Hello, ye who read this

Humankind, more so than any other 4X with the possible exception of something from Age of Wonders, is extremely dependant on strategy and planning. Both macro planning and city managment, and micro decisions and unit movement. It will be much harder to hide the flaws of the AI systems used by 4X games so far by simply giving them buffs with increasing difficulty, or having them throw more units at you. I understand that the current state of affairs is not representative of the final game, but my experience with the AI in the scenarios leaves much to be desired. It isn't bad, at least in comparison to other games like the Civs, or Endless Legend, but roughly the same. Unfortunately, it can become a much more glaring problem here in particular. And we haven't seen how the system handles long-term progression and planning at all..

Having like four greatswordmen units one-by-one rush the same pike unit on a walled hill, lose half their HP, and then die to longbows, is understandable in previous games or like Civ or something, but surely, as Humankind innovates in every other field, so too must it progress on the AI front. Let's not compare it to other games, but surpass them!

Love and respect to the AI devs, pls give them more time/resources <3

Not to burst your bubble, but surely Amplitude knows this and their AI programmer(s) is(are) doing their best job on the game.  And/or they are trying to hire more people to the AI team as best as they can.

I kind of feel like really there are just limits to what is possible using current game tech around AI, and we will not see leaps and bounds of improvements over other 4X games. I would love to be wrong, but either way, I can assure you nobody probably cares more about this than Ampllitude. But yes, it's frustrating when the AI still seems insufficient to the task.

As far as game design goes, what is the correct choice? Do you make a simple game that the AI is very good at, or a more complex game that is more interesting to a player, but more difficult for the AI to be effective in? History tells me that it seems to be the "Impossible 4X Question". I am glad I'm not a game AI programmer. I feel they must have a small heart attack any time a new feature is proposed :D.

The problem as I have observed from Endless Legend, is that good AI takes time and resources. It is hard to convince your boss that these are being used to their best effect when they have more features or innovations to add to the game.

The ELCP (Endless Legend Community Patch) significantly changes the game AI for the better. That's not just an opinion it is factual that winning before and after the mod is applied are markedly different in difficulty.

So I was always wondering...why don't Amplitude just use the parts of the mod that are related to AI to make the base game better. And the unfortunate answer is that it takes dev resources and time to implement and test AI changes. Even if they are from a mod that you know works. And that's a cost that most companies are not willing to pay. Simply because it does not make you more money as a developer to have better AI past a certain point. What makes you more money is DLC and add-ons that you can sell.

I really hate that this is the case...but it is one of the constants of 4x developement that you cannot handwave away. The best we can hope for is that this game is given a significant AI programming team and plenty of time to get it as good as it can be for launch and then hope that the more glaring AI flaws are looked at over time.

Unless Nvidia, Intel, etc... create new hardware that allows learning AI the players won't see any significant changes for the AI part of the games. 

The way I see it there are three types of AI.

1#  An AI which creates rules in-game. Commonly known as different enemy types. By challenging the player via pre-establish rules. E.g. Flanking, flying, charging types of AI...

2# Actor AI. These are hard scripted AI types that act intelligently within the given script values. These are commonly used in strategy games. E.g. When reached X amount of gold build X, this can be varied by weight values to give a big of "thinking" feeling. 

3# Learning AI by try and error the AI knows what to do on its own, establishing the goals at first that is.

The problem with the actor AI that the game is ever-changing! Even the slightest balance patch could effect AI effectiveness. Not to mention the AI programmer doesn't have to be the best at the game to create a good meta to challenge thousands of different humans.

Since we are still dependent on the scripted AI the rules of the game should be designed around the AI. That doesn't mean the game becomes boring in fact the AI challenge is the fun of it. It's actually not the "pressing a button and doing something awesome" feeling. Not in the strategy genre at least. Unless its competitive strategy game. I believe the developers should drop the multiplayer part in 4X games. The player numbers are extremely low on every game of 4X. So rather than designing a game for 2 different styles.

When these two clashes with each other. Single-player: rules for AI and Multiplayer: rules for Humans. The game suffers. I wished the game designed around only automated the battles. But affect the battle outcome by reinforcements, positioning, unit composition, and appointed general. The rules could be much more simple but the overall fun would still be there. I mean Total War is quite boring. Almost every battle is the same. You try to flank, send archers, etc.. Now they manage to stop the decline of the series with the Warhammer because of the fantasy element. I don't know If anyone could remember but I loved the Spartan (2004) its mechanics were solid and the battles were fun to interact too. 

Every time the player tries the same thing. Get higher ground, flank, etc... as I said the real fun part isn't clicking and destroying things. It's the design, positioning, unit composition in 4X. Unless it's a party management game. But we get to create an empire, while still trying to micro soldiers like an officer.