Will the AI be able to handle all that uniqueness?

[JtW]

After watching that last interview where they disclosed that every civ will have 4 uniques, I must say I feel conflicted. On one hand, it's fun to have so much unique elements in the game: it looks like not only civ, but also every city state, every great person, and even many wonders will work in pretty unique ways. On the other hand, I am pretty sure it won't be possible to balance all these uniques, but more importantly, I doubt that the AI will be able to understand them. Effectively, it will make the AI appear more incompetent than it really is.

What do you guys think? Is there any chance the AI will be able to deal with all that variety? Or do you share my apprehension?

 

[AriochIV]

Will it be able to use them "optimally"? Probably not. But even in Civ V I haven't noticed any serious problems of the AI being unable to use its unique units or abilities.

 

[SupremacyKing2]

For me, it is not a big deal if the AI cannot handle all the uniques super well because ultimately the unique bonuses are meant for the player to enjoy. I am the one who will be enjoying the different play styles that each unique bonus brings.

 

[Eagle Pursuit]

It's definitely a concern, but we won't be able to quantify it for some time yet.

 

[Stringer1313]

I think this is a small issue compared to the much larger issue of AI generally. People will never be satisfied with the AI, period, in a game as a complicated as this. I am not sure I have read a single post of someone saying they were happy with the AI in any particular game (not just Civ).

 

[m15a]

I don't think 4 uniques is necessarily more difficult for the AI to understand than 3 uniques. It's more a question of the nature of the specific civ bonuses and era bonuses used, which could potentially be difficult to program for. I doubt there's any synergy between all 4 bonuses that would make there be greater complexity.

It'd be nice if the AI could use all the bonuses, but personally I would prefer interesting bonuses that the AI has trouble with vs bland bonuses the AI can handle. (Although a lot of the Civ V bonuses are both interesting and apparently not too hard to make work.)

 

[AriochIV]

Let's put it this way: handling 4 uniques is at most 25% more of a challenge for the AI than handling 3 uniques. I have bigger worries for the AI.

 

[m15a]

~33%. But, yeah, computers are good at doing more of the same thing.

 

[stiiknafuulia]

As long as they get the *combat* AI to perform half-decently (no embarking in the water to be slaughtered tyvm), I don't really care that much. To me the combat AI is the most important aspect of the overall AI, because if it's trivial to beat, then you could in principle ignore most other mechanics and just steamroll the whole world with your one army, if you could stand the enormous micromanagement. The only reason to play peacefully will then be because you don't want all those extra clicks and head-problems from the otherwise optimal wars that you'd have to wage all-game-long. That and role-playing reasons, ofc.

Otoh if the AI can't use its districts or Great Walls (etc) optimally, but can fight correctly, then it doesn't really matter that much, as they will get economic bonuses on the higher difficulties and be able to compete that way.

 

[Krajzen]

I think combat AI may be better this time because of seemingly smaller traffic jam problems (cities apparently occupying more space, support units, 2-3 unit stacks possible etc)

It's however balanced by my fear of AI handling of district placement and tile usage

 

[Ryika]

Let's put it this way: handling 4 uniques is at most 25% more of a challenge for the AI than handling 3 uniques. I have bigger worries for the AI.

That's not how math works.

But yeah, @topic: I think Uniques aren't much of a deal for the AI. It'll just get number bonuses and play a more generic game. Which is fine, the important thing is whether Uniques do the job of creating interesting and diverse gameplay for the players... and on the part of the AI... well, whether the Combat AI has been improved.

 

[Babri]

It's however balanced by my fear of AI handling of district placement and tile usage

IMHO city placement would be more tricky business for the AI. If they get it right, then it will be much simpler to implement district logic properly.

Sent from my HTC Desire 820s dual sim using Tapatalk

 

[Amrunril]

IMHO city placement would be more tricky business for the AI. If they get it right, then it will be much simpler to implement district logic properly.

The AI will presumably be quite competent when it comes time to actually place a district on the map. What may be harder for it is taking future district placement into account when laying out cities. This could be an issue for city placement, but would also impact district and wonder organization. The AI should always be able to put down its scientific district in the best spot, but what it may not be able to do is, given two relatively even spots, choose the one that better sets up adjacency bonuses for a future culture district.

As for AI use of uniques, I feel compelled to mention Babylonian Great Scientists, but really this is more a case of a UA exacerbating a preexisting strategy issue. Otherwise, I'll admit I haven't really noticed how well or badly the AI uses its uniques (the buildings are generally hidden from view, and the AI uses all units badly), but I don't thinkit should be any harder for it to handle 4 abilities than to handle 3. Multiple leaders per civ (or leaders interchangeable in advanced setup) could actually be a bigger issue, as I think a lot of the AI's use of uniques is built into individual leaders' strategy flavors (that is, Alex being individually programmed to focus on city states rather than focusing on them in response to his UA). That's still a fairly hypothetical concern at this point though, and presumably the developers would make some effort to address it if it did become relevant.

 

[dexters]

A lot of uniques will be UUs with a combat bonus, a resource requirement tweak, or being simply being the flat out best unit in its class type for an extended period. These are things the AI can handle as it works along a continuum of the battle mechanics.

UUs that do unique things like build improvements/districts/roads/ may be more of a challenge and these tend to get short changed especially when coming in balance patches, expansions etc. where the feature is added but the devs simply forget to fix the AI to use it.

UAbilities/Traits/Agendas may or may not require additional AI work.
Traits/abilities that are rules based (ie: all cities on same continent get X) will probably need some AI flavour modifiers telling the AI to settle as many cities as possible in its home content,

Agendas will require the AI the proper tools to understand the concepts. This is actually my biggest concern. You can't make agendas the AI can't handle properly, but at first glance, they've done well.

China's wonder agenda is a pretty easy metric the AI can measure
so is Egypt's agenda of being friendly with another Civ with a larger military and hostile to those weaker than them.

America's agenda is most challenging, but since continents now have discrete labels, and one assumes all landforms can be tracked by the game , the big stick policy makes sense since the AI can easily check the relevant info for its home continent Y.

 

[Thormodr]

I'm not to worried about the AI being able to handle their unique abilities. I'm more concerned about them handling 1UPT. It was utterly befuddled by it in Civilization 5.

 

[Acken]

Depends what. Units buildings and districts should be fine unless they again make micro intensive units like Keshiks.

Abilities will depend on whether or not the ability is AI friendly. For example, Civ5 poland UA is very AI friendly because its very passive. Babylon isn't because the AI doesn't understand how to use great scientist or how to really generate them. Or worse, something like Denmark.

 

[shaglio]

Will it be able to use them "optimally"? Probably not. But even in Civ V I haven't noticed any serious problems of the AI being unable to use its unique units or abilities.

William of Orange notoriously refused to trade away the last copy of his luxuries despite retaining half the happiness. But, that's the only one I can think of off the top of my head.

 

[stealth_nsk]

It's not a problem with AI. You don't need specific algorithms to handle unique units or abilities, you just need corrected "weight" values in tables. Some things may require specific algorithms, though, like Great Wall UI, but I expect them to be not that common.

 

[joncnunn]

I don't see additional additional UUs and UBs increasing the complexity. These just have different values on the weighting table than the units & buildings they are replacing. UUs & UBs that are better than standard will naturally be higher weights. And if for whatever reason Firaxis were to introduce each civ also getting a UU that's worse than the standard one in addition to the normal UU better than the standard one, the "weak UU" will naturally have lower weights.

The more complex part is teaching the AIs how to exploit their UA.

 

[King Jason]

The more complex part is teaching the AIs how to exploit their UA.

Really this - There's a huge difference between a civ that gets a bonus to early (or all) wonder production as a base trait - versus a civ that needs to use a unit in a different way, even if it's a slight difference, as in the case of Quin's worker being able to rush wonders.

Also wonder-spam in general in older versions of civ versus now adds more complexity to Quin's trait and agenda.

Before, it would just be "build lots of wonders, all the time, and they will build really fast"

Now, wonder placement, district capacity, and worker chargers all need to be used properly.

I remain optimistic, however.