Community Feedback Needed: Biggest AI issues
- Thread starter Kael
- Start date
Monkeyfinger
Deity
- Joined
- Oct 24, 2006
- Messages
- 2,002
It's annoying and adds nothing to the game. The AI is as fair as ever when trading for resources it lacks, so it's clearly meant to still do fair deals. Just, thanks to a coding snafu which you can read about in my second link, the feature is broken and not working as intended.
CruX
Chieftain
- Joined
- May 14, 2008
- Messages
- 28
I think this would be a simple one. Force the Luchuirp to protect Barnaxus in a stack of a least a few units rather than leaving it wandering alone, even during peacetime.
In one of my games, I don't know why the AI send Barnaxus alone close up to my border. I subsequently declare war on them just to kill Barnaxus and then bring the Pieces of Barnaxus back to my city. I made no effort to push forward and simply wait out the next dozen turn and then ask for peace. I think the AI need to know how to protect Barnaxus better.
Secondly, could it be made that the AI actively seek back the Pieces of Barnaxus? And is it possible that the Luchuirp has a diplomacy penalty with whoever that own the Pieces of Barnaxus as well? (Well, this is obviously a balance feedback rather than AI issue, but if this could help the Luchuirp to actively seek back the Pieces of Barnaxus, then why not?)
Lastly, do the Luchuirp know how to rebuild Barnaxus?
In one of my games, I don't know why the AI send Barnaxus alone close up to my border. I subsequently declare war on them just to kill Barnaxus and then bring the Pieces of Barnaxus back to my city. I made no effort to push forward and simply wait out the next dozen turn and then ask for peace. I think the AI need to know how to protect Barnaxus better.
Secondly, could it be made that the AI actively seek back the Pieces of Barnaxus? And is it possible that the Luchuirp has a diplomacy penalty with whoever that own the Pieces of Barnaxus as well? (Well, this is obviously a balance feedback rather than AI issue, but if this could help the Luchuirp to actively seek back the Pieces of Barnaxus, then why not?)
Lastly, do the Luchuirp know how to rebuild Barnaxus?
Caradoc
Emperor
- Joined
- Dec 21, 2005
- Messages
- 1,010
It might help if each AI had a few 'scripts' to use. These would consist of a series of techs leading to steady development along one of the paths used by players (melee, archery, sorcery, religion, etc.). Match each step with a 'best unit' for the AI to build. There may be a simple mechanism for this, since when the player selects a tech from the chart, all the prerequisites are factored in and queued up for you.
Roghar
Warlord
To go with the terraforming posts, casting Sanctify on hell terrain needs to be incorporated, as well as perhaps strategy reflecting AC - determine whether or not the civ wants a high AC, and make some choices accordingly.
Building and protecting high xp troops is another area where players differentiate a lot from the AI. Putting high xp troops back in the attack order within a stack could help to protect their elite units.
Building up and protecting mages/archmages etc is a related issue, the AI is a lot better at using spells than it used to be, need to somehow get it some casters! No idea how best to do that, I'm sure it’s a difficult one
Building and protecting high xp troops is another area where players differentiate a lot from the AI. Putting high xp troops back in the attack order within a stack could help to protect their elite units.
Building up and protecting mages/archmages etc is a related issue, the AI is a lot better at using spells than it used to be, need to somehow get it some casters! No idea how best to do that, I'm sure it’s a difficult one
- Kuriotates does not settle their cities in good locations.
- The AI does not form their groups in a smart way. Combining 2 move units with 1 move units for example, makes the whole group move at 1 speed.
- AI need better tech priorities. Elves shouldn't be forced to get Fellowship of the Leaves first, just heaily weighted towards it. Instead they should adjust their strategy depending on terrain and situation.
I can't think of more at this moment cause I'm tired and I haven't played a game recently enough.
Ecofarm
Deity
These 2 heroes defend until dead with uninjured rangers in tile, on occasion:
Kithra (it's like he has guarding promotion)
Alazkan (perhaps because of strikes)
Also:
Moe, Larry and Curly
onedreamer
Dragon
yeah, the AI is indeed designed to do so. But in regular Civ there is a limit of 3 missionaries at once which keeps at bay the AI, in FFH there isn't such a limit.
onedreamer
Dragon
This is quite questionable. With such statement you are assuming that movement is the most important element in designing a stack, but I strongly agree with the AI that it's not.
No, but for example;
A stack containing 2 horsemen, 4 hunters, 5 axemen, moves towards your city. The horsemen can attack and kill units 2 tiles away and return safely to the stack. But they wont, cause they won't separate from the group.
A stack containing 2 horsemen, 4 hunters, 5 axemen, moves towards your city. The horsemen can attack and kill units 2 tiles away and return safely to the stack. But they wont, cause they won't separate from the group.
On raging barbs the A.I gets wiped out to often because they defend their cities with just one unit. In one game I got open borders with cassiel early, retook a cassiel cities from the barbs and gave it back to him. I stationed a warrior in it and saw that bevore he built a warrior, he built a settler. The A.I should be forced to allways have two units in every city, when raging barbs is enabled.
Marksman77
Aethernaut
AI won't attack from a city. In most cases when invaded, it will just sit in cities.
I had a Horseman end his move next to enemy city to test that; the AI didn't attack even though it had 5-6 units perfectly capable of killing my Horseman on spot and return to city in the same turn.
I had a Horseman end his move next to enemy city to test that; the AI didn't attack even though it had 5-6 units perfectly capable of killing my Horseman on spot and return to city in the same turn.
when an AI civ founds a religion first, make it a higher priority for them to go after the techs they need to get their religious hereos. I always chuckle to myself when I lose the race to Fol or Rok,etc. and yet I alwyas get Kithra and Yvain or Bambur and Arthendain, etc. before the AI.
Kuriotates cities need to be built further apart to take advantage of their 3rd ring
Lanun especially and other civs in general need to be using ships need to make more use of naval city defense bombardment
AI needs to get much better at upgrading its adepts into mages
Luchuirp need to be able to better take advantage of their ability to have fireball-weilding golems
need to see the calabim vampire unit more willing to feast on city populations
Kuriotates cities need to be built further apart to take advantage of their 3rd ring
Lanun especially and other civs in general need to be using ships need to make more use of naval city defense bombardment
AI needs to get much better at upgrading its adepts into mages
Luchuirp need to be able to better take advantage of their ability to have fireball-weilding golems
need to see the calabim vampire unit more willing to feast on city populations
Not just the Kuriotates: The AI in general builds cities too close together, which leads me to raze them and build them a tile to the left, for example.
So a national limit (say 10) on each type of religion-spreading unit should suffice.
And for Vitalise: If city x has tundra, desert (not floodplain, unless this doesn't remove it) or ice, cast Vitalise. If city y has not enough food and no eligable places to build a farm, cast Vitalise on a plain.
And can you please make it that forts are not auto-built (except where they are on a hill 2 tiles away from another fort at least, forested hill for elves, or when only forts can be built), and act like a city, but can't be built on an improvement?
Heh, I generally find the AI build the cities too far apart, except the Kuriotates that is. However, they do found them in bad spots most of the time.
Slvynn
Duke Vector fon Pixel
- Joined
- Apr 3, 2005
- Messages
- 2,073
another thing that connected to spell using, but domestic one, so its easy to control it:
i had game where sheaim started on huge island, covered with jungles. Well, they sucked. But seesm it easy to programm adepts to cast blaze if tile is jungle. Flaming Sheaim land - sound cool, its also make them kinda terrafiorming their land. Anywasy adepts are units they use alot. Very strange they not blazing all those jungles (at least in 0.31 they did not).
i had game where sheaim started on huge island, covered with jungles. Well, they sucked. But seesm it easy to programm adepts to cast blaze if tile is jungle. Flaming Sheaim land - sound cool, its also make them kinda terrafiorming their land. Anywasy adepts are units they use alot. Very strange they not blazing all those jungles (at least in 0.31 they did not).
As of .31 at least Sun II spell (blinding light?) stops entire stack. The non-frozen units will wait for the stunned units to recover. By casting the spell each turn you can indefinitely stop a large stack of troops.
I don't think i've ever seen the AI use a fireball. Not sure if this is due to lack of mages, fire promos, or targeting.
AI sacrificing units in hopeless attacks. One unit against a stack. Could the AI check to only attack if they outnumber or way overpower the enemy before attacking?
Been said but important:
Bad AI city sites. Too crowded (especially the Kuriotates) and often 1 tile away from river sites. This is mostly a selfish frustration.
Little or no AI terraforming, especially Spring.
I remember it being said the AI will check each tile it's on and cast if applicable. Could a Water I unit be set to do 'explore' with the limit 'only within cultural borders' for X turns?
Overly complicated alt idea: give the AI a unique unit that's identical to the adept but has a worker code and is only built if the civ has Water mana. Make them limited to a civ limit of 2 or set to national unit if that is easier.
The AI does seem to be getting better - thank you for working on this!
I don't think i've ever seen the AI use a fireball. Not sure if this is due to lack of mages, fire promos, or targeting.
AI sacrificing units in hopeless attacks. One unit against a stack. Could the AI check to only attack if they outnumber or way overpower the enemy before attacking?
Been said but important:
Bad AI city sites. Too crowded (especially the Kuriotates) and often 1 tile away from river sites. This is mostly a selfish frustration.
Little or no AI terraforming, especially Spring.
I remember it being said the AI will check each tile it's on and cast if applicable. Could a Water I unit be set to do 'explore' with the limit 'only within cultural borders' for X turns?
Overly complicated alt idea: give the AI a unique unit that's identical to the adept but has a worker code and is only built if the civ has Water mana. Make them limited to a civ limit of 2 or set to national unit if that is easier.
The AI does seem to be getting better - thank you for working on this!
1. This was mentioned by someone else: the AI needs a better balance of the early-game expand vs. defend mentality. The rapid expansion only leads to zero research and, in many cases, civs being wiped out when cities are poorly defended or even have no defenders at all. The expansion really takes the raging barbs feature out of the game as the map is quickly filled.
2. The raging barb option really doesn't work beyond early-game. One of the main reasons for this is the AI capturing and defending (often with many defenders) cities. By mid-game you have more of a barbarian empire spread around the map rather than anything resembling raging barbs. My suggestion would be to make the barbs raze every city rather than being allowed to capture. If you want barb cities, you get them from the Barbarian Lands option and barb cities spawn very prolifically already IMO.
3. The Aggressive AI option doesn't work in most games. While I have played a few games where it works, it seems like most of the time wars in the games are very rare, and, if declared short-lived. Another related problem is the way the AI goes about creating so many Defensive Pacts in the game, especially later. This could be resolved by allowing the pacts to be valid for a specified number of turns only and/or having an event which would cancel all Defensive Pacts and, perhaps, create a very interesting game.
4. Although I was lucky to build the Grand Menagerie in my first .32 games, would it be possible for the AI to stop its "Exterminate the Animals" (PETA is not happy
) campaigns? As many have said, animals are virtually wiped out by the time the regular barbs appear on the screen or you have a chance to get Hunting and Animal Husbandry. One reason for this is that the AI, with no intention of capturing animals, just goes out to slaughter the poor critters. Stop this mindless slaughter of those poor Polar Bears now, or I will call Al Gore. 
2. The raging barb option really doesn't work beyond early-game. One of the main reasons for this is the AI capturing and defending (often with many defenders) cities. By mid-game you have more of a barbarian empire spread around the map rather than anything resembling raging barbs. My suggestion would be to make the barbs raze every city rather than being allowed to capture. If you want barb cities, you get them from the Barbarian Lands option and barb cities spawn very prolifically already IMO.
3. The Aggressive AI option doesn't work in most games. While I have played a few games where it works, it seems like most of the time wars in the games are very rare, and, if declared short-lived. Another related problem is the way the AI goes about creating so many Defensive Pacts in the game, especially later. This could be resolved by allowing the pacts to be valid for a specified number of turns only and/or having an event which would cancel all Defensive Pacts and, perhaps, create a very interesting game.
4. Although I was lucky to build the Grand Menagerie in my first .32 games, would it be possible for the AI to stop its "Exterminate the Animals" (PETA is not happy
) campaigns? As many have said, animals are virtually wiped out by the time the regular barbs appear on the screen or you have a chance to get Hunting and Animal Husbandry. One reason for this is that the AI, with no intention of capturing animals, just goes out to slaughter the poor critters. Stop this mindless slaughter of those poor Polar Bears now, or I will call Al Gore. 
leo.
Monkey Monster
- Joined
- Aug 2, 2007
- Messages
- 85
I've seen Calabim a little bit weak in the beginning sometimes. One game, the first better unit it had after warriors were royal guards (even before they got hunters), they're obviously on their tech path.
The problem here is that not only Calabim get bronze working too late in many cases. It seems there are more AIs which dont have a priority for bronze working. Even after it teched its religion or more expensive techs, like code of laws or feudalism... that leads often to a very bad economy for AI, which results in a bad research, because it doesn't get a decent science up and stone-age warriors are still protecting their cities after a quite long time (and this still happens).
Another thing i've seen is that a couple different AIs are sometimes super-farming their cities (for no reason). Only farms in a city, that doesn't work, even if AI can run a good number of specialsts. And the AI doesn't run the proper civics for that.
Contrary to that i've recently seen AIs capital stayed at size 3 for over 150 turns quick speed (worked gold, cow, forest, all plains). I have no idea what went wrong there. Civ was Sheaim, and the capital built definitely no settlers and units all the time, and there were other improvements workable by only the cap and even free tiles, grass river, unimproved.
Combat:
Axe- and swordsmen are of course not the best defense unit for the AI. Archers are much tougher to take out. AIs using area damage spells (ritualists) or blinding light, entangle etc. aren't easy to beat, and healers (priests) in their cities give them a good defense ability, too. AI using catapults is good, but the problems with movement on offense was already mentioned above. Btw. AI knows how to use catapults on defense good enough i think.
Perhaps assasins fit better in an offensive stack for the AI than cats, but there is still the problem that there wouldn't be a bombardement for the AI. The good thing is that assasins are often effective against the human player. But building defenses for that is easy, and it looks silly to have 10 workers in a stack or build 10 hawks.
Something must be changed with assasins, only invert the order of attack alone isn't sufficient and doesn't make much sense imo, and guardsmen like it is now is not the solution. Probably assasins always attack/defend assasins? I don't know...
The problem here is that not only Calabim get bronze working too late in many cases. It seems there are more AIs which dont have a priority for bronze working. Even after it teched its religion or more expensive techs, like code of laws or feudalism... that leads often to a very bad economy for AI, which results in a bad research, because it doesn't get a decent science up and stone-age warriors are still protecting their cities after a quite long time (and this still happens).
Another thing i've seen is that a couple different AIs are sometimes super-farming their cities (for no reason). Only farms in a city, that doesn't work, even if AI can run a good number of specialsts. And the AI doesn't run the proper civics for that.
Contrary to that i've recently seen AIs capital stayed at size 3 for over 150 turns quick speed (worked gold, cow, forest, all plains). I have no idea what went wrong there. Civ was Sheaim, and the capital built definitely no settlers and units all the time, and there were other improvements workable by only the cap and even free tiles, grass river, unimproved.
Combat:
Axe- and swordsmen are of course not the best defense unit for the AI. Archers are much tougher to take out. AIs using area damage spells (ritualists) or blinding light, entangle etc. aren't easy to beat, and healers (priests) in their cities give them a good defense ability, too. AI using catapults is good, but the problems with movement on offense was already mentioned above. Btw. AI knows how to use catapults on defense good enough i think.
Perhaps assasins fit better in an offensive stack for the AI than cats, but there is still the problem that there wouldn't be a bombardement for the AI. The good thing is that assasins are often effective against the human player. But building defenses for that is easy, and it looks silly to have 10 workers in a stack or build 10 hawks.
Something must be changed with assasins, only invert the order of attack alone isn't sufficient and doesn't make much sense imo, and guardsmen like it is now is not the solution. Probably assasins always attack/defend assasins? I don't know...
![[NWO]_Valis](https://civfanatics-data.community.forum/avatars/m/110/110973.jpg?1475248361){kind=avatar}
[NWO]_Valis
A bad kitty!
* AI buiding tons of units that they cant afford and stopping them tech - some kind of balance system would be required that the AI does not spam units before it gets to a better version of them/in peace time/when has no gold for maintanance
Similar threads
