RHQ Artificially Intelligent AI Mod RHQ AI Mod 2.07

[notque]

I started a 2.07 deity game (my first try on deity)
I use this mod now, set up a game of me as Himiko/Khmer, and 3 sci non-military AIs...

A-Age, turn 50, it is all peace - I try to make friends with Ada, who is at 1 city (?), while the other AI is at 4/5 settlements...
I have 3 settl. - and do all the endeavors I can...

I haven’t seen anyone at 1 city at turn 50. That’s interesting… not sure what is happening there

 

[V. Soma]

I haven’t seen anyone at 1 city at turn 50. That’s interesting… not sure what is happening there

I have yet to discover Ada's territory - perhaps IPs are there?

 

[ReneTS]

At turn 38 I have 2 other civs with 1 city and 2 with 2. Wouldnt it help if we could give the AI extra settler units at game start? I tried adding it in, but they didnt show up in my game. Not much experience with modding, so maybe someone else will have more luck with it.

 

[Morgawr]

In my experience, with this mod some AI can indeed be a bit slower to get expand in the first half of Antiquity (~70 turns) but they eventually do get there. I believe it's because they focus more on wonders vs vanilla. Whether that makes the game easier or harder is up for debate.

I have also seen rare instances where an AI will be stuck at 2 settlements for a while, but when I place scouts and watch them closely , I find it's because they are having trouble with an IP. That's no different from vanilla. They eventually reroute the settler or wipe out the IP. (If they're trying to befriend the IP however things get more complicated..)

P.s. my advice is to play the full age and then decide of the mod makes it harder or easier.. first 50-80 turns can vary widely

Also not saying something else might not be going on - just my 2 cents

 

[notque]

In my experience, with this mod some AI can indeed be a bit slower to get expand in the first half of Antiquity (~70 turns) but they eventually do get there. I believe it's because they focus more on wonders vs vanilla. Whether that makes the game easier or harder is up for debate.

I have also seen rare instances where an AI will be stuck at 2 settlements for a while, but when I place scouts and watch them closely , I find it's because they are having trouble with an IP. That's no different from vanilla. They eventually reroute the settler or wipe out the IP. (If they're trying to befriend the IP however things get more complicated..)

P.s. my advice is to play the full age and then decide of the mod makes it harder or easier.. first 50-80 turns can vary widely

Also not saying something else might not be going on - just my 2 cents

It's difficult to make decisions about. I can certainly try to push settlers harder, but there's always some tension between one thing vs another. I haven't seen it feel all that bad in autoplays, but I watch their production queues, so maybe I'm justifying that getting that particular wonder at a point is better than a settler, or whatever.

 

[mielony]

2.07

 

[V. Soma]

I finished Antiquity age of the deity game I have with 2.07.
Set: Tiny map, Science victory only (set with mod).
I had AI Ada and AI Kongzi on my continent.
At about 70% of the age Kongzi had 800/t sci, I had about 200/t.
So I attacked Kongzi. I had Ada as ally - and we worked together and just before the end of the age
we eliminated the Kongzi AI. Ada AI took one of his cities - and that was even not easy terrain!
I finished the age with way over the settlement limit: 10/6 - Ada has 5 settlements.

 

[V. Soma]

I started Exp. Age - met AI Franklin, he is on the other half of the world: 8/9 settlements.
He has 500/t sci, I have 400/t - now...
I have 5 treasure resources settled, and now I am at 13/11 settlements.
Ada AI is also not bad: about 350/t sci and more culture than I have...

 

[Renouf]

I think AI wonder chasing is currently a little too high. Rolled a Deity game as Hatshepsut-Egypt and only managed to secure one Wonder, Gate of All Nations (the AI hates the Discipline mastery for some reason). I was a little shocked so I did some testing, again as Hatshepsut-Egypt, this time on Immortal and Sovereign. Leaned very hard into %prod boost towards wonders, rerolled until I had elephants, etc. In my Sovereign games I was averaging 4 Wonders, in my Immortal games just 2. Wasn't sure if this was a skill issue at first but checked out their capitals and more than half the civs were racing the others to the newest Wonder. The production and culture boosts on anything higher than Viceroy mean the player is really going to struggle at finishing the civic to unlock the Wonder before a few AI are already halfway through building it. For example, in one game me and Ibn Battuta unlocked the Byrsa wonder on the same turn. We had very similar capitals, but it would take me 24 turns to finish it - from start to end it took Ibn just 3 turns. I even play with that JNR mod that adds the cut Great Library wonder but it doesn't slow them down. It looks like they even prioritize Wonders over expansion which is a little odd - the AI in the wonder 'lead' would rarely have more than 3 or so settlements, whereas 'loser' civs would actually funnel that prod into settlers and expand.

 

[Morgawr]

Yeah exactly the point I've been thinking about as well. I think meeting their settlement cap (and maybe even going 1-2 higher) should be higher priority vs wonders because it provides much larger benefits in the long term and makes AI more competitive.

When an AI misses a wonder they're not funneling the "loser bonus" production to pop out settlers - not sure what they're doing with it, likely more wonders and buildings.

Other than a few critical wonders (those giving extra settlement limit, free general, war support etc) , wonders in Civ 7 are like glorified buildings... Not drastically game changing IMO. At deity the AI should still be able to meet / exceed their cap AND beat player to wonders after that, but not sacrificing the former .

 

[notque]

I think AI wonder chasing is currently a little too high. Rolled a Deity game as Hatshepsut-Egypt and only managed to secure one Wonder, Gate of All Nations (the AI hates the Discipline mastery for some reason). I was a little shocked so I did some testing, again as Hatshepsut-Egypt, this time on Immortal and Sovereign. Leaned very hard into %prod boost towards wonders, rerolled until I had elephants, etc. In my Sovereign games I was averaging 4 Wonders, in my Immortal games just 2. Wasn't sure if this was a skill issue at first but checked out their capitals and more than half the civs were racing the others to the newest Wonder. The production and culture boosts on anything higher than Viceroy mean the player is really going to struggle at finishing the civic to unlock the Wonder before a few AI are already halfway through building it. For example, in one game me and Ibn Battuta unlocked the Byrsa wonder on the same turn. We had very similar capitals, but it would take me 24 turns to finish it - from start to end it took Ibn just 3 turns. I even play with that JNR mod that adds the cut Great Library wonder but it doesn't slow them down. It looks like they even prioritize Wonders over expansion which is a little odd - the AI in the wonder 'lead' would rarely have more than 3 or so settlements, whereas 'loser' civs would actually funnel that prod into settlers and expand.

The only way it's too high is if the argument is they are doing worse from it.

That isn't what I see in general. Perhaps I'm missing it, but overall they seem to be doing better getting the wonders at a priority.

The player struggling is an argument it is effective, and reasonable. It means the AI is utilizing it's bonuses much more effectively in general.

I do need to increase their desire for Gate of All Nations then. Am I misunderstanding something? It also sounds like I need to check mods out, and make sure I support the AI taking all the wonders in mods as well.

 

[notque]

Yeah exactly the point I've been thinking about as well. I think meeting their settlement cap (and maybe even going 1-2 higher) should be higher priority vs wonders because it provides much larger benefits in the long term and makes AI more competitive.

When an AI misses a wonder they're not funneling the "loser bonus" production to pop out settlers - not sure what they're doing with it, likely more wonders and buildings.

Other than a few critical wonders (those giving extra settlement limit, free general, war support etc) , wonders in Civ 7 are like glorified buildings... Not drastically game changing IMO. At deity the AI should still be able to meet / exceed their cap AND beat player to wonders after that, but not sacrificing the former .

I agree I wish I had more control in funneling the loser bonus.

Wonders are less strong than in previous civilizations. But given the AI's production bonuses, Wonders are better than the vast majority of things the AI can build, and they don't take too long. In Autoplays, wonders are a good indication of high yields, and effective play from the AI.

What am I missing here? With their production bonuses, wonders are awesome for the AI, and the only negative is the everyone building it, which for sure, I can try to do a little more about, but gets complicated.

Often, if the AI is not building the wonders, they are just building more units. It's really hard to ever find a time where you'd argue the AI building a wonder is ever bad except in the case they haven't built enough cities.

And that's a hard calculation to do, because you don't know how big the map is. I don't have any particular way to manage how much space they have, and what makes sense.

 

[Morgawr]

I agree I wish I had more control in funneling the loser bonus.

Wonders are less strong than in previous civilizations. But given the AI's production bonuses, Wonders are better than the vast majority of things the AI can build, and they don't take too long. In Autoplays, wonders are a good indication of high yields, and effective play from the AI.

What am I missing here? With their production bonuses, wonders are awesome for the AI, and the only negative is the everyone building it, which for sure, I can try to do a little more about, but gets complicated.

Often, if the AI is not building the wonders, they are just building more units. It's really hard to ever find a time where you'd argue the AI building a wonder is ever bad except in the case they haven't built enough cities.

And that's a hard calculation to do, because you don't know how big the map is. I don't have any particular way to manage how much space they have, and what makes sense.

Ideally there would be a check like "if # settlements < cap, increase settler priority, lower wonder priority". I am talking about those edge cases where an AI is stuck at 2 settlements and only rushing for wonders thru most of ANT - that's not good. But to be honest I haven't seen anything that drastic myself. It did feel overall w/ your mod though that they were happier being 1-2 under their cap most of ANT in favor of wonders

As a player it's very tough to compare vanilla vs. mod because I play one game w/ vanilla, another w/ the mod the week after, but there are soooo many factors that go into AI performance (map, locations, civs, leaders, etc) that it's misleading / premature to draw conclusions from 2 games. You are doing your autoplays so I trust you have a much fairer comparison than us. I am just theorizing..

 

[skybsky]

I can confirm that with this mod, getting even 4 wonders done is way harder (I wasn't able to get 7 even when focusing hard on them).
AIs build wonders in 2/3 turns and just rush for them.
I was thinking I'm just bad with culture legacy path, but when I tried vanilla for the first time in many months, I was shocked that I was able to complete the cultural legacy path for the first time
I don't think wonders rush benefit them too much, though. Since when you win a war, they're quite often willing to give you that city with 5 wonders.
All of it is about playing on Deity, of course.

 

[Morgawr]

I don't think wonders rush benefit them too much, though. Since when you win a war, they're quite often willing to give you that city with 5 wonders.
All of it is about playing on Deity, of course.

Yeah that's a good point to add my argument above as well - maybe in AI-vs-AI autoplays this doesn't come up but as a human deity player, the counter-strategy to a wonder-focused AI is that human builds units and then captures their wonders, OR even easier DoW and pressure them to give you the wonders in peace deal. (But in the vanilla game the counter-strategy is to build more units and walls because the AI is also expanding more and building units.. so harder IMO)

I did exactly that in my modded Antiquity game and now I'm flying so high in EXP that I won't play MOD for sure because the game is essentially won.. I'm only playing to see how fast I can finish EXP. EDIT - I am bored, quit the game, whats the point knowing the outcome will be same as my previous 13 EXP games

In other words - a wonder-focused AI in early/mid ANT can be easy prey. (exception is e.g. they are far isolated on the map) An AI that has maxed their cap, built enough units and THEN spams wonders is something to be scared of --> which a deity AI should be able to do w/ all their bonuses, in that order

 

[Morgawr]

Sometimes a pic is worth a thousands words

Why aren't they at/over their cap yet?? This is Exp and the age only lasts 80 turns on average, I am already over halfway thru. I then force end the age (hitting milestones) and AIs left with empires way under their cap. Then they are building settlers in MOD era which is a joke, too little too late.

I bet AI-AI autoplays ages also last longer because AI is not playing like a human. This is what I mean when I say the AI loses its sense of urgency starting in mid-EXP. This is not mod related however, also in vanilla games (too SLOW to expand in EXP - I have no doubt they will get there by turn 130 or whatever but the age is not that long by design)

Modded game, mid-EXP

 

[Morgawr]

This one from my previous VANILLA game, end of Exp age - as an example to AI's lack of urgency in (not) meeting their cap:

My grand conclusion after playing 13 EXP games = in this age there are more THINGS for AI to build (vs in ANT there are fewer things to build) so it takes away from settler focus. It shouldn't because Civ 7 compared to other civ games is so much more about EXPAND and EXPLOIT

My asssessment from a few months back still holds today, in terms of AI performance (vanilla or mod)

Antiquity = (although we started this by discussing wonder priority, this age is still relatively solid IMO. Problems start in EXP per above)
Exploration =
Modern =

 

[notque]

I can confirm that with this mod, getting even 4 wonders done is way harder (I wasn't able to get 7 even when focusing hard on them).
AIs build wonders in 2/3 turns and just rush for them.
I was thinking I'm just bad with culture legacy path, but when I tried vanilla for the first time in many months, I was shocked that I was able to complete the cultural legacy path for the first time
I don't think wonders rush benefit them too much, though. Since when you win a war, they're quite often willing to give you that city with 5 wonders.
All of it is about playing on Deity, of course.

That has no impact on decision making from my end because it's a bug Firaxis needs to solve. Nothing I do has any impact on that.

 

[notque]

Sometimes a pic is worth a thousands words

Why aren't they at/over their cap yet?? This is Exp and the age only lasts 80 turns on average, I am already over halfway thru. I then force end the age (hitting milestones) and AIs left with empires way under their cap. Then they are building settlers in MOD era which is a joke, too little too late.

I bet AI-AI autoplays ages also last longer because AI is not playing like a human. This is what I mean when I say the AI loses its sense of urgency starting in mid-EXP. This is not mod related however, also in vanilla games (too SLOW to expand in EXP - I have no doubt they will get there by turn 130 or whatever but the age is not that long by design)

Modded game, mid-EXP
View attachment 731663

I am not following. I'm sorry. If the argument is the AI needs to push settlers a bit more consistently, okay.

 

[V. Soma]

I think in Antiquity AI (and also later) should pay attention getting a big enough army to defend, at least.
Especially so, as in a peace deal AI so easily gives away settlement...