OK so the AI is bad; any good alternatives?

[Rapidly]

I'm trying to think if I've actually played any games that did in fact have good AI. Seems like every one that I've gotten at all serious about has had the same complaints. I'm wondering: are we expecting too much from the technology? Is good, or maybe even just competent, AI just not possible yet in the current state of gaming? If there are good strategy/tactical combo games out there with AI that really gives the single player at least a reasonable facsimile of a humanoid challenge, I for one'd sure like to hear about 'em. Please share!
p.s. When I was playing Sid Meier's Gettysburg and Antietam 5-6 years ago, those sure seemed to challenge the human player. But they'd prolly be unplayable in the XP world, more's the pity.

[ZZR Puig]

Yeah, I really enjoyed Gettysburg also. One of the best RTS I've played.

Let me think if I remember another one worth it... Civilization IV seems also far in the future and it's TBS... no, I can't remember any recent RTS that were challenging because of AI behaviour. But I think it's not a technology problem, it's just that seems to be easier to make nice graphics than smart AI.

The only games I know that have some similitudes with RTW are Imperial Glory and Stronghold 2, but I haven't played them, so I don't have any idea about how good or challenging they are.

If you are successful in your search, please, let all us know.

[Slyspy]

Very few AIs can challenge a player of any ability, because what we call AIs really are nothing of the sort. They cannot, after all, learn. However a game AI should be able to do two things to improve the challenge: It should be able to cheat to some extent to help itself, but this should not be so obvious as to annoy the player. This seems ok in RTW. Secondly the AI should be able to use the game's own internal mechanics. On this requirement RTW falls short. The AI is unable to build useful armies, unable to move those armies purposely around the map, it is unable to utilise the game's diplomacy system, it is unable to employ units in a tactical fashion and it is unable to respond usefully to events on the tactical battlefield. We are left with an AI that cannot even play its own game. Coupled with the bugs this is why RTW provides so little challenge and entertainment.

[Marcus Maxentius]

With games with "good" AI it that knows your moves too quickly and counters them too easily or efficiently(read robotically). Usually, these games give the AI cheats galore such as being able to build up faster than humanly possible. That's what turns me off to RTS's in general.

Rome Total War with it's faults is a step towards what I think is good AI. I'm not saying Rome has good AI, in fact it's moronic at times. What I liked when I started playing, was how the AI is beginning to play to the same rules as the human. It still gets some cheats. But I liked how fallible, or human, it seemed to be. In battle it makes tactical mistakes, but it is also capable of pulling off brillant manouvers. And in campaign each faction is not some kind of generic Napoleon or Sauron with different units. Each his different goals for bettering their nation-states. Spain for example is Isolationist and in my game Allies with other factions surrounding my own in self-defense. It's intent is to protect not expand its borders.

Now obviously these features are implemented poorly in the game. Spain doesn't build up much of a self-defense force and is crushed. But they're there for future games/expansion packs to build upon.

I'd pick this game over RTS's because I don't feel like this is The Matrix and Agent Smith is controlling the enemy. I don't want to have to be Neo.

[Red Harvest]

Civil War Bull Run has very strong AI. SM's Antietam and Gettysburg were also good.

It's not that good AI can't be written. It is that most companies don't have a strong interest in it.

[Spino]

Activision Value has a great little title called "Civil War - The Battle of Bull Run: Take Command 1861" (the full title is even longer!) that was created by a tiny (as in 2 guys and some helpers working for free) developer called Madminutegames. Now CWBBR won't win any awards for graphics and at first glance it looks like a souped up version of Sid Meier's Gettysburg but it is far more complex and challenging than most tactical strategy games you'll ever play. Forget about the point and click RTS style combat of Gettysburg or the TW games because CWBBR simulates (in an abstract fashion) the challenges and limitations confronting a 19th century commander at various organizational levels; army, division, brigade, etc. This includes leadership ratings, realistic fatigue and morale and above all, delayed execution of orders. For example, let's say you're commanding a division and want that brigade held in reserve to double time it to the front. You tell it where you want it to go and what formation and posture to assume when it arrives and then a mounted courier gallops from your general to its commander with the orders. To add to the suspense there's no guarantee that the commander will get the orders because couriers can be killed or captured by enemy units in the vicinity!

Best of all the AI in CWBBR seems to conduct itself with greater intelligence and ability than that found in the TW series. It's not brilliant but it rarely self destructs and massacres its troops like the AI in RTW does with alarming regularity.

The game is full of little moments that make it all worthwhile. My pet favorite is taking command of an entire division or army and sending the entire formation down a given road and deploying once they reach their destination. I also love those nail biting moments when you think you've gotten the best of an enemy brigade only to see fresh enemy troops arrive out of nowhere from a nearby treeline. You immediately request reinforcments and pray that your commander grants your request and sends those troops in time to save your bacon from the fire! CWBBR just feels right and it sucks you into the action in a way that no shallow game loaded with eye candy and little else can do. Excellent stuff.

Much like the Combat Mission series CWBBR has given computer wargames a serious kick into the modern age. It's a steal at $20 and is worth every penny.

http://www.madminutegames.com

[Navaros]

i don't know of any alternatives but i will throw in my 2 cents on the AI issue:

it always bewilders me that mankind can send rovers to Mars yet can not even program semi-decent AI into PC games. what the heck is up with that?!

i also do not think that the devs beefing up the AI's stats to godly levels is a suitable reason to consider a game as having "good AI". yet that seems to be what most games i've played (in which the AI can put up any sort a fight) do. they don't make it challenging because of good AI, but rather it's challenging because the AI cheats. that is a sorry state of AI, i say. mankind should be ashamed of itself for not yet evolving beyond the incredibly primitive state of PC game AI even in 2005.

[Crazed Rabbit]

Civ 3 has some good AI.

Rise of Nations, though a pure click fest RTS with some very abitrary limits (like the limits on the economy, and making every new unit of the same type cost more, as if you got worse at training units the more you trained), has good AI too.

Crazed Rabbit

[Uesugi Kenshin]

RON has good AI at times, but my favorite AI is Operation FLashpoint. So far I have not found squad or platoon level tactics simulated better in any game, it is an FPS, but remains hyper-realistic. I have been replaying it after my upgrade, the sad thing is even with its bad graphics no rig I have ever seen is able to deal with max settings and the largest draw distance, 5km.

[Gregoshi]

Best of all the AI in CWBBR seems to conduct itself with greater intelligence and ability than that found in the TW series. It's not brilliant but it rarely self destructs and massacres its troops like the AI in RTW does with alarming regularity.

Spino, I think you've just stumbled onto something brilliant! CWBBR's AI doesn't sound too true to the Civil War. CWBBR should be sending troops in to be pointlessly massacred - like RTW does. It seems to me that RTW should have CWBBR's AI and CWBBR should have RTW's AI! I wonder if a trade is possible?

Navaros, the AI for those little robots has been tested for years and backed by millions of dollars. If you've ever messed around with AI just a little, you can begin to appreciate how difficult it is to program it. If you want a simple taste of what creating AI is like, try writing down detailed instructions on how to make a piece of buttered toast. Take into account any problems you may encounter during the process and any corrective actions needed to handle the problem. You don't even need to know how to write a program. Use plain old english. It is not as simple as it sounds and we are just talking about toasting some bread and then buttering it. If you think it is easy, then you aren't thinking detailed enough. Example of things to consider: Is there bread in the house? What if there are no clean butter knives? How does the bread fit into the toaster? What if the "light/dark" knob is set too light or too dark? What if the butter is too hard (right out of the fridge)? Is there butter? What if there is butter and margarine - how do you tell the difference? And so on... Remember a computer is very literal - you have to tell it everything because it knows nothing. Now think about what it takes to move one unit on the battlefield and have it attack - or should it defend? My mind boggles at the thought.

Some day (hopefully soon) there will be a major breakthough in AI creation that makes it easier to build. And when we figure out how to make the AI learn, then games will get real interesting. Until then, we will have to put up with what we have today in AI or just play other humans.

[Benny Moore]

I'm also a Flashpointer, or was. It is the most realistic land combat game out there, yes, but it is not hyper-realistic. I'm not going to delve into its host of unrealisms and bugs that the developers will not fix because they feel it makes their game unique. Once more, people are victim of the erroneous belief that a realistic game is automatically not unique. Truly, I tell you, that among the most unique of games are games that strive for and excel at displaying realism. They are very few and far between. Indeed, I can think of no more than two in any type of game that even try, and only one in any type of game that succeeds at all.

[Red Harvest]

Gregoshi,

CWBR will slaughter its troops, but it does so at brigade level, rather than at corps level as in RTW. Individual commanders do their own thing and you have to learn what sort of stance orders to give to prevent them from self destructing. And regiments will often do something stupid, like wheeling or charging at inopportune moments. I've read a lot of Civil War after action reports and battle descriptions, and this fits reasonably well with what I've read.

On the other hand, left to their own devices, sometimes the brigades perform brilliantly and save your bacon. The AI is very good with artillery.

One thing I've noticed with CWBR is that all else being equal, I have my hands full winning a new scenario the first time out. I wish I could say the same for RTW.

It really is impressive what 2 people have done with the combat engine.

[PanzerJaeger]

I just looked over the posts about CWBR - one question?

Does it just cover one battle or many?

Thanks.

[Gregoshi]

I may have to take a second and closer look at CWBR. I checked out the web site a few months ago and watched one of the trailers but wasn't too impressed. Though I am impressed that it was done by two guys in a garage - old school game programming!! Thanks for the info Spino & RH.

[Spino]

I just looked over the posts about CWBR - one question?

Does it just cover one battle or many?

Thanks.

CWBR features one major battle (Bull Run/Manassas) and two minor ones (a new map, Morgan Hill, was introduced in the patch). The two smaller battles use fairly large maps so there's plenty of real estate to play on if you decide to fight a large engagement using the Open Play feature. The one cool thing about the large Bull Run map is that it also features the same ground fought over during the second battle of Bull Run/Manassas.

The 'Open Play' feature I mentioned allows for ahistorical battles with varying victory points, objectives and whatnot (you can also set many of the variables yourself). It greatly improves the game's replayability factor.

The good news is there will be a sequel to CWBR and it should be released sometime in 1st Quarter 2006. Madminutegames has stated their intent to proceed with the series in chronological order so I think it's a safe bet that the next installment will feature the battle of Shiloh. MMG also stated it will feature two minor battles as well. The developers are also in the hunt for a more powerful 3D engine for rendering purposes so I assume they're going to abandon the current one and try to up the detail levels and lessen the use of sprites.

I may have to take a second and closer look at CWBR. I checked out the web site a few months ago and watched one of the trailers but wasn't too impressed. Though I am impressed that it was done by two guys in a garage - old school game programming!! Thanks for the info Spino & RH.

Yes, the graphics are a bit hard on the eyes but you quickly forget about that after awhile. For a first time offering CWBR is a phenomenal achievement. For $20 it's a far better deal than paying to see two bad movies and losing four hours of your life you'll never get back!

[Fragony]

Wheeeeee another good reason to pimp Sacrifice. Yes, Sacrifice has good AI, actually it will tear you apart. Did I ever mention I like Sacrifice?

[Lord of the Isles]

Galactic Civilizations has pretty decent AI. What's more, at higher levels it actually performs more intelligently (rather than just getting bigger bonuses). Add different behaviour for each civilization on the good-evil plane and you get great replayability.

Pity the graphics are crap though.