TotalBiscuit thinks the AI is good. and i'm now willing to believe it...

[Lucius_]

Huh? I had XCom quite early and don't think it was buggy at all! XCom 2 as well.
Couldn't argue about BE or starships since I never played them... I'd say starships (and even BE to a lesser extent) were more of a 2nd /3rd priority release and not comparable with CiVI, the 25th anniversary edition of THE strategy flagship they have (no pressure, Ed!! ). OTOH CiV was obviously flawed and buggy like hell in vanilla... :hmmh:

I don't have the time to watch each and every playthrough but I see a lot of small bugs, inconsistencies in UI or small missing Infos from what I'm watching.. . This gives me hope this release is further away from the release version many here might fear...
It is build 1.011.xx or something alike , right? (on my mobile right now so I cannot check). Is there a rule of thump which version likely might be the release version? (2.02.x or 1.22.y or something like that) Or is it impossible to guess that?

XCOM had a major bug on the, IIRC, supply ship map? All aliens would spawn at the entrance of the ship. You'd run into a situation where you were fighting 8-12 aliens at once, depending on how many groups you already killed before triggering everything else on the ship. It ruined numerous of my ironman games.

There is no rule of thumb really, but usually version 1.xxxx is a vanilla version. Firaxis doesn't seem to change version numbers significantly between expansions, ie going from 1.x to 2.x. Anyway, I don't know the release build for CiV, but the first patch was 1.0.0.17. Relating that to the current build number of Civ VI in anyway is likely not possible.

 

[juanpavo]

Also where are these "epic AI battles" Ed Beach claimed the Firaxis teams were watching in May ? I haven't seen anything epic about the AI so far.

Maybe he was talking about the Apostles and Missionaries epic spam

 

[stealth_nsk]

When I read this forum, it looks like people treat the AI as some kind of lesser human being or such. It's just part of game mechanics, really. The AI from other games will not work in Civ6, because it has different rules. The AI will behave differently in different builds due to different rules/values. The same AI may spawn military units in June build and apostles in September build. AI CAN'T BE SMART - it has no mind. It can be challenging or not - in particular game, in particular part of the game, on particular difficulty level.

 

[Stringer1313]

Folks also don't seem to recognize that the introduction of agendas -- which folks seem generally to support -- has further hamstrung the AI even more b/c they are forced to play a role. I have no problem with giving the AI random bonuses to compensate

 

[Acken]

Folks also don't seem to recognize that the introduction of agendas -- which folks seem generally to support -- has further hamstrung the AI even more b/c they are forced to play a role. I have no problem with giving the AI random bonuses to compensate

If the AI used to play rationally you'd have a point but I seriously doubt agendas have much to do with AI performance.

 

[stealth_nsk]

If the AI used to play rationally you'd have a point but I seriously doubt agendas have much to do with AI performance.

Strategic decisions (like who to attack) are the easiest to code, so AI could be potentially strong in them. Agendas cripple this, making inefficient alliances and conflicts. But that's necessary, because "play to win" AI is much less interesting to play against - at some point human player will take lead which would cause all AIs to ally against the player. Possibility to ally with winning human player is surely ineffective, but is required for diplomacy to work.

 

[Acken]

But the AI was always bad at making alliances or chosing its ennemy. I fail to see how its suddenly worse.

Also its not their agenda that make them have issues teching or making units.

 

[Randall_Flagg]

I have watched many, many hours of the media release, I have seen nothing to suggest the AI on prince level will give even the lower than average player a challenge. It really seems like a Settler setting for learning. Our only hope is they at least get MP done right from the get go. That and a lot of players will play MP. I cannot ever see playing the AI other than 1st game to understand how it all works.

 

[Lord Shadow]

Nothing suggests the AI performance seen in the many weeks old preview build represents the current state of the AI, nor that said state is permanent and inflexible. There's no sensible reason to equate the past to the present nor the future.

 

[Mathias Rex]

Nothing suggests the AI performance seen in the many weeks old preview build represents the current state of the AI, nor that said state is permanent and inflexible. There's no sensible reason to equate the past to the present nor the future.

I would like to believe in this. Do you think that there is a possibility of the launch AI will be better than the preview build AI? The devs are on holiday now, probably the game is getting on gold this time around.

 

[juanpavo]

I would like to believe in this. Do you think that there is a possibility of the launch AI will be better than the preview build AI? The devs are on holiday now, probably the game is getting on gold this time around.

Remember that just because the Let's Plays came out last week, it doesn't mean that they were using the latest version as of last week. They likely used a stable version that was frozen a few weeks prior. Because of this pipelining, I do believe the launch AI will be better than the preview build AI.

 

[Mathias Rex]

Remember that just because the Let's Plays came out last week, it doesn't mean that they were using the latest version as of last week. They likely used a stable version that was frozen a few weeks prior. Because of this pipelining, I do believe the launch AI will be better than the preview build AI.

I hope you're right, i try to be optimistic

 

[ExemplarVoss]

Nothing suggests the AI performance seen in the many weeks old preview build represents the current state of the AI, nor that said state is permanent and inflexible. There's no sensible reason to equate the past to the present nor the future.

Just because it is potentially fixable doesn't mean we should discard past precedent or causality.

 

[Lord Shadow]

Just because it is potentially fixable doesn't mean we should discard past precedent or causality.

Sure, I never suggested that. Just that there's room for optimism, and not much of a point to jump to conclusions.

 

[PendragonWRB]

One thing I have not seen mentioned here is the lack of AI 'algorithms' that CHANGE with difficulty level. They need to add this instead of just the old AI 'cheats' mechanic. The 'cheat' mechanic giving them boosts is fine, but at each increasing level of difficulty, better 'algorithms' should be enabled so the AI actually gets 'smarter' as you raise the difficulty level.

I also hope these 'algorithms' are exposed to the modders so they can evolve with the game rather than just better 'settings' getting applied.

 

[Cyberian]

I have now watched about 50+ hours of total gameplay in the youtuber/streamer NDA release and the AI in there is totally failing. Civs are either collapsing to warriors or the barbarians alone or not growing/expanding. Civs not rising above total pop 5 after 50 or even 75 turns without player aggression seems to be a common occurence.

Hopefully this is already fixed in a Firaxis build or will be with Day1 patch.

 

[Lord Shadow]

One thing I have not seen mentioned here is the lack of AI 'algorithms' that CHANGE with difficulty level. They need to add this instead of just the old AI 'cheats' mechanic. The 'cheat' mechanic giving them boosts is fine, but at each increasing level of difficulty, better 'algorithms' should be enabled so the AI actually gets 'smarter' as you raise the difficulty level.

I also hope these 'algorithms' are exposed to the modders so they can evolve with the game rather than just better 'settings' getting applied.

If there were such a thing as a "too good" AI, staging algorithms like this would make sense. But there isn't. It's always desirable to have the AI play at its best, because that ideal scenario is purely theoretical, and in practice that "best" is still considerably below the capacity of an actual human. Cheats are currently the only means to give an AI an advantage on an otherwise level playing field, unfortunately.

 

[Acken]

To be honest i just want them to release the modding tools so that we can work on it ourselves.

 

[Smith of Lie]

Looking at some recent videos, mostly Marbozir's war against Spain in Greece game, combat AI seems little bit better compared to V. He is still crushing the AI, but once higher difficulty level bonuses kick in the AI might be a half decent challenge in a war.

 

[Olleus]

One thing I have not seen mentioned here is the lack of AI 'algorithms' that CHANGE with difficulty level. They need to add this instead of just the old AI 'cheats' mechanic. The 'cheat' mechanic giving them boosts is fine, but at each increasing level of difficulty, better 'algorithms' should be enabled so the AI actually gets 'smarter' as you raise the difficulty level.

I also hope these 'algorithms' are exposed to the modders so they can evolve with the game rather than just better 'settings' getting applied.

The problem is that the best algorithms we have are awful. There is no need to make them worse for anything but the very lowest difficulties (settler, warlord). Its easier and cheaper to give the human bonuses at those difficulties, and good enough.

What you are asking for is a better AI at high difficulties. We all want that. But saying it needs to be done is entirely moot as nobody knows how to do it in anything resembling sensible time and cost scales.