Has anyone heard anything about the AI?

[olly]

Haven't seen much about CA saying how good the AI is going to be on the battlefield.

Has anyone heard anything as I hope that it will be worked on as well as the sea is going to be. The AI more or less makes the game IMHO.

[Fisherking]

Yes they said new AI and some stuff about handling military and diplomacy together so it doesn't offer a treaty and attack next turn...

[Trax]

As far as the battle AI is concerned, we've started from the best of the previous AI incarnations (the RTW:BI one), rebuilding it in the new codebase while adding several key new features such as goal-based planning and improved time awareness - for the first time in the TW series the battle AI will be projecting your unit's positions forward in time and evaluating your possible strategies at a high level in realtime, somewhat akin to what a chess program does offline, and it will be a lot better at making pre-emptive countermoves. We're still in full flow in the development, but it's starting to show some good promise.

link

[Bijo]

link

It does not remove my skepticism.

[Rhyfelwyr]

I've got into a battle routine in M2TW of pinning the enemy down with infantry and then flanking them with my cavalry from both sides. It would be nice to see the AI predict this and move spearmen to the flanks. Well the probably wouldn't have spearmen in ETW but you know what I mean.

[johnmcd]

I'm concerned that the emphasis seems to be on producing (beautiful) screenshots again rather than delivering AI that understands the game it's playing.

[Slyspy]

Haven't seen much about CA saying how good the AI is going to be on the battlefield.

Has anyone heard anything as I hope that it will be worked on as well as the sea is going to be. The AI more or less makes the game IMHO.

I would argue that the AI in RTW and MTW2 more or less broke the game rather than making it.

This time I'm going to assume that the AI will be rubbish so that I'm pleasantly surprised if it isn't.

Besides which I've long since decided to ignore dev's comments on what will be in a WIP since the most significant features promised hardly ever seem to make it into the final product.

[TB666]

I'm concerned that the emphasis seems to be on producing (beautiful) screenshots again rather than delivering AI that understands the game it's playing.

Well the person doing the screenshots isn't the same person doing the AI .
Anyway, the game is still young(I guess you can say that) and we got Lusted in now so hopefully the AI will be better.

[abdecken5]

I would argue that the AI in RTW and MTW2 more or less broke the game rather than making it.

I have to agree. the AI acts differently every turn, making it seem clueless, and in battle only has one (passive) tactic making it very easy to predict and defeat.

Besides which I've long since decided to ignore dev's comments on what will be in a WIP since the most significant features promised hardly ever seem to make it into the final product.

This is true and the problem is the track record of these last two games that dosnt seem to place Ai very high on the list of priorities. I cant really see what has changed we are always promised better AI.

not saying it wont be better but then again like the last few games it could be worse!

[Daithi MacGuillaCathair]

The AI is the major weekness in the game, its to easy to defeat and too unpredictable to bother with much in the way of diplomacy, basicaly any nation which border you will end up at war with you no matter how much you help them out.

the only time i have difficulty is when the faction lacks cavalry and i cant do the pin and flank with the cavalry.

or whe you see there general charging headlong for you spearmen, has the mentality of a world war one commander, Charge,and if it doesnt work the first few times its bound to work later

I hope they live up to there promises, time will tell.

[Puzz3D]

I think Jerome is wrong about the battle AI in RTW/BI being the best of the Total War series. Of course, he wasn't with CA when the older, first generation Total War games were designed. The older battle AI was well suited to the battle mechanics used in the older games which had greater tactical depth and a more statistically robust combat model than the newer second generation battle engine. The main deficiency of the older AI was it's inability to utilize ranged units to best effect, although, there was nothing wrong with its choice of targets. The problem was that the AI moved its ranged units too much.

Although the RTW AI was based on the older AI, it wasn't adapted to new features of the battle mechanics such as proper use of phalanx, proper use of shields or proper use of secondary weapon. This trend of introducing new features without making the AI aware of them is a chronic problem in the Total War series. For example, artillery was introduced in MTW, but the AI doesn't know enough to protect its artillery crews. The artillery can move in RTW and tries to stay with a moving army, but the AI still doesn't provide protection for the slow moving artillery.

Improving that AI's use of ranged units took a back seat to introducing new features, and old problems that had been solved (such as suicide general) returned in the new battle AI. Some of those problems were corrected in BI, but the battle AI remained inferior to the old AI. Units with shields routinely turn their backs on enemy ranged units and sustain heavy losses as a result. This happens because the AI sends single units forward, changes its mind after they get within proximity of many enemy units and then turns them around and walks them back to their lines. The RTW/BI AI will also allow an enemy ranged unit to shoot into the right (unshielded) side of a unit and not react at all even to move its unit back. The old AI at least moves its unit back, and into cover if it's available. Another issue is the RTW/BI AI doesn't know enough to keep it's army out of archery range of castles. The battle AI in MTW/VI knows enough to do that until a breach is made in the wall or the gate is knocked down.

In ETW, there will be no archers, no phalanx, no shields, no sieges (apparently), so these issues go away. It has artillery which will need protection and muskets which the AI will have to be made to use effectively along with a secondary weapon (bayonet). The artillery finally is the anti-personnel weapon that CA has insisted on making it in its ancient and medieval warfare games. So, tactically the land battles look simplier than the older games, and this should make it easier to make the land battle AI which is perhaps why Jerome can think about making an AI that projects the position into the future. However, if someone like Mike Simpson insists on lots of esoteric units, it's going to undermine the effort to make a competitive battle AI, and adversely affect the playbalance. ETW also needs to have a naval battle AI developed which is something that CA has no experience making, and judging from other games such as Privateer's Bounty it's not an easy task to make a good naval battle AI.

The PR for the game emphasizes the graphics. CA has said before that they have limited resources. There is a competition for resources between graphics and gameplay, and despite Jerome's best intentions, he has to do what Mike Simpson decides. Right now Mike Simpson thinks that being able to aim cannonballs at the men on the decks is a good feature worth spending time implimenting. It's rubbish from a tactical gameplay perspective because you couldn't aim cannonballs that accurately. Much more important would be talking about how you are going to prevent the AI ships from colliding with one another because, if they all just try to cross the T of enemy ships, it's going to be a mess.

Of concern is that Total War is no longer being being made for the fans that post here or anywhere else. Palamedes said that it's being made for a phantom majority that never posts.

[Noir]

Originally posted by Puzz3D
I think Jerome is wrong about the battle AI in RTW/BI being the best of the Total War series. Of course, he wasn't with CA when the older, first generation Total War games were designed. The older battle AI was well suited to the battle mechanics used in the older games which had greater tactical depth and a more statistically robust combat model than the newer second generation battle engine. The main deficiency of the older AI was it's inability to utilize ranged units to best effect, although, there was nothing wrong with its choice of targets. The problem was that the AI moved its ranged units too much.

I agree and yet because the engine is different it might have to do with the 3D men; i mean re-enacting the AI of STW/MTW and working from there might not be possible for ETW that will feature the 3D men and associated mechanics much like the RTW/M2 engine.

On top of that its interesting to hear CA saying that the RTW/BI was "the best AI" after all the marketing M2's AI has received; apparently it wasn't as enhanced as advertised i guess.

Originally posted by Puzz3D
The PR for the game emphasizes the graphics. CA has said before that they have limited resources. There is a competition for resources between graphics and gameplay, and despite Jerome's best intentions, he has to do what Mike Simpson decides. Right now Mike Simpson thinks that being able to aim cannonballs at the men on the decks is a good feature worth spending time implimenting. It's rubbish from a tactical gameplay perspective because you couldn't aim cannonballs that accurately. Much more important would be talking about how you are going to prevent the AI ships from colliding with one another because, if they all just try to cross the T of enemy ships, it's going to be a mess.

Developing two tactical AI's (naval and land battles) will definitely require lots of work and i hope that they'll dedicate the resources and hours for the task. The TW AI has never excelled in using shooters; as you said he can't provide support to firing lines while keeping their movement at a minimum to utilise and maximise fire output. AI shooters start by default on skirmish so far that really kills off AI use of them, especially for the more short range projectiles (grenades, javelins).

Also "hybrids" that is, shooter/melee units controlled by the AI so far needed to empty their shots before they engaged on; the era ETW is set will require alternate use of the melee/shooting capabilities many times possibly during the course of the battle as appropriate.

All this makes disciplined line tactical warfare impossible for the AI in all earlier TW releases. MP battles only can feature this sort of tactical play where the shooters keep the front aided by melee lines and cavalry harass and cover the flanks. The AI's timing in taking out shooters at the presence of enemy units is too early and that was part of the reason why shooter use wasn't effective. The so far TW AI is at its best in making melee matchups i feel and at producing a decent flanking every now and then.

I am really curious to see how all this will be adressed - if its done well it will offer great tactical depth for the SP game and interesting modification options for many periods. However if the AI isn't properly tuned with firing lines and the like the whole may end up too easy in the SP game.

Noir

[GFX707]

Of concern is that Total War is no longer being being made for the fans that post here or anywhere else. Palamedes said that it's being made for a phantom majority that never posts.

That's right. It's being made for the "holidays" crowd who'll see it sitting on the "number 1 game of all time" shelf with its inevitable 99% "best game in the entire universe" review by the reviewers who didn't bother to play the campaign or hang around long enough to see all the bugs and crap AI and buy it on impulse, and it doesn't matter because these are the parents of fickle kiddies who don't stick around long enough to complain or wait for patches.

[DisruptorX]

That's right. It's being made for the "holidays" crowd who'll see it sitting on the "number 1 game of all time" shelf with its inevitable 99% "best game in the entire universe" review by the reviewers who didn't bother to play the campaign or hang around long enough to see all the bugs and crap AI and buy it on impulse, and it doesn't matter because these are the parents of fickle kiddies who don't stick around long enough to complain or wait for patches.

Have you played other RTS games these days? They are all C&C clones.

Total War is easily the best Real Time series out there.

[Daithi MacGuillaCathair]

unfortunatly so, if its the best it doesnt say much for the competition, i think many of the silent majority would also like an enhanced AI, three of my brothers play it and never post on forums but hold many of the same views expressed here. i think many people dont post becuase they dont think it would make a difference.

[GFX707]

Have you played other RTS games these days? They are all C&C clones.

Total War is easily the best Real Time series out there.

I agree, and it's still its own genre, which is what puzzles me after 7 years of the series and not one real clone.

[TosaInu]

I agree, and it's still its own genre, which is what puzzles me after 7 years of the series and not one real clone.

Yes. I recall it being said that it's quite a complex job to do.

[Fritz]

Has anyone heard anything about the AI?

What does it matter what anyone from CA/SEGA says until the game is released?

As the saying goes, talk is cheap.


As far as a competent AI, I'll believe it when I see it. M2TW is too much style over substance. Looks, bells and whistles over challenging and logical gameplay. Can we expect anything different with Empire?

I'm damn cynical and I'm abusing way too many cliches but I'm frustrated by the lack of emphasis on good AI in TW.

CA: please get it right this time.