The AI doesn't cheat as such that I'm aware of, although at higher difficulty levels it gets significant buffs. It performs reasonably well in general, I've seen a lot worse AI, especially at the time of release, although I am personally hoping that the developers spend some time improving the AI in future patches. If you've played a reasonable amount of strategy games then you won't want to play the game on any less than Challenging, and once you're used to the mechanics you'll want to go higher. I play on Expert and I'm thinking of increasing the difficulty; a significant number of people on these forums seem to play on Insane, although that sounds a bit like hard work to me. Personally I tend to concentrate on empire building and questing, only stopping to crush AIs that declare war on me or are in my way.

Are you sure?  How do you know this?

Really?  I thought in LH the player can raze a city instantly... or am I imagining that change?

I think you're being a bit unrealistic. It took decades for computers to learn how to beat the best chess players and this game is about a million times more complicated. Giving the AI some bonuses to compensate for its weaknesses is fine with me. The important thing is that the AI can provide the player with an appropriate challenge to keep the game interesting, not that it be forced to play exactly the same way a human would.

Wow, I couldn't disagree more. There are few things, I so categorically disagree with, but this is one of them.

Oh, my god. You completely miss the point. All points. If there are points to be missed, you go and methodically miss each and every one of them. And here is why:

Nobody likes playing a cheater!

Cheaters are hated, scorned and frowned upon, and for a good reason:

Cheating destroys the game!

Absolutely, and totally. Once a single occurrence of cheating is admitted, the game is gone, and utter chaos emerges instead.

Let me give you and example - chess endgames. Chess is still the deepest, richest and best turn based strategy game no computer game ever has a chance to rival. It offers both immense strategic and tactical depth. variety of positions which all play completely differently and require vast array of skills. Imagine a rook+pawn vs. rook ending. The theory is well researched, and most masters know it by heart, yet many positions are demanding and require accurate play to win. Hours of practice. Now I can either learn to play them... or I can cheat and plop a queen on the board. Now I am winning comfortably. No finesse needed. Did you just win a piece by a clever combination ensuring you a winning advantage? Tough luck, I am plopping a rook on b6, now I am winning. 

Playing with a handicap, on the other hand, is completely different matter, because it's a transparent process of giving an advantage to the weaker player, to get (as you inappropriately mentioned in association with cheating) more challenging and interesting game. For example, I will play against you by normal rules, but without my knight on g1. Or I will remove my pawn on a2, and give you two moves to play in the beginning. Or anything else. Anything clearly defined before the game has started.

Because with game where one side cheats, there is no accomplishment, no sense of struggle or victory. Did you just win a game and feel: "Wow, I maxed my cities nice, and outplayed the opponent in the battle, what a sweet game"? Then you are delusional, you are lying to yourself; you did not win because you bested the computer ability (you are not fighting that, remember), you owe your victory to the computer's (arbitrary) decision not to cheat at the moment. Because, if it wanted, the AI could have cheated in a dragon. Or two. Or ten. Or it could receive 15 techs from another AI for free. Or it could win and unwinnable quest. Or whatever. 

Cheating vs. Handicap.

And don't tell me "it can't be done because current technology...etc. etc." It can be done, if the game is designed around the AI, not vice versa. The example of that is a game called "AI Wars". The AI has infinite resources, and allocates them according to transparent rules (AI progress), and everything is cleverly and well explained in the game logic - humanity is just one nuisance the ultra-powerful AI has to deal with, and you have to accomplish your objectives before the AI realizes you are a threat and sends so many ships you have no chance of winning. No cheating necessary. At all.

But the game clearly states that when playing "Challenging" difficulty, the AI uses its best algorithms but does not enjoy any special bonuses, but it's a plain lie because it cheats.

And here I stand, waving with my Bill of Gamer's Rights and I shout: "Please, don't lie to me, and don't cheat when playing with me. I want fair victories and I want my sense of accomplishment, I want to know that I have beaten an opponent in the game played by the rules we both know, not that I have been let to win in some obscure cheatfest. I don't mind handicaps at all, but I abhor cheating, for it is a murky pool of distrust, spreading foul miasma of suspicion and uncertainty."

Thank you for reading it all, and when replying, thank you for attacking not my person from the position of fanboism, but instead my arguments from the standpoint of proper logic and common sense. Thank you. 

The thing is bonuses are not the best way to do this then. The best way to do this would be to forget the AI economy, spawn units in cities at a rate and quality that is consistent with the difficulty level. When the AI declares war, have it send its offensive stacks toward the opponent's cities. Level up the heroes and troops according to a predefined rate scaled to challenge. Give the heroes the tactical spells that make the game most interesting, and let them use them at a rate consistent with challenge. Give it the strategic spells automatically, and let them use them at a starting time and a rate also scaled to challenge.  

I have nothing against this approach. I am sure most strategy games do something like this, and I am sure this approach would have saved Frogboy many hundreds of hours over his career. It just doesn't make "AI" very interesting.

Fair AI is just a super-duper hard way for Frogboy to play everyone himself blindfolded to the board and with all his moves predetermined when the game is installed. Just like making a chess engine, if chess was so complex you could never use the minimax algorithm (or read that ... so complex it could never see more than its own move ahead).

*Sometimes my IPad cheats me, it happened here and I hit the submit button while I was still writing this. I was probably going to erase this and never submit it, because I wasn't going anywhere with it, but what the heck, too late now and maybe I will finish my point someday, when I find out what it was.*

It is interesting you use chess in this way. There are a couple of ways to think of games, how rich they are to human experience, and how vulnerable are they to either brute force calculation or mathematical proof that one strategy is best. Checkers is a rich game to human experience, but by bridging opening and endgame databases, the game is completely solved because of brute force calculation. Chess is very rich to human experience, but an opening database, surprisingly simple algorithms, and an endgame database are better than all human players. Chess is not solved, but someday it will be.

The thing about these commercial strategy games is that they seem more shallow than chess, because they both push the limit of the human mind, but in chess we can find opponents that push limits as hard back at us. Chess has other advantages too, the game plays very fast, 20-40 moves are decisive in most cases, and a single move can provoke one side to resignation. But the same solutions that work on chess and checkers are impossible to employ on 4X games like these, because a single move can contain dozens of moves, and if go positions are like snowflakes, then Legendary Heroes positions are like fingerprints on atoms. 

So then there is the word complexity. Chess is about as complex a game as a human can handle, the game could be more complex, but it wouldn't make a difference. A 400 lb barbell and a 400 ton barbell are equally heavy to me, I can't life either. I can struggle the same amount against them though, and that is much the same way with games that can be brute forced and those that can't. Except when it comes to designing an algorithmic solution to them. Then that extra weight makes a huge difference. A crowbar can move the 400 lb barbell, but a whole different concept is going to be needed for the 400 ton barbell. 

Regarding the AI, I've not noticed that the AI does do things differently than the player. It is not cheating in the same sense as people here seem to think, but it is playing a different and supposedly equivalent game. Their rules are different from the rules that you follow and for good reason. There is also several AI that to which people refer.

For example, the Monster AI, the game they play is not to dominate the world and kill everything around it, it is more of survive and roam around randomly, chase when their is food, don't attack if they think it is too dangerous for them. Obviously this AI is playing a seperate game and a lot of people have fun making this AI lose at their game, survival.

Then you have the AI players. At challenging or below these players do not get any economic bonuses to their gameplay, nor do they get research bonuses, vision bonuses, or diplomatic bonuses. (There are bugs with the AI, and they are addressing these bugs as they arise and become reported, with reports that give appropriate nice concise feedback). There have been many reports of the AI cheating, only to find that the save games that were posted were players playing the game on Insane difficulty, or expert. The AI is supposed to get these bonuses at these levels and it is told to the player.

The AI player does not do the quests like the player, but also for the informed the AI cannot do ALL quests either. There is a tag for those who make quests to determine if the AI player may go on that quest, and if the AI cannot go on that quest, then the AI player can never do that quest. Thus, this is the compensation between the AI rolling dice to do a quest (which I don't know how an AI is to choose correct answer to a riddle without a random choice selection).

I play the game for fun and does the AI detract from the fun no. Does the AI provide a challenge YES. Does the AI play exactly the same rules as the player NO, but do they play by an equivalent set of rules as to the player, in my humble opinion, yes.

Frogboy did a GREAT job with the AI and it is ALWAYS getting better, especially if people show him new tricks to have the AI do.

Thanks everyone for info.

I saw lot's of info about cheating AI and problems mentioned earlier, but I don't remember any of the stardock staff replaing or telling us if they are planning doing anything with this - that's weird, I thought that stardoc comunnicated with players more, especially with problems reported so many times.

And yes, I know that for many it's not a problem, for me it's game breaking, like rubber banding in racing games or autoleveling monsters in skyrim.

I already have FE and I'm on edge of buing LH