Additional AI issues:
In addition to AI unit design, pathing, citybuilding, settling, and risk assessment (on all of which the AI still desperately needs help!), the AI still needs work on
- Champion building: The AI will often, to its great disadvantage, choose the wrong strategy in equipping and fielding its champions. I've seen the AI (unsuccessfully) attempt to field Procipine as a warrior. I honestly do not know if the path Warrior or Defender was chosen, or if she remained a mage and simply failed to cast spells, but rather continued to wield her axe, while decked out in armor, in all of our encounters (in which she was soundly thrashed).
- Selecting when to accept peace. Warhungry AIs will attack when they deem fit, even if it means their own downfall, but in at least one situation I've had Tarth decide to kill me, then it succeeded in losing all but two of its cities in quick succession, not to mention most of its troops, while I did not lose a single unit. I wanted to offer Tarth peace, but found that it valued peace with me at around 800 Gildar (I owned about 200). If anything, Tarth should have begged me for peace, but would not accept it when offered "for free". And hence it was subsequently destroyed.
I can't resist saying it again. Gosh please considering looking into a method of allowing units to pass through / stack with non-hostile entities (including structures), some were suggested above. Please
These things should not rely completely on MODs, because the AI does not work optimally with non-vanilla data.
- Make strategic spells which hamper enemy production less expensive. Have you ever seen anyone, AI or Human, cast those spells (whose name sadly escape me and which I sadly cannot look up here on the website) with wintery-sounding names that slow production? I haven't. Perhaps because they are not "worth it"? Even spells such as Curgeon's Volcano, should that be the name, do not seem all that appealing, considering the fact that for less than 1200 mana I can teleport in to that city and sack it after a quick 100-mana-costing battle.
- End-game researched items and spells should be more powerful. I've never really seen anyone cast Blight, or the spell whose name I cannot remember but perhaps starts with Q.
- Experience rewards for ALL quests, lairs, and creatures needs to be seriously readjusted. Some rewards might be relics of earlier iterations of the Elemental franchise. Sometimes the difference in experience between comparable challenges is 10-fold. It's frustrating to have the same team (say, 2 champions and 5 troops) fight 10 comparable battles, gaining in each 10 exp for the champions, while fighting an 11th battle of comparable difficulty, to be awarded 150 each.
- The "value" of many items in the game need readjusting. Some might be relics of earlier iterations of the Elemental franchise. There are some items of comparable use (e.g. weapons which deal similar damage and have similar effects and similar initiative), only to have some be "worth" 10 times more .
Again, I am not suggesting that most of these changes be made primarily because I want an aesthetically balanced game, rather I am suggesting these changes be made so that the AI can make a much more reasonable judgement regarding what to build, cast, research, and equip.
It doesn't have to be anything groundbreaking, but there needs to be more spells which are not available through research; each node of the Elemental / Life / Death branch should have at least 3 choices, if not more.
Some obvious ones (locus classicus, surprisingly not already in):
- flaming arrows; tactical; mid-range fire tree: gives all bowcasters on your team +x fire attack for the rest of the battle
- mirror image; tactical; early-mid air tree: creates x mirror images of self, which disappear when hit
- Raise Zombie; tactical; mid-range death tree: create a zombie from a wounded unit on the battlefield
- Disease; tactical; late death: Champion suffers an Injury (Disease) which has a chance of causing a further random injury each turn to that unit and any other Champion in its army
- Flight; strategic; mid-late air tree: unit blessing, grants the flying ability
- Cure; strategic; late-game life: rids a champion of an injury or all injuries
- Frost; strategic; early-game water: unit bless, successful attacks have a chance to slow target
Again, unlocking cross-class spells, such as Mud (water and earth), Rust (water and air), Sleep (Air and Life), Boil (water and fire), Virus (Death and Air), Banish (Fire, Life), Angel (Life, Air, Earth), Bog Golem (Death, Earth, Water) or Sulphur Golem (Death, Earth, Water, Fire) would be nice to have, eventually.