Problems still remain

[Haggbart]

Yes. But the problem is - AI doesn't seem to threat the conquest victory, or at least significant conquest. If I play on continents, I expect the other continent to sometimes be dominated by a single AI - usually a militarist with a good start. If this never happens, it would look a bit wrong to me.

I would guess that the agendas would make the human player the victim of DOW on higher difficulty levels. Because the bonuses should make the AI easier to fullfill each other's agendas, while the human player will struggle to maintain friendly relationships with several AI's.

You can even argue that tuning difficulty levels is a lot easier without one snowballing AI in a high percentage of the games. But I agree that at least the aggressive leaders should try (and succeed a reasonable amount of times) to conquer the other AI.

 

[skyclad]

No. The AI is the best I've seen in civ games. But the game (not AI) still have some problems.

Complains often come from not understanding. "Oh no, so many warriors, everything is lost".

The AI appears to have severe problems:

It settles cities across the world but not next to the wonder next door.
It goes to war but sends no units.
" joint war" does not work because neither AI will attack.
It doesnt use settlers theyve had for 100s of turns.
It doesnt repair damaged improvments.
It will sell cities for great works.
It has the tech and district for science victory but doesnt go for it.

Problems you can blame on "the game, not the AI" is resource requirements = lack of upgrades, and warmonger penalties being too high = avoids wars in late eras. The rest seems to be with the AI.

 

[qadams]

City state unit spam. Give them a hard cap on number of units. I honestly don't care what they do with their production, it has zero game effect. As long as they can place their district, their improvements, and their builders (for both improvements and repairing). Once those things are done or unnecessary, force them to do a special project to produce gold (which they can use to upgrade units).

I agree with this, along with having an upgradable line of units from era to era that does not require strategic resources but is not as strong in each era as units that do.

 

[stealth_nsk]

The AI appears to have severe problems:

It settles cities across the world but not next to the wonder next door.
It goes to war but sends no units.
" joint war" does not work because neither AI will attack.
It doesnt use settlers theyve had for 100s of turns.
It doesnt repair damaged improvments.
It will sell cities for great works.
It has the tech and district for science victory but doesnt go for it.

1. Not a problem. AI settles remote cities in rare and understandable conditions.
2-3. On low difficulty levels AI has reasons to attack, but not forces to do so. Not sure whether it's problem or not.
3-4. A small issue of building priority.
5. A small bug in diplo values.
6. It's not a problem as AI don't have a goal to rush for victory.

 

[Al Bundy]

I agree with stealth_nsk. Those problems can (and will) be solved. Let's stay positive and hope for the best. We all want the same thing: a CIV game as good as possible.

I already preordered, so tomorrow you won't see me here in the forums. I'll be busy...

 

[fay902]

Take it easy, man. I am also a big fan of CIV. I already played CIV I when I was a teenager, but the AI looks bad and unimproved since the last stream. That's a fact, whether you like it or not.

Yeah they should just whip out a new and awesomely improved AI in two weeks.

 

[skyclad]

1. Not a problem. AI settles remote cities in rare and understandable conditions.
2-3. On low difficulty levels AI has reasons to attack, but not forces to do so. Not sure whether it's problem or not.
3-4. A small issue of building priority.
5. A small bug in diplo values.
6. It's not a problem as AI don't have a goal to rush for victory.

Lol you act like you know more than the rest of us. We all saw the streams and made our own conclusions.
At least you admitted that the AI isnt perfect, even if we disagree with what is a problem or not.

 

[gfeier]

After watching quite a few playthroughs I see the problems more as balance issues than AI issues. The first patch should fix the most egregious ones, so I'm less concerned than some others here. I'm still looking forward to playing.

 

[Ikael]

Very good analysis, Stealth. There are lot of problems that looks like AI failures, yet they aren't, such as city states building tons of units since they have far less building options than regular civs (no settlers nor wonders nor city projects to build).

Perhaps a way to avoid it would be to add an increasing production cost for each city state military unit already produced, a la settler penalty. That way, city states would be able to replace war casualties easily and have a middle sized army, but would also be unable to generate carpets of doom.

 

[Paramecium]

I have to agree, that the AI might still look weak, but most problems seem to be fixable. They just have to crack a nut what to do with those unit upgrade paths. What most people might forget is the fact, that you can get quite huge numbers of Gold if you sell your "old" units. So if the AI cant handle to get its hand on iron or niter to upgrade its warriors, it should sell most of them and replace it with stronger units available. Just another thought on solving the modern age warrior swarms. Or yeah, just add all gold costs together if you already can build musketmen after swordmen, but if you dont have iron but niter. Or in general upgrade to the most up to date unit, so the AI can mass upgrade its warriors when it has reached replacable components.

In general, I hope we will see a lot of patches in the coming months which will improve those bugs/tweaks/weaknesses we have seen so far. But I think most problems will be handled it. But we will see tomorrow how it will play out. Especially that the AI doesnt seem to be that aggressive should be quite easy to fix.

 

[KrikkitTwo]

I think changing the strat resource effect (for AI and players) is a good effect

1-upgrade to this unit, or build this unit in encampment/harbor
2+ upgrade to this unit, build unit everywhere
Then
0-upgrade to this unit for 3x normal cost, build this unit in encampment/harbor for 3x normal cost

Also make all upgrades "direct upgrades" warriors direct to infantry, slingers direct to field cannons,

 

[Seek]

I agree with the posters above who suggest lowering warmonger penalties, especially for CB wars - the AI is too passive. I wonder if the AI is able to tell how much military other civs have and make that a factor in its decision-making process wrt DoWing and upgrading units (DoW more aggressively if neighbor has low military, upgrade more aggressively if neighbor has high military).

The AI appears to have severe problems:

It settles cities across the world but not next to the wonder next door.
It goes to war but sends no units.
" joint war" does not work because neither AI will attack.
It doesnt use settlers theyve had for 100s of turns.
It doesnt repair damaged improvments.
It will sell cities for great works.
It has the tech and district for science victory but doesnt go for it.

The unsettled Natural Wonders, agreeing to Joint Wars, stuck settlers, unrepairded improvements, and great work trade issues actually all seem like bugs to me, and hopefully will be patched/modded in short order.

The other issues (not sending units to war, not pursuing victory) don't seem like it would take much to correct, just in terms of adjusting values for the AI (though Civ5 had/has these issues too so who knows).

Also make all upgrades "direct upgrades" warriors direct to infantry, slingers direct to field cannons,

Direct upgrades are already in the game, so there's really no excuse for not upgrading. If the issue is with the AI not prioritizing upgrades because of gold (both upgrade costs and maintenance), then give the AI a higher priority and/or further bonuses for gold. It seems fairly straightforward.

 

[gunnerxtr]

Nice to see some positivity coming around! Especially for those of us who pre-ordered! Tonight will be fun and we will work through the bugs.

 

[Furycrab]

Yes you are right if the goal is to make the AI play well. But I don't think that is the goal to achieve. If an AI is able to surprise another by having more units then that is fun and dynamic. I don't need the AI to be a master strategist. I just need it to be competent in its actions. For example if it choses to go to war, it really make an army and attack. I also want it to be dynamic and able to put pressure on a player.
I personally enjoy an overagressive AI, to the point it neglects defense, a lot more than one that turtles for 300 turns.

And from what the lead AI designer has blabbered about I'm a bit worried he wants something totallly different for his AI. Not sure it will work well with the fans.

If that's what you want, at least label it properly... You want an AI that is effectively worst at the game. If an AI is able to surprise and kill another AI, it means the defending AI somehow didn't know whatever was happening was possible.

The problem often in Civ is the mechanics, often Gold. You prepare an army, declare, set out to attack, I buy a wall and units closest to where you are attacking, maybe levy an army and all the sudden it's certainly possible that the right decision is to stop, you've forced your opponent to spend it's gold on units. The dumb thing is if you still decide to attack, lose most of your army and then your opponent takes the units you forced him to buy to take some of your cities. Maybe you think it's fun and dynamic, but once your players learn how to properly exploit an AI that does this, you then need to compensate for this dumb behavior with more bonuses. Next thing you know... Deity AI gets Hanging Gardens on turn 12.

If the AI was really good, I'd argue that it should be almost impossible for one AI to crush another AI early unless someone had a ridiculous start, or there were some really bad underlying balance decisions, or if diplomacy can be leveraged properly to produce a 2 vs 1 scenario. If I was an AI Designer, I think that's the goal I would aim for first, and then let the players decide how they want to generate asymmetrical gameplay by perhaps modding the AI the be overly aggressive and take more early game gambits.

 

[Acken]

If that's what you want, at least label it properly... You want an AI that is effectively worst at the game. If an AI is able to surprise and kill another AI, it means the defending AI somehow didn't know whatever was happening was possible.

The problem often in Civ is the mechanics, often Gold. You prepare an army, declare, set out to attack, I buy a wall and units closest to where you are attacking, maybe levy an army and all the sudden it's certainly possible that the right decision is to stop, you've forced your opponent to spend it's gold on units. The dumb thing is if you still decide to attack, lose most of your army and then your opponent takes the units you forced him to buy to take some of your cities. Maybe you think it's fun and dynamic, but once your players learn how to properly exploit an AI that does this, you then need to compensate for this dumb behavior with more bonuses. Next thing you know... Deity AI gets Hanging Gardens on turn 12.

If the AI was really good, I'd argue that it should be almost impossible for one AI to crush another AI early unless someone had a ridiculous start, or there were some really bad underlying balance decisions, or if diplomacy can be leveraged properly to produce a 2 vs 1 scenario. If I was an AI Designer, I think that's the goal I would aim for first, and then let the players decide how they want to generate asymmetrical gameplay by perhaps modding the AI the be overly aggressive and take more early game gambits.

And I dont feel why you had to repeat yourself. It was pretty clear the first time. I dont have anything more to add than I already did. I also think my point of view was pretty clear the first time.
Ill just add that if its so simple to defend as you describe then the defender advantage is too big.

 

[lorrydriverjr]

For those fearing the AI is to passive: TPangolin just got a joint war declaration by Germany and Brazil after 34 turns.

 

[jlc22]

For those fearing the AI is to passive: TPangolin just got a joint war declaration by Germany and Brazil after 34 turns.

but will the AI actually attack with anything... According to the Battle Royale, then no.

Either way I'm still excited to get my hands on the game.

 

[Danzo_87]

Okay I am off to sleep, couple of observation:
Played as Greece, on King difficulty, Pangea, standard speed

Rome declared surprise war against me on turn 20, attacked with a slinger and 4-5 warriors, I had 2 slingers and my warrior was quite far away scouting, Trajan got my capital down until the red before I fend off his army

then few turns later Gandhi (!!) and Pedro declared formal war together, and they did sent troops towards my cities

I feel like the harder difficulty you play, the easier you will trigger the AI's negative agenda towards you, therefore expect loads of hostility from the get go

 

[DSveno]

I'm still trying at King difficulty, and honestly I have no idea what's happening but after turn 45, there are 3 reports that someone has been defeated. I would suspect they actually attacking each others or something. Wonder if I can save a replay and rewatch it after finishing.

 

[chazzycat]

Everyone should keep in mind the "Battle Royale" was on king, and all previous vids were on Prince. According to the other thread for the German live-stream happening now, immortal looks to be very different.