AI Artillery usage

[Berrern]

I always get so tired of the game in the late stages because I can easily capture any AI city with my massive Artillery stacks.

Now, what I'm curious about is why doesn't the AI build Artillery? This makes it so unfair..
Yes, they sometimes have an artillery unit or two in their cities, but I've never seen a SOD with Artillery. It seems that they like to keep their Artillery units inside the cities, and only use them for defense.

Now, is this because the AI is poorly programmed when it comes to AI? Or is it simply because they prioritize other units? If the latter is the case, isn't it possible to mod the AI to produce more Artillery in the editor? Or will this only result in them keeping enormous stacks inside their cities?

I want a challenge - I want the AI to attack me with 50+ Artillery pieces! They DO use Bombers a lot when they've discovered Flight, but that doesn't compensate for the lack of bombardment units in the Ancient and Middle ages

Anyone have some extensive information about this issue?

 

[sealman]

Your right. The AI does not know how to properly use Artillery to attack cities. It primarily uses them as defensive measures for cites. That does not really annoy me, what does is that if say an AI city has 2 pieces of artillery in a city and I just sit my SOD right outside, the AI will rarely shell my units. However, the AI will bombard me with naval forces and bombers.

But alas, I do not know of a way to get the AI to use artillery more effectivly

 

[Tomoyo]

I think Ision made a thread were he saw an AI invade him th 10+ cannons, and he actuallly lost the city in question!

 

[Dogmeat]

Originally posted by Tomoyo
I think Ision made a thread were he saw an AI invade him th 10+ cannons, and he actuallly lost the city in question!

I've found out that the AI plays really smart on some rare occasions as well.

 

[Jon Shafer]

AI can't use artillery. After hundreds of hours of playing Civ 3 since it was released I've only seen the AI send out a bombardment unit and attack with it 3 or 4 times. And in those times it was only 1 unit.

One of the reasons the AI's use of artillery was not previously a priority was because the developers were concerned about the AI sending out a large stack of bombardment units and the human would then focus everything on the AI's stack in order steal the artillery.

 

[Smellincoffee]

I've only seen artillery used in cities as defense. And not very well, either, because my knight armies don't really CARE about their pathetic trebuchets.

One of the reasons the AI's use of artillery was not previously a priority was because the developers were concerned about the AI sending out a large stack of bombardment units and the human would then focus everything on the AI's stack in order steal the artillery

And we would. I know I would. Won't the "requires escort" tag take care of most of that, though? Especially if that tag being checked meant the AI would REALLY protect it?

 

[Gogf]

Well, this would kill newbies who didn't have access to a forum like CFC, had no idea how to play.

 

[Tomoyo]

Oops. DP.

 

[Tomoyo]

If the AI marched 30 arty with 30 infantry into my territory, I would bombard all of it, and chop it down, and steal the arty during the retreat.

 

[RegentMan]

If they do change the AI to bring 50+ artillery into your territory, then they'll need to change the way artillery is captured. Perhaps have only a small % survive?

 

[SuperBeaverInc.]

If the AI is going to use artillery in large stacks, then it would definetly have to be revised. Otherwise, I could capture enough artillery to supply my army iwthout building any myself.

 

[King Alexander]

It's not a case of bad programming: it is a case of intentional programming. The developers could make the AI extremely hard to defeat(if impossible), but that would take the player's joy away for qonquering AI cities(e.g.: if the AI have railroads and the player invades, then the AI could move ALL it's infantries and artilleries on the front cities right away). I've read many topics on the Internet and magazines about what difficulty should the AI have, as I'm sure you've read, too.

 

[yoshi74]

I think the main problem is that artillery can be captured. Programming the ai to make much more artillery would ultimatly result in much more artillery for the player, which in turn would make him stronger.
While in several games the ai is sometimes toned down to allow humans the victory, i doubt its the case for strategy games like civ. A modern civ war between humans would be interesting to watch. I think both sides will actually build less artillery, because the risk to lose a battle and granting the enemy so many free guns would result in the defeat at all.
So the tactic of mass artillery is the usage of an ai weakness. A more competent ai would result in less artillery used.

 

[Samson]

It's not a case of bad programming: it is a case of intentional programming. The developers could make the AI extremely hard to defeat(if impossible), but that would take the player's joy away for qonquering AI cities(e.g.: if the AI have railroads and the player invades, then the AI could move ALL it's infantries and artilleries on the front cities right away). I've read many topics on the Internet and magazines about what difficulty should the AI have, as I'm sure you've read, too.

Are you sure? I am not dissing Fraxis, or saying it is *bad* programming, but I always assumed the AI was given such massive bonuses was because it was either too difficulkt to make the AI clever enough, or would take too long for them to play their turn. Surely civ would be a much more fun game if the AI had less bonuses, but made war more like a human player.

I do not know what I am on about though, I have not read any articales, on or of line.

 

[King Alexander]

I know what you mean: Our current technology doesn't permit the AI to perform like a human player. Maybe, it's a combination of both the technology and keeping the AI somewhat limited in options(even at higher difficulty levels), so the player can still enjoy the game. The advantages the AI gets at higher levels can be overcome from a really good player.

 

[yoshi74]

I don't like the term 'bad programming' in this case anyway, because it is not bad programming. Programming the ai for such a complex game like civ is far from trivial. The programmers did a very good job.
No one should think because modern chess programs can beat the chess world champion, all problems of ai programming are solved and a not so bright ai results out of poor programming. Actually, chess computers work very differently from the human mind when it gets to playing. The computer just trys every possible move, calculates some moves ahead and chooses the best one. There are some algorithms who stop searching on really bad moves to shorten the time, but still it is mostly dumb trying. Also the success of the computer chess came mostly due to faster computers, most professional chess machines are specificly build in hardware to get the maximal speed.
However, such a approch can't be used for a game like civ. A chess board has 64 fields, a civ map up to 10000 and more fields. The ai could be programmed much better, but then you have to wait for the next turn several weeks or more. Not an option, right.

 

[Samson]

No one should think because modern chess programs can beat the chess world champion, all problems of ai programming are solved and a not so bright ai results out of poor programming

Quite. You only need to look at how the best programs do at go (about as good as a human who has just about picked up the basics I belive) to see how hard the problem is.

Quite why the computer sends its units in a few at a time, rather than waiting a few turns to send in a decent SoD that may have a chance of doing some damage, I do not know. I am sure there is a reason though.

 

[yoshi74]

Actually the ai already sends stacks quite often. However, planning several turns ahead is a difficult matter, because the problem is a exponential one. Good civ players plan often (especially before wars) 20+ turns and more ahead. Many decisions are bases on plans even longer ahead. Unless the computer is teached to think in a similar way as the human that would cost too much time.

 

[Samson]

Actually the ai already sends stacks quite often

I have seen stacks when I first go to war against an AI, as they have had time to build them up. It always seems to me that once the war is ongoing, they just send the unit over as and when they are built. Last game, I had a city at a choke point, and every turn the AI would send between 3 and 6 units at it, not always in the same square!!? Great for producing leaders though

However, planning several turns ahead is a difficult matter, because the problem is a exponential one. Good civ players plan often (especially before wars) 20+ turns and more ahead.

I know what you meen, but when I sit there, watching the AI units come at me like lambs to the slaugter, that a simple bit of code like this would help -

if(units_I_have_to_send_this_turn < units_likely_to_be_needed_to_take_city)
{
wait_untill_next_turn_to_attck();
}

I am sure that the developers have thought of this though, and there are very good reasons why it would not work.

 

[yoshi74]

Waiting too long could also result in more human units in this city.

But your right, many attacks are just rediculous. But when waiting to get enough units to beat the 20 units you maybe stacked in this city, the ai may never come to the conclusion to attack. And then people cry about the passive ai

And there is again the problem to give the ai a good enough ai to actually come with a plan to hit where you don't expect this.