic tone, etc...). It shouldn't be too difficult to do, but we would have automatic voices, thing that some find rebarbative...

Oh my, I read an article about a RPG where they did that, and it worked quite well in the development stage.
Now, I'll be damned if I could remember the name -_- .

 

[Leyrann]

Wait, where is this coming from?

Take your pick:
-Common sense logic.
-Steam statistics.
-Statements by Firaxis (if they exist).
-Comparing to other games, using statements of their devs.
-Comparing Civfanatics averages to whole community averages.
-etc

 

[vorlon_mi]

It likely depends on what you mean by "decent AI".

I'll share the 3 things I always indicated I wanted when we had this conversation about Civ 6, and a 4th I've added after the experience of playing Civ 6:

1. I want the AI to act as a pace car. The speed of the pace car should depend on the difficulty setting, but I should have a reasonable idea, at the outset of the game, that if I don't win within X turns, one of the AI is likely to have won before me.

2. I want the AI to act as a speed bump. I don't want the AI to passively ignore what I'm doing, I want it to take steps to mess with my plans and make me adapt.

3. I want the AI to enhance flavour and re-playability. Having Shaka as your neighbour should be a different experience than having Gandhi as your neighbour.

4. I want the AI to avoid player-obvious, immersion-breaking incompetence. I don't care if the AI can't play all the game mechanics "well", as long as it doesn't blatantly display the things it does poorly continually and regularly to the player.

I'd like to think these are reasonable goals for Civ 7, but I'm not sure any of these will be priorities for the current design team, so who knows.

I agree SO much with these goals. They align with my observations about CPU players in other franchise games which seem to follow the following priorities:

1. Survive: deal with imminent threats or war declarations
2. Thrive: develop cities and population, grow their economies
3. Win: pick a victory condition and pursue it

Granted, I've rarely seen a computer player pick "Conquest" or "Domination" as their VC, but I have often seen them pursue each of the other VC.
Even in Civ6, if one allows the game to go past turn 300, you will observe the computer players building towards a victory: spamming missionaries & apostles, trying to build the Potala Palace or Statue of Liberty, accumulating lots of great works.

To expand on Trav'ling Canuck's ideas:
Pace Car: Yes, if I try some different approach and execute it poorly, I want the AI pace car to pass me by. To build more cities, to build more tourism, to spread its religion to my neighbor. If I play poorly and fall behind the pace, I should lose.

Speed Bump: If I send out a settler with no escort or neglect to protect my city near a resource, I would expect the computer player to take advantage. If I forward-settle a computer player, I might get attacked before I can get my defenses up.

Enhance Flavour: Yes, having leaders with different attributes and tendencies as neighbors should affect how I plan my game.