Humankind difficulty, Green/Huns declared a surprise war against me. I had the war score advantage, as I had several grievances and he had none. He sent a single scout to attack me. I sent 2 scouts back, and sieged his city, despite the levies having significantly higher combined CS than my scouts. He repeatedly sortied me, and once my scouts were at low health I stopped attacking his units. My scouts had high ground, adjacent unit bonus, and defending bonus at that point, and despite the fact he could've easily killed them of anyway, that seemed to be enough to deter him from attacking. The flag spawned on low ground right next to my scouts, so I just swooped in at the last turn to retake it, that is, if he even bothered taking it in the first place. After several sorties he ran out of war score, and I force vassalized him. 

I finally finished a game on Humankind difficulty.
The final score after 200 turns is:

- 1st the Spiffing Brit (black) with 11547 (he could only overtake me because of the Contemporary, otherwise he had 7990 points)
- 2nd me with 8611
- 3rd Lewis with 5819 (brown)

Early game the AI was overwhelming because of their massive bonis, but I managed to survive barely.
My overall gameplan was defensive/tall/ late game.
Babylon - Carthaginans - Khmer - Joseon - Italian

The biggest weakness of the AI is strategical decisions in war. It just didn't know what to do with it's armies and this lead to me winning against the Mongols even though he was vastly superior in forced.
I could always choose my battles and dictate how the war goes which made winning possible, even with way weaker military might.

The AI was keeping up well with research, until Turn 120 when I began to overtake them. (Joseon :) )

On humankind difficulty: In early game the ai seems to advance very quickly through the eras, ignoring to gather more fame by staying in the eras for longer. However as the game progress the ai seems to fall behind, like I'm last to ancient era but first to reach contemporary era, also I have more fame than the ai.

What is the AI doing and what do you think it should do?

I blocked the green guy in the south west corner by taking early the two territory with a natural wonder. He started to be angry and hateful but as i built an army to protect and lock the border it seems that he understood the only way to go out was not the war. He started to ask for trades and become pleasant and reverential although i refused all these offers. Once i was sure that he was not a danger anymore i accepted these offers and it quickly asked for an alliance. I accepted and he sent his troops conquering far away lands. Very clever, i never saw such a behavior in Civ for example.

On the other hand the yellow guy who had a lot of rooms to expand attacked me with nothing but an outdated pile thus firing a bullet in his foot...    

I am so happy how the AI evolved compared to the last open dev. Last open dev, it was somewhat easy to win the game on humankind difficulty. This time I played 2 playthroughs on Empire difficulty. For me it was an appropriate challenge on Empire difficulty. I managed to win in both playthroughs, but I carefully had to plan my turns to keep up with the AI. So, I was really happy with this difficulty and there are 2 higher difficulties that I can try, once the game is released.

Does the AI play in a way that feels consistent with their Archetypes and Biases seen in the diplomacy screen?

• I must admit that I didn`t spend much attention on that topic. The only big concern I have, is that the AI didn`t focus on settling the “new world” enough. Some AI players are settling very aggressive in the beginning, but after researching settlers, I had the new world completely on my own. There should be some AI Archetypes focusing on settling the new world as a priority. 

How did the AI evolve throughout the game? Did it feel equally challenging at the start and end of your game? Does increasing the difficulty increase the challenge as much as you expect?

• I think the AI is more challenging at the start of the game. Manly because it can aggressively forward settle you at the beginning and an early war can be quite challenging. As mentioned above, I think the AI must focus more on settling the new world again in mid game. In addition to that, I think the AI is transcending in a new era too quickly. For me it always felt that it is a good idea to stay in the current era a little bit longer to earn some close 2 or 3 star fame points within a few turns. So, I think the AI should focus on earning some more close era stars instead of transcending in a new era immediately.

Does the AI commit any grave strategic missteps? What is the AI doing and what do you think it should do? (Please provide saves if you can.)

• It felt that the AI was only building one unit type (normally it`s unique unit). I think it would be important, that the AI builds up a more diverse army, with more unit types like archers, melee, and cavalry. As already mentioned above, I think the AI could be even better in earning easy fame and settling the new world (which is producing a lot of fame with new territory, population, and districts).

How does the AI perform in battle? Does it make any obvious mistakes it should avoid? What would you have done in that situation? (Please provide a save if you can)

• I think AI is performing quite good in battle, I didn`t see major strategic missteps, as it was using the terrain quite good, I think.

I am playing current company on NATION difficulty.

Here's the situation: I have Vlad as ally of high level (all treaties are signed), we both have an adequate amount of troops. So does Isabella. She has constant pressure on her from us and Yanga all around her from Turn 30-40.

Turn 59: Vlad is in footsteps on Isabella's capital (it's her only city), he have my troops as reinforcement: power balance in 73 to 64.

Turn 65: My troops are frozen for 6 turns because they stuck not my battle and I can do nothing to retrieve them or something. Besieging power balance is now 103 to 57. 

Turn 66: Vlad surrenders to Isabella, becomes her vassal and joins her in her battles.

It was like: "Hello, I want to destroy for the half of the game and I siege your capital for 10% of the game. I could raise it to the rubble, but I have a better solution: I surrender you, and you take full control of me"

There was no threat to Vlad at all, he was severely threatened Isabella and still he surrendered to her. It is damn weird issue.

I couldn't move my troops for ages just because allied AI sieged the city nearby.

I agree that the AI seems to have early bonuses that keeps it a challenge. I also think after few games that those bonuses are flat, that is why AI fall out in later turns.

AI in the battles seems very balanced at first actually. Enough to give you challenge but not perfect, so you can find edge with being clever. The only problem seems later that the Ai is only good at aggressive plays. In my games every unit AI had wanted to do something. If it could fight, it would even if that was suicidal. AI didn't understood defensive play. It understood playing around terrain, and blocking my access to attack it, but didn't understood how to change position without attacking just to get a better one, that would win it the battle later. And because it seek to deal as much damage possible it never blocked any spawn/reinforcement point at all even when it easily could, and that would win it the battle. In the end AI in battles is tactically very good, but strategically poor.

And outside of battles it also seems the same way with some certain exceptions (what Fojibo mentioned about green before happened to me too, and was clever. It also happened with other AI too at some points) under certain conditions. Most of the time AI felt opportunistic. It made very good tactical decisions in certain situations, that was bringing them an edge somewhere, but strategically in the long term they felt like they didn't have a game/win plan. That makes the AI very good if it's stronger, because it will seek a short term opportunity again and again, but if it's not, it doesn't know what to do outside of few again short-term exceptions (like again in example Fojibo here mentioned, green wanted new land, couldn't get there by fighting, so was working towards open borders/alliance so it could go past the point that was blocking it, and get that land.Witch is really nice. But after that it didn't know what to do next to actually profit from that and turn that into fame, resources, advantage). Also it seems that AI patterns are mostly driven by territory. It very often just wants to default to state "get more territory", and in later eras/turns it can basically turn suicidal just because it cannot satisfy that need, and cannot find anything else to work on. Like people mentioned, it also doesn't seem to know how to prioritize research or build army compositions. Most of the time it will built the strongest/fastest to build unit it can (with most of the time means it's culture unit) and will never adapt it's army composition or even make it varied. The only variety it can get in army is with leftover units.

Was playing few sessions on empire difficulty.

Also, I don't think the AI had much choice here. The game feels very warlike. Influence and religion generate grievances, if you want to use that -You need to prove it with army. So go to war over them. There isn't much diplomacy here to engage with. 4 basic treaties that don't mean much, because there is no cost accompanied with enacting or breaking them. IF yo have military strength you can do whatever you want and there is no backlash. The only way to alleviate warlike behaviour are the trade routes, and mostly distant ones, but they are insecure and there is no way to protect them. Alliance doesn't mean much, course there is no shared victory or diplomacy victory.

To improve the AI strategically we need to give players and the AI more options to play. Right now it's all about keeping up with military strength to protect oneself or expand, and if you have that secured you can do whatever you want, you will get fame with anything. And if you need to invest in military anyway, why not use it. So warlike patterns are prioritized. And AI follows that.

Also about AI and diplomacy - vassalage seems way too easy to obtain. In my game Yellow had like 80% of continent taken. I went to war with him because I was superior in technology and just had few better units that decimated his low technology army (this play pattern AI doesn't understand as well). After winning the war I could take from war reparations like 3 of it's territories (not even cities), that wasn't even 10% of what it had, or for the same amount just make him my vassal. I was like what? I had a small country and was out of nowhere a liege to a powerhouse 3 times bigger than me. Also it never wanted to gain independence from me, but was constantly harrasing and fighting green on neutral waters that I had an Alliance with (that I wanted to keep). And I as a liege couldn't do anything about that. He could do whatever he wanted, it generated grievances for green, broke my Alliance, and the only way I could do anything was to declare war on yellow that was my vassal (either by surprise war or by making him independent and after few turns of war support just declaring normal war on them). I couldn't tell yellow anything, or do anything else to stop it from behaving like that, the game just doesn't have mechanics for that.

So AI feels poor and doesn't know what to do when behind, but also doesn't really have too many ways to find better solutions than being warlike.

I felt the AI had way more production queues than me or way more industry and/or food.  I cannot fathom have the population the AI must have had in order to field armies of the size I was seeing from them.

The-Cat-o-Nine-Tales wrote:

• How did the AI evolve throughout the game? Did it feel equally challenging at the start and end of your game? Does increasing the difficulty increase the challenge as much as you expect?

This was the science chart of one of my games on Civilization difficulty:

As you can see the AI is competitive until about turn 80. Then it just can't compete anymore. Note that two times my science was increasing between turns by more than the total amount of science that the best AI could have per turn. This was thanks to Joseon's coastal science, not due to science mode, I never used that in this game.

But even without Joseon I still snowball with science and all other resources while the AI does not. This is how the science chart looks on Civilization difficulty without choosing any science culture:

Industry and Influence looks similar:

With fame I am ahead of the AI for basically the entire game:

Here is the corresponding save file this this game: Italians Turn 167.ctr

The main take-away from these charts is that a skilled player is able to grow resources exponentially, while the AI scales its economy only linearly. When looking at what the AI built across my games, it looks like the AI does not sufficiently prioritize food and industry districts. Those two districts enable rapid growth since industry and population are the key to a stable and scaling economy. Once these districts are in place, campuses are important to research the buildings that make effective use of the districts that are in place. As you can see in the save file, Aminata went for food districts, but had has basically no industry:

The yellow player did the opposite:

A mix of both is important for a successful economy that scales into the late game. It is also important to finish those before constructing a lot of campuses or market quarters. I know the AI has huge economic bonuses on higher difficulties. But it seems like it is not making good use of that, because in practice their economy is much weaker.

The AI should prioritize industry districts first, then food districts, and then whatever their agenda is.

Also the AI needs to be better at exploring the world and starting new colonies on different continents. As of now the AI mostly leaves the empty space to the player, which means that this is an easy win. If the player does not have to compete for new land with the AI, an essential part of human history is also lost. So that is essential for the fun of the game. On higher difficulties I would expect the AI to colonize different continents shortly before the human player can so that the AI will be competitive in the long run.

Battle/siege related AI :
-The AI palying the huns keeps moving around, making the battle very slow (1 mvt + 1 mvt+ 1 attaq + 1mvt + 1mvt .... ) and very annoying
- Auto resolving can enable to win siege we cannot win (with close combat cavalry). 
- AI start some siege with only cavalry, so may be unwinnable because of same wall 
- AI don"t shoot at wall with siege weapon if there is no unit on it (very useful for defence but a bit strange). A normal player would shoot it to make is cavalry roam the city.
- Maybe the AI tend to "destroy them all" instead of playing a defensive play around the flag, Seen some siege where they sacrifice the militian where they could have win just waiting beside the walls.

Difficulty : 

- AI higher than empire difficulty cheat a lot, and it is very frustrating/not very fun to play. Empire do "cheat" a little but it is manageable and fun.

Other:
- We don't see a lot of AI on the water (one indepedant people of the "new world" in my playthrough, + 1 factions)
- I did enjoy diplomacy AI.
- Didn"t check carefully city construction by the AI but seems ok
- AI territory claim aim at very close terrritory. Doesn't seem strange but it could be fun to see AI with other  strategy (exploring seas for islands, exploring and claiming distant territory,...)
- AI beat me in religion. But I could have keep my religion with some more effort on the start of the game.

- AI seems a bit ahead, sometime on science (For units mostly). Graohs at the end didn't major differences.

It's due to the era unlock based not like civ game with bottleneck in each era. But based on X tech (most people literally beeline). Like being in industrial era while having galley tech. Or medieval without the early swordsman.

It’s difficult for me to comment on the AI, as my only interaction with the other empires was mostly as an aggressor in war. But I’m pleased that the AI will vassalize and kill each other off and not simply twiddle their thumbs. I even had the Mexicans attempt to attack me near the end of the game, showing that they were at least aware that I was about to win and wanted to do something about it!

Good evening :). I'm Kevin. Thanks for develop this.
I play on a Omen 15-dh1076ng. Intel Core i7-10750H. 32GB ram. GeForce RTX 2070. It is a mid gaming laptop.

-Early ships explore "automatically" to ocean tiles and, of course, they get destroyed. I think that they should "know" that this is not safe.

Thanks :)

Playing on Empire. Finished a close 3rd in my first game and finished a very distant 1st in my next game. 

• Does the AI play in a way that feels consistent with their Archetypes and Biases seen in the diplomacy screen? I suppose they do. It's not something I really noticed that much. Maybe it would be better for the AI to be slightly more exaggerated in these traits?
• How did the AI evolve throughout the game? Did it feel equally challenging at the start and end of your game? Does increasing the difficulty increase the challenge as much as you expect? It took me one game to figure out the AI on Empire difficulty. They are more difficult than they were in Victor, but still don't play intelligently. They are just receiving unfair boosts. I don't really like that kind of game design and was hoping for something better. If this is the way you are going to approach higher level AI, then I would take some time and figure out some subtle ways to give them a lift.  
• Does the AI commit any grave strategic missteps? What is the AI doing and what do you think it should do? (Please provide saves if you can.) I took over a few AI cities and they were spamming market districts. Now, that *could* be a viable strategy, but market districts kind of stink currently. I think they need to give better explanations for the diplomatic decisions they make, because I don't understand why they do things. As long as the AI is programmed to try and win, then should decent decisions. A good way to balance them is to watch human players attempt a balanced playthrough of the game. Adjust from there based on the AI's current culture affinity and traits. 
• How does the AI perform in battle? Does it make any obvious mistakes it should avoid? What would you have done in that situation? (Please provide a save if you can) I think they need to value the lives of their units more. Rushing into battles they have no chance of winning right now. Take defensive positions and guard your flag better. They fight surprisingly well, but make odd decisions like only having their army be made up of their unique unit or sending one units against an army of 20. The combat system in this game is really fun and I think it has a lot of potential in the future too. With the AI having better strategy (staying in formation and utilizing rear attacks) and adding in more than 2 combatants, combat will be incredible. 

One more thing that annoys me - AI does everything on the map at the same, even before the turn fully loads for a player. Endless Legend had like moments of waiting between AI actions - this game need something similar if you want to keep it real-time like it is.