The Direction of Age of Empires IV What im curious and concerned about

If it looks wrong, it probably is.

Considering corporate planning and execution measures by months or years, the budget and personnel may have been drained long ago.

Even at its heyday AOE4’s support was still notablely slower than other games in the series. I know some people will go into denial mode immediately. But that was plain fact. That was in sharp contrast with the huge marketing investments, the good reputation the game earns among wider audience, and all the business triumphs WE had been claiming.

Then considering Relic, their external contractor who probably have access to most of the product lines, had major financial problems in 23~24, and such problems should have been deeply rooted before they surface. It is likely the internal coordination was bad and the management at WE had limited long-term visions of how to sustain their game(s) production.

These are only speculations. But we’ve heard a lot of other failure stories and they all end up sounding familiar. Those are not failed games on their own but failed product management.

Well then I hope they make a properly supported antiquity or early modern game.

But they had just started hiring Unreal engineers. The production can take a long time, and the current decline of support feels too early.

I’ll answer you in a moment… follow along.

I would say that the AI ​​in AoE4 is broken in some parts.
We also noticed that in some patches the AI ​​fixed some bugs, but worsened some of its tactical behaviors. There were times in the past when it was quite aggressive and then they made it worse, making it slower.

But it seems they are having trouble fixing it definitively. Let’s see if, over the years, they reach the ideal balance.

Well, I and my family and friends and I have been playing AoE4 since its release, always in private PVE matches.
So we have a lot of experience in PVE in AoE4. Since we don’t play PVP.

Actually, you seem to have played very little AoE4 in PVE.
Many things you say don’t make sense.
Showing that you play AoE2 or other RTS games much more than you constantly play AoE4 in PVE.


Let’s break it down:

That’s not true. The AI ​​adjusts to the player’s behavior.

If you insist on producing a specific type of unit, the enemy AI will produce specific units to fight against you; however, if you create armies designed to fight against the AI, that same AI will alter the production of its units.

Example:
Yesterday I was facing the HOLY ROMAN EMPIRE, and they came with a huge rush of “Men-at-Arms,” ​​then when I started making large quantities of crossbowmen, it stopped making “men-at-arms” and started making light cavalry, archers, and Landsknecht.

Same thing if I produce a massive amount of “men-at-arms” and “spearmen,” the AI ​​will start making catapults.


There are some strange bugs in AoE2, depending on the map. Where the AI ​​can’t build anything or simply stands still.

I’ve seen this kind of extremely defensive and inert behavior in AoE4 in the previous patch (it was more common to see in the Allied AI), I don’t know if it’s been fixed now with this new patch.

In the last few patches, I’ve never had a problem with the enemy AI. It might hesitate to attack you because it knows your army is superior to it, but if you leave openings, it will move more confidently.


You are daydreaming.

And this is widely used in many other RTS games. Having variations, but always following a list of scripts.

All RTS are like this. It’s not yet an evolving AI that learns from the player.


Lie! Many people play AoE4 precisely because the AI ​​is fun.

Imagine what AoE2 would be like, which is much more technologically limited and quite repetitive as well.
My wife stopped playing AoE2 because she found AoE4 more replayable.
That’s a matter of taste.

Some people think that “Counter-Strike 2” is far superior to a Battlefield or a Call of Duty, regarding replayability. This already proves a lot about each person’s individual opinion.

—> NOTE: AoE4 is the second most played game in the franchise by the community. Most of them are casual players who play a lot of PVE. What you said makes no sense at all. If it were complete garbage, it wouldn’t be the second most played game by casuals. If it were complete garbage, it wouldn’t be the second most played AoE by casuals.


I didn’t like Red Alert 3, I thought the design was childish.
But I played many other Command & Conquer games, such as C&C3TW+KW and Red Alert 2+YR, and other older ones as well.


Now you’ve just proven that you’re not a PVE player in AoE4.
You played a few hours of PVE, didn’t enjoy it, and preferred to play other games, without paying attention to the details of the combat, perhaps focusing more on the economy and other details.

It’s obvious that the army is marching in an organized way, all together!

What’s chaotic is AoE2, where the army isn’t controlled, but rather each unit seems to be controlled separately, where there’s no organization.

In AoE4:

  • It can ambush. —> (I’ve seen this behavior several times. The AI ​​can leave an army standing still). Behind the trees, waiting for you to leave your base, only then will she emerge and attack you when you are unprotected. She can also place a hidden army inside camouflaged forests, so that they remain stationary or move silently, waiting for you to pass by to attack from behind or attack your unprotected villagers. I have witnessed this several times).
  • He can divide the army, so that a small group of cavalry can go around and attack the enemy trebuchets and cannons from the flanks/back. —> (In AoE4 this behavior has been seen many times, usually using 4 horses for this. You can see them being controlled separately, circling your army and going behind you to pick up the trebuchets).
  • He can divide the army so that a small group of lancers can accompany and defend his trebuchets and cannons. —> (I have seen this a few times, where the enemy army attacked and the trebuchets stayed behind, being accompanied by a group (of a lancer who stayed with them protecting them).
  • It can control the entire army, making it retreat and regroup to attempt a counter-attack —> (does it quite often, usually when its base is stressed).

Neither AoE2 nor AoE4 do that.

The only time I’ve seen a unit get stuck in AoE4 was because of surrendered enemy armies along the way, which can obstruct it. But that only happened in my army. I never saw the AI ​​get stuck (maybe I saw it in the first year of the game), or maybe you were very unlucky.

However, regarding defenses, I saw in AoE4 many times, in various situations, the AI ​​army retreating from better-defended areas and opting to attack less defended areas. However, if it has a cannon or trebuchet or battering rams, it will probably attack its defended areas.


I think now you want an AI that thinks like a person.
It’s better to wait for a new AI for AoE5, with “ChatGBT” and other AIs included, to then reach the level you’re looking for. Rsrsrs.

Well, the most official thing, in the opinion of the developers, was the “Exclusive Developer Interview with ZeroEmpires” that Corvinus broadcast a few days ago.

Overall, I’m very happy with the direction of the new variants. Many are now more “historically accurate,” possessing many unique units, buildings, and new bonuses, very different from the first four variants.

I was unsure if “Crucible” would be a good single-player mode, but as Beasty announced, it looks beautiful and very well-crafted, on par with a special campaign scenario, but with many, many modifications and the possibility of changing civilizations each time you play.


Regarding the new civilizations:


  • Sengoku, seems like a civilization that would benefit greatly in team games (protecting the yatai).

  • Golden Horde, instead, look more like a 1v1 civ because in teams, making a lot of ovoos (by stealing Stone Ores) or stealing relics to make more ovoos has bad synergy with Japanese, HRE, and OftD.

  • Tughlaq, seems like it will be good in the late game, but in the early game, the high cost of many of its bonuses will probably make them vulnerable to rushes.

  • Macedonian Dynasty, I really want to try the Arsenal and see if it’s as broken as I think it is or not.


About the patch of Season 12:


  • I’m glad they nerfed Templars’ control marion.

  • The mansion rework is nice. What I like most is that it’s based heavily on a suggestion made about them, since in theory “resources didn’t come from thin air” but from the overworked labor of the peasants. Now it represents that. Whether or not it needs a boost for the bonuses, I think it can be fixed with more balancing.

I think the mayor question was by AoE4, the self-doubt and Identity lost of the genre.

Too many games just didnt feel right, no matter how “good and honest” the intention of developers was, not like they delivered bad products on purpose, but too many products that deviated from the formula were simply really bad, in terms of feedback, sales and ratings.

That’s why AoE4 did make the bold move, just to be the game people wanted.
Now alter it, will simply feel like a betrayal.

What AoE4 needs is an AoE5 that does pick up the torch, does own thing, but doesn’t forget its origins.

The idea that opinions are undeniable fact is at times funny, and at other times disappointing. Is that what online discussion has come to in the 2020s?

1 Like

I think you wrote such a long reply just to prove that you’re right, but the problem is that in trying to prove yourself correct, you’ve mixed up a lot of things.
What you described sounds more like your own imagined version of the AI rather than what it actually is. In Age of Empires IV, the AI’s complexity and learning ability are nowhere near what Red Alert 3 achieved. The AI in AoE4 is based on static scripted logic — it has no learning or real-time adaptive reaction. It’s not self-adjusting; it just follows pre-set strategy scripts. It’s honestly not even as good as the AI in Company of Heroes 3.

So when you talk about the AI “changing units” or “adapting to counters,” that isn’t real dynamic learning.
And no matter how you phrase it, it doesn’t change the fact that the AI does not adjust its overall strategy based on map layout, resource distribution, or terrain conditions. Its positioning, expansion, and building placement all follow fixed patterns and predictable logic. Before boasting about how “smart” the AI is, I suggest you actually check with the developers — they could give you a more accurate answer.

If AoE4’s AI were truly as intelligent as you claim, and you and your friends have played so many PvE matches, yet you still can’t tell the difference between dynamic learning, real-time adaptation, and pre-scripted behavior, then you’ve basically been playing without observing. And please don’t confuse player-controlled AI behaviors with the actual non-player AI logic — the non-player AI in AoE4 doesn’t use stealth in forests, doesn’t hold chokepoints, and doesn’t defend river crossings.
That alone already proves your argument wrong — you simply can’t demonstrate its so-called “intelligence.”

As for saying the army formation in AoE4 is “more organized” than in AoE2 — seriously, is that even an argument?
I suggest you try to understand what AI design means before getting proud of such surface-level details. Even if you asked the devs, they wouldn’t tell you “better formation = smarter AI.” That logic makes no sense. Formation pathfinding and movement cohesion ≠ actual AI strategic intelligence.

If you want to talk about that, maybe explain why during large-scale battles, the AI’s target selection is so unbalanced — why do multiple units randomly focus on a single target instead of forming proper firing lines or pressure fronts, especially given that AoE4 was initially designed with formations similar to Total War?

And what does the game’s player count have to do with AI quality?
I genuinely don’t understand your logic there. It seems like you just wanted to mock me while dodging the actual issue.

You‘d better wait until the developers officially confirm that they’ve actually fixed the AI issues — and that the AI can properly take advantage of high-ground positions — before trying to convince me. Lol.
In fact, you’ve already lost the argument: the moment the developers themselves admit they’re working to fix and improve the AI, it already proves that the AI has problems.

=================================================

The real, proven AI problems I mentioned still stand:

  1. The villager allocation logic isn’t dynamic, leading to idle workers or resource overflow in late game.
  2. The pathfinding and micro behavior are inconsistent.
  3. You clearly don’t understand what “adaptive AI” means, nor the difference between scripted and intelligent systems.
    Maybe the things you’re describing are just general engine behaviors — not actual proof of smarter AI.

Of course, you can’t prove me wrong either, and you certainly can’t prove yourself right.
When I mentioned Red Alert 3, you dismissed it without trying — so you can’t compare or disprove my point that AoE4’s AI is inferior.
And as of now, the issues I’ve pointed out in AoE4’s AI remain unsolved.

Can the AI actually distinguish between defensive and offensive situations dynamically?
Can it use terrain, chokepoints, or rivers as part of its tactical setup?
You can’t explain those things away, because they simply don’t exist.

If you really want to make your argument sound reasonable, you should focus on evidence, not blind praise or irrelevant side topics.
All you’ve proven is that you haven’t observed the AI carefully at all.

I didn’t even compare AoE4’s AI with AoE2’s AI. What I said was simply that I play AoE2 more often because it offers more game modes and greater variety.
I honestly have no idea what kind of logic makes you bring up AoE2’s AI as if that’s the issue here.
If you really want to make a comparison, why not take a look at Age of Mythology: Retold? I’m not saying its AI is particularly smart, but at least it improves the AI while offering different behavioral styles — defensive, aggressive, and so on.

So if AoE4’s AI is supposedly “smart,” then why doesn’t it hide in tall grass? Why doesn’t it take advantage of high ground?
That’s the whole point — those are actions only player-controlled units can perform under direct commands.
So why can’t the enemy AI do the same? You can’t explain that, because it simply can’t do it.

=================================================

If you’ve never experienced the refinement of Red Alert 3’s AI, you’ll naturally assume that AoE4’s AI is impressive. Since you brought up AoE2 for comparison, let me make it clear what RA3’s AI is capable of.

RA3’s AI adjusts its behavior dynamically and in real time. Yes, it starts with a general direction, but every behavior—construction, offense, defense, harassment, tech progression—changes and learns throughout the match according to how the player acts. That’s what adaptive AI truly means. Meanwhile, AoE4’s AI is still stuck at the scripted preset stage. I suggest you learn a bit more before calling it smart—don’t be a frog at the bottom of the well.

Moreover, RA3’s AI is intelligent enough to bait and lure the player into ambushes for encirclement, to kite units, and to fight guerrilla-style, striking without overcommitting and minimizing losses. It can also select and combine different tactical templates dynamically—offensive, defensive, harassment, etc.—depending on the map, faction, and game situation, switching between them instantly at higher difficulties. On the hardest level, the AI can perform multi-front operations, attacking from several directions or targeting your logistics to force you to split your army.

Its goals are clear and purposeful—it knows exactly what to do and where the player’s weak points are. Its tactics are threatening: it prioritizes destroying key structures, disrupting mining operations, and locking down resource zones, following short-, mid-, and long-term objectives to completely choke the player’s development.

Even on the highest difficulty, its resource income and production speed are identical to the player’s—no cheating. What makes it strong is its intelligence: faster reaction time, more precise scouting and targeting, dynamic unit composition, and real-time adaptation based on the player’s choices.

The AI also interacts with the map extremely well, using chokepoints, rivers, and bridges to ambush or defend, and lastly—it remembers. It records the player’s tactical habits within a single match: if you keep using the same strategy, you’ll quickly find yourself dead.

The reason I’m explaining all this is because I can tell you’ve likely never played RA3, since you’re convinced AoE4’s AI is great. And your reasoning—comparing AoE4 to AoE2—makes little sense. You say AoE4 has more players? You might want to check the average and long-term player counts of both AoE2 and AoE4 first, otherwise you’re just making yourself look ridiculous.

=================================================

If you really want to prove your point, the best thing you can do is ask the developers directly — ask the official team how the AI was designed, what their original ideas were, why certain things weren’t implemented, what improvements they’re planning, and what their overall vision is.

You should also ask them what they think of Red Alert 3’s AI, and maybe even contact Relic to find out why their own engine ended up producing such a limited AI system. Then, ask the World’s Edge and Relic teams what kind of technical or logical issues they’ve encountered in the AI programming and game code.

Finally, ask them why, despite being built on the same engine, Age of Empires IV and Company of Heroes 3 have such a huge difference in AI quality.
And perhaps Relic will tell you something like: “Sorry, we’ve always worked with squad-based unit health systems, not individual-unit models.”
If they say that—well, that would indirectly confirm that AoE4’s AI does indeed have problems ,At the end of the day, it really comes down to just one sentence: if AoE4’s AI were truly that good, smart, and problem-free, why would the developers need to fix and improve it in future updates ? You should try to explain this issue more thoroughly.

I believe I’ve pointed out the issues with AoE4’s AI from the very beginning, which already indicates problems on a technical level. I think you didn’t really pay attention to what I wrote and just reacted based on your own assumptions without thinking it through , i never denied that AoE4’s AI has strengths; I was only pointing out its problems. That doesn’t mean it has no merits, but your understanding clearly has some issues.

You and your people just love to pick on the tiniest thing possible against whoever who was not in your fanboy club. Is that what online discussion has come to in the 2020s?

When did I ever say opinions are not deniable? I’m simply pointing out someone will definitely deny anything that does not go in WE’s favor, regardless of what it is, and that someone has just shown up to prove that. There is no judgement, but I think you already have your own idea of whether such an action is a genuine opinion or a simple hypocrisy.

I won’t even compare it to C&C, which only has one type of resource… or compare it to Company of Heroes, which is quite simplistic economically because it doesn’t have gathering units. These are games where you can work on the programming in very different ways.

Age of Empires is quite complex… it has huge bases, 4 types of basic resources, and worker units that gather resources and build and repair.

However, I understand the differences in AI movements, because I’ve played all 3 franchises extensively. So you don’t need to explain it to me.

Well, you’re going around in circles and there’s not much point in answering you… because EVERYTHING has already been answered.

Because you say that the AI ​​in AoE4 doesn’t adjust itself.
Even after giving you some previous examples.

Well, wow… the community would have already abandoned AoE4 if that were really true… considering AoE4 is the second most used game in the franchise by casual players… even playing in PVE.

I understand that there are problems with the AI, it most likely doesn’t work well on maps with very few resources and where the resources are very far away from it. But those are terrible maps even for a beginner player, imagine for an AI that is difficult to make like the one in Age of Empires. Maybe if it were a C&C with a single resource or Company of Heroes just to conquer points on the map, it would be easier.

But if you choose more balanced maps, the Age of Empires AI works well.

Overall, sorry
But I’m not going to waste my time answering you.
It doesn’t make sense.
As you yourself said, the Age of Empires AI follows a fixed pattern. It doesn’t change at all.
Well, for me it changes.
—> I understand that there are some tactical patterns that are strong and can be used consistently, but this also happens in C&C and Company of Heroes, where some tactics are repeated, but they can vary as you counter them.
But if for you it NEVER changes, no type of new unit can be altered on the battlefield, it’s always the same 100% of the time. Then the discussion ends here.

Thinking about it more…
You must be right!
I must have played and faced the same units 100% of the time.
I’ve never seen any kind of change. I must have been sleeping while awake, it must be some kind of delusion.
Speaking of which, I’m going to sleep, maybe that will help me see better in the morning.
Have a good day, thank you.

I’m not interested in the other games you mentioned.

I don’t want to know if the AI ​​in AoE4 is inferior or superior to another game.
That’s not the point of the discussion.

Because each game has a different AI, and some are better in some aspects and others worse in other aspects, that’s extremely obvious.

I don’t like C&C:Red Alert 3, I think it’s ugly and boring. I prefer C&C3:Tiberium Wars + Kane’s Wrath, which for me is much more fun and not so childish.
Note: I didn’t play the regular version of C&C:Red Alert 3, I played it with mods… And the mods had the same problem that all previous C&C games had… it was a problem that appears exactly at the end of the game, when resources are scarce. So I can’t consider the C&C AI perfect. At the end of this post I’ll explain what the problem is with the collector units.

You’re taking the conversation in a direction that doesn’t make sense.

Each game has its own AI… period. They’re not the same; they’re from different companies and developers.

Of course, the AI ​​in AoE4 has bugs… they’re constantly fixing them.

Probably, if you choose bad maps, where resources are scarce and too far apart to be collected, you run a higher risk of encountering errors.

But regarding idle villagers… there are patches that mention fixes… and if that’s the case, it’s probably a map where the AI ​​is bugged. This type of problem can even appear in other games, when the AI ​​can’t read the path on a particular map.
In the maps I usually play, I haven’t seen villagers standing still… In the maps I use, it takes 2+ hours for that to happen, for wood to become scarce, and then for them to be afraid to approach my territory to collect.
—> Even if you say that C&C:Red Alert 3 doesn’t have problems at the end of the game, I doubt it, because I’ve played MODS for C&C:Red Alert 3, and it also makes a lot of mistakes at the end of the game… it can become idle or, as far as I remember, send the collectors to the enemy base and lose them in the process, just because there were resources inside the enemy base (or enemy territory). I bet this happens in the standard C&C:Red Alert 3, since it happened in the C&C:Red Alert 3 Mods.

And obviously, AoE4 is much more complex than C&C.
C&C only has one type of resource. It’s much easier to balance gathering and production, and it has far fewer unit types.

—> But I’ll present official proof of AI behavior in the next post… since you doubt it. Then bear with me. So wait and see.

Since you don’t believe my PVE gameplay reports, I will prove with official information from the developers, themselves that it does many things you say it doesn’t. (See end of this post.)

An account of a situation I witnessed several times:
I’ve seen AI hiding in camouflaged forests many times. I’ve also seen it behind small forests… this one was funny because the AI ​​positioned its army right up against the trees and just stood there… and I did a test to see what it would do.
Generally, the AI ​​stands still for a moment, hidden/camouflaged, as if it were waiting for some safe route calculation or a higher probability of victory. Often this army comes with some siege unit, and in many of the times I managed to see them, they were smaller armies, nothing very large.
It waits for you to leave the base and then advances to attack the unprotected base, or some place of interest.
It really does have some kind of logic and programmed behavior to do this in some very specific situations.
I’m not going to film or take pictures to prove this to you.
If you had been playing for a lot of hours and chosen good maps, you would have already witnessed this type of situation.
Note: At the end of this post, I will prove that the developers themselves mention that AI has logic, and that it may or may not feel confident in certain situations. Honestly, it’s as if it’s calculating probabilities and choices that are worthwhile or not.


I said that it adapts to the player’s behavior.
I didn’t say that it learns from the player.
I also said that there are variations, but they always follow a list of scripts.
Dude, I don’t need to go into so many minute details about AI programming… I believe you must understand something.
Because if I have to explain it in more detail, I won’t finish today.
I may have used some incorrect expressions… but my native language isn’t English, and Google Translate is terrible.

Coming back…
Which means that, just like in “AoE4”, “C&C:RA”, and “Company of Heroes”, the AI ​​will adapt in real time to the player and will make changes to the production of some units, according to its script = pre-programmed programming + behavioral logic.
This is used in most RTS games. It’s nothing new.

The problem is that you insist on saying that in AoE4, it doesn’t have any kind of variable behavior, no matter how the player plays and and that the map does not influence their behavior. Which is an absurd lie!

Because she does have the ability to alter her behavior and make choices, as she is already programmed to have these options, as mentioned by the developers themselves and from what I witnessed while playing.

The fact that you insist on talking about C&C:RA3 and CoH3 shows that you played a lot of those two games in PVE, but played very little of AoE4’s PVE.

Basically:
The Age of Empires 4 AI (Artificial Intelligence) actively attempts to counter players’ units, especially on higher difficulty levels, by employing the game’s built-in “rock-paper-scissors” counter system.
There are also changes in behavior depending on the map type.


You asked for this (official information from the developers themselves).

Let’s look at the evidence!
If you still insist that it’s not true, then you should ask the developers yourself if these descriptions they provided are true or false.

Since you don’t believe my stories… The evidence is described in the official patches and updates themselves, which the developers shared with us.

—> In these patches/updates, you will see mentions about the AI: that it has a behavioral logic, that it uses formations in armies, that it behaves differently in certain situations, that it adaptive to the player in real time, that it monitors the player and makes decisions with the information it collects from the player, that it chooses routes, that it can devise strategies according to the information collected while exploring the map and according to the information collected from the player.

However, I repeat, it follows its database, which is in its script = pre-programmed programming + behavioral logic, in which it will be responsible for making decisions (which is in its data list), and it will change its behavior in real time according to the information it collects and participates in certain situations.


Let’s begin.

Note: I included the numbering of each patch/update to show that it is true. You can check it yourself later on the official website.

The evidence → Here is the official list (That’s a lot, I must have forgotten something, but this is enough to see that there is an elaborate program that makes the AI ​​change its behavior in real time, which confirms my stories):

7.0.5861:

  • AI resource exchange via market has been optimized, so that AI will only go for resource exchanges that make sense.
    —> AI players will now correctly begin scouting on Four Lakes and Continental maps.
  • Pathfinding improvements allow AI to handle walls more intelligently on campaign maps.

15.1.6970:
—> AI players will now take more conservative paths towards their enemies’ Wonders.

14.0.4963:

  • AI players will now have increased defense for their or their allies’ wonders.

13.1.4420

  • AI combat ships, villagers, and military units will now correctly continue to flee when faced with enemy Incendiary Ships, military units and converting monks.
    —> AI players will now avoid producing fishing boats if no deepwater fish deposits are present on the map.
    —> AI players will now avoid building fishing boats on maps that do not have fishing spots.
  • AI players will now correctly build naval military on lake maps.
  • AI players will now avoid building naval military when it isn’t necessary to do so.

13.0.4178:

  • AI military units will now flee from monks attempting to convert them.

10.0.576

  • Japanese AI now makes better use of Shinobi units.
  • Adjusted how the Easy and Intermediate AI chooses to advance to the next Age.
    —> Easy AI is now adaptive, speeding up or slowing down its Age-up based on the other players in the match.
  • Hard AI can now research all economy technologies.

11.0.782:

  • AI will now attempt to fortify contested islands on naval maps.

9.1.370:

  • AI now only uses a handful of units to guard its siege and the rest of the army continues to fight.

9.2.628

  • Units will no longer get stuck on generated wall blockers, this would previously happen most commonly when Villagers are chopping wood near wall ends.

6.0.878

  • The AI has had several improvements to how it handles its economy as well as its military and naval strategies.
  • AI now researches military upgrades more effectively.
    —> Developer’s Note: Military research used to heavily rely on knowing what units the enemy brings to a fight. Now it will get generally good upgrades based on it’s own army composition if it is unable to get enough information on the enemy forces.
  • AI will now always try to build the first lumberyard near a forest.
  • AI will prefer shore positions for building defensive fortification on naval maps.
  • For the AI, we increased the production priority of Transport ships when land military units are available.
    —> AI transports will now avoid unloading their units at dangerous areas on enemy islands.
  • AI Fire ship targeting selection has been improved, and AI fire ships will attack the enemy ships that will try to attack them at their dock.
  • Fixed a bug where AI attempted to build docks and fishing ships on hybrid maps with small ponds.

5.2.131
Hardest AI military production changed to start earlier than before to allow the AI to make attacks earlier.
----> A IA tem uma logic que trabalha para avaliar situações, exemplo na nota do patch “Patch 5.2.131”.
—> Adjusted the AI logic for producing anti-building siege units to improve the AI’s production of rams and trebuchets when faced with enemy walls and defenses.

  • Adjusted the AI logic for producing fishing boats to lower their priority when the AI has a stronger need for naval military ships.
    ----> Adjusted the AI logic when it has an allied Wonder on its team to prevent it from going overly defensive too quickly. The AI will now consider the time remaining on its team’s Wonder victory timer and its comparative military strength before deciding whether it should be defensive or to continue aggression.

5.1.148

  • Fixed an issue where the AI would still desire fishing boats when its naval military is at a large disadvantage.
    —> Developer Note: The AI will now deprioritize land units more often when more navy is needed. AI will also deprioritize navy more often when its island is being invaded by land units.
  • AI will abandon capturing a sacred site if its base is under attack and is lacking defense.
  • AI villagers will now retaliate against wolves.
  • AI will now focus on gathering more wood at the start of the game to increase naval gameplay, especially on island maps.

24916

  • AI’s Scouts are now more aware of threats, avoiding them when possible and running away when attacked. They should avoid running through areas of high enemy concentration and no longer trail a large number of sheep into the opponent’s base. Sorry, no more free sheep deliveries.
    —> AI’s Scouts will move more aggressively to locate the player earlier. On harder difficulties, this should make the AI able to execute early rushes and enhance their difficulty. (With Concealed and Explored map settings only)
  • AI with multiple Scouts now divvy up the world for scouting instead of crossing over into each other’s territories. (On Concealed and Explored map settings only)
  • AI will use fishing boats for scouting as minimally as possible, prioritizing fishing. This has helped increase the speed of the AI’s start on water maps.
    —> AI Naval Scout scouting patterns have been improved and Scouts will actively seek out the enemy islands.
    Threat Assessment
    ----> The AI has learned some new ways to interact with its environment:
    ----> AI will now detect some map choke points (including river crossings and bridges) and will build Outposts and Keeps in these areas to control them.
  • How the AI scouts and utilizes resources on a water map has also been improved:
  • AI will colonize other islands on water maps and will take advantage of available resources.
    ----> AI land Scouts on water maps will make better decisions about where to land and monitor other players.
  • AI will move extra Villagers to colonized islands on water maps, which will result in better gathering and economy.
  • AI Fishing Boats will now garrison docks if they are threatened.
  • AI will now send Fishing Boats to gather deep sea fish.
    ----> The AI has a better sense of threats that players might pose, and can respond better to them:
  • When facing multiple monks with relics, the AI now recognizes the extra monks as threats instead of only the first monk.
    ----> AI’s military units will always try to run away from a monk-unit performing conversion now (unless they are confident that they can kill the monk very quickly).
  • AI monk-units will not attempt to collect relics in dangerous areas and will flee if attacked while on its way to pick up a relic.
  • Improved AI combat logic to reduce instances where the AI’s ranged units could get distracted and attack harmless buildings instead of helping the rest of its army fight enemy military units.
  • Fixed an issue where the AI could get stuck and not Age Up while at maximum population because its military is too weak.
  • Fixed an issue where sometimes groups of AI units would get distracted from their main objective by a side combat. The AI should complete the side combat and then continue onto, if possible, to their original target.
    ----> AI’s retreating armies won’t use formation anymore, so that they can get away faster.
    Approaching the End Game
  • How the AI approaches and responds to the end game has been improved.
    —> AI now will prioritize finishing off enemies when it has advantage to do so.
  • Fixed an issue where the AI could save towards building a wonder even though it has a very strong military advantage.
  • Tuned the AI so that it is less picky about its army strength when the enemy religious victory timer is getting low. Before this could cause the AI to not attack enough even though the situation was desperate.
  • AI will now build more farms later in the game to support its larger armies.
    Other AI Changes
  • Fixed a bug where the AI was not researching the technology Horticulture.
    —> Fixed an issue where a group of the AI’s units could get separated after combat. The units should now maintain better group cohesion before and after combat.
  • Tuned AI market trading for resources so that the AI will trade when the needs for it are more important (such as Aging Up or being under immediate threat).
  • Fixed an issue where AI players were not attacking landmarks.

17718
—> Improved AI scouting and sheep gathering behavior.

  • Chinese AI have been tuned to prefer to place farms near Granaries and production buildings near the Imperial Academy to better take advantage of their influence bonus.
  • French AI have been tuned to prefer to place military production buildings near Keeps to benefit from their influence bonus.
  • Rus AI have been tuned to prefer to place siege workshops near the High Armory to better take advantage of its influence bonus.
  • Sultanate AI have been tuned to prefer to place research buildings near Mosques and the Tower of Victory near Barracks/Archery Ranges to better take advantage of their influence bonus.
  • AI Chinese Imperial Officials will now take safety by garrisoning buildings if threatened.
  • AI will consider gathering from resources with better gather rating.
  • AI can now use their demolition ships properly and can use them to destroy enemy ships and buildings.
    Hard/Expert AI early game economy tuned to prioritize resources and resource upgrades more efficiently.
  • Sultanate and Abbasid AI tuned to prioritize gathering from berries.
  • AI scouts now collect sheep in Revealed games.

14681

  • AI will use Khan as scout at the start of the game to find stone deposits faster and place its Ovoo on it

will focus on a few key points in response.

I have never denied AoE4’s AI strengths, so I want to emphasize it again. No one ever expected AoE4’s AI to be perfect or reach some extreme level of sophistication. What players really need is an AI that can interact with map features and terrain to achieve deep tactical gameplay. Simply pointing out that it works well on balanced maps doesn’t prove anything, because the AI still can’t do this. And if it performs poorly on unbalanced maps, that only shows the developers’ own limitations — the AI still has problems. As for AoE4’s complexity or comparisons to RA3 and CoH3, I never insisted on those; I only brought up RA3 because you mentioned AoE2’s AI.

===

Regarding AI hiding in forests, what is the probability and frequency of this behavior? Where are the trigger points, and what is the timing? Campaign or skirmish? If you say skirmish, then it indeed doesn’t happen reliably. There’s no need to claim that AI on higher difficulties adjusts production according to unit counters. At any difficulty, the AI’s main, routine behavior is quite shallow: harassing villagers and gold mines, early small raids → phased harassment, attacking traders, mid-game medium-sized flanking → two medium-sized attacks. Even when the player’s army is stronger, the AI doesn’t utilize terrain to defend river chokepoints or set up proper encampments, nor does it take other advanced tactical actions. When the player’s army is larger, the AI simply stops attacking and circles around its base. I’ve never seen it adjust adequately to counter a player’s army. If I field a full-strength heavy cavalry army, I can easily wipe out the AI forces. The AI doesn’t produce enough countermeasures dynamically; its commands remain primitive. These are exactly the shortcomings I was highlighting.

===

As I said before, the reason I mentioned RA3’s AI was because you brought up AoE2. You argued that comparing with other games is meaningless because each RTS has different AI systems. But CoH3 is point-based, with simplified, repetitive control, giving it more flexibility. RA3 doesn’t suffer from resource depletion because many maps have oil that can be repeatedly harvested, and the pace is faster, so resources don’t run out. Do I want AoE4 AI to reach RA3’s level? No, because RA3 is a fast-paced, competitive RTS. But within the RTS genre, comparisons are meaningful, because AoE4’s technical AI logic, flexibility, and single-unit control are even inferior to CoH3, which uses the same Relic engine. For instance, in AoE4, when siege engines are built, soldiers outside the center still circle around them — that’s a persistent issue.

===

Regarding AoE4’s technical problems, I emphasize intelligence because its AI cannot perform basic map interaction, cannot effectively use terrain to control advantageous positions, or adapt tactics. For a historical RTS with cold-weapon-era battles and tactical map design, shouldn’t dynamic tactical AI be fundamental? I know the AI is constantly being improved, but that only proves it currently has problems. Over the years, tactical-level interaction hasn’t improved. When CoH3 has a population cap of 100 and AoE4 a cap of 200, it also highlights that AoE4’s AI cannot manage larger populations effectively.

===

Regardless of whether AoE4’s AI can eventually be improved, if the developers cannot solve it, I hope they provide more AI options, like Age of Mythology: Retold did. AoM: Retold offers offensive, defensive, and passive types, each with subtypes, giving players more choices. Among AoE1 through AoE4 and AoM, AoE1 and AoE4 give the least choice. I personally enjoy AoE2 because its battles and destruction feel more immersive, but I had hoped AoE4 could surpass AoE2. Currently, I’d rather see AoE2 update its unit visuals than rely on AoE4, because AoE4 really offers very little. No matter what the studio’s intentions are, they have disappointed some players. As a player, I won’t spend every day criticizing the studio, but I have to consider player interests. I’ve bought the game and its DLC — that counts as support. Mistakes made by the studio must be acknowledged; it’s not the players’ fault. The studio should maintain its core design vision, but if technical issues remain unresolved over time, they cannot give players satisfactory solutions.

It’s described in the patch notes that she uses parts of the map to create tactics. Look for it and you’ll find it.

And she really does… yes… but it depends on the map… it depends on her economic situation… and obviously she doesn’t do it as well as a person would.

If you want something perfect… you’ll never find it in the AoE series games.

It’s a very complex game. With randomly generated maps, and that drastically increases the difficulty to be able to develop that.

—> If you want something that reminds you of another player… maybe only if one day they add these new AIs that are in production today, to be put into a game.

And if you don’t believe me… I can’t do anything about it.
I’ve done my best to show descriptions in the Official Patches, which mention that she uses logic and analyzes situations.
Maybe you don’t like AoE4, and you play it very little PVE.
And that’s okay.
Everyone plays what they want.


Skirmish

Maps I’ve played the most and where the scenarios described in the patches and updates generally occur.

Note: It was on some of these maps that I witnessed funny and unusual events with the AI. Even if you don’t believe it, they happened… I don’t know if there’s any map that increases the chance of certain types of behavior occurring more easily than others. (Generally tested on the Fourth difficulty, called “hardest”)

The maps are (Sorry, I don’t know if I translated the map names correctly, the game is installed in my language):

Maps with a good amount of resources:

  • High Forests
  • Danube River
  • Hideout
  • View from Above
  • Mountain Pass
  • The Pit
  • Volcanic Island

Easier maps to enclose with walls (good resources too):

  • Mountain Lakes
  • Rugged
  • Altai
  • Ancient Pinnacles
  • Confluence
  • Hill and Valley
  • Mongolian Highlands

Note: All these maps are balanced in resources, allowing the AI ​​to develop adequately.
Some maps have a higher chance of happening than others.


I made this suggestion a while ago…
However, I compared it to the game “Command & Conquer 3: Tiberium Wars + Kane’s Wrath”.

Where there were some very interesting AI personalities… such as:

  • Turtle (a slow but strong AI)
  • Rusher (attacks quickly and with systematic tactics)
  • Guerrilla (attacks quickly but with groups of lighter and cheaper units)
  • Compressor Roller (evolves everything and attacks with overwhelming force, using everything)
  • Defensive (could be an AI that would build walls and focus on attacking only sacred points and want to win the game with sacred points and wonders)

I think it’s time for us to let go of all expectations. From my perspective, this might be a chain reaction issue — if they never lift the 200 population cap, then more dynamic AI or more diverse AI types might never be possible. I believe there’s a connection here, and the root of that problem could lie in their engine code — or simply in the fact that the studio’s technical ability isn’t strong enough to fully handle the engine. Either way, Relic has disappointed me deeply. Company of Heroes 3 next door only allows 100 population total, and I honestly have no idea how they even dare to do that.

As for AI interaction with the map and with players, I’ve seen them constantly stitching random elements together and shoving them into Age of Empires IV. I can’t really judge — but if they truly knew what they were doing, they’d understand the underlying logic behind all these actions and how to properly integrate and optimize them. Maybe they’ve already lost control of AoE4 and its overall development direction, which is why they’re now frantically patching things together, throwing in whatever they can. The real question is — can they even stay true to what AoE4 was supposed to be?

From what I can see, AoE4 is currently a complete mess. Ever since 2021, they’ve been endlessly tinkering and “improving,” which might actually suggest that the game’s foundation and direction were empty from the start — there was never much depth to begin with. If that’s the case, then AoE4 may never truly inherit the Age of Empires legacy, because it’s fundamentally unstable — always clashing, always reworking itself. When will it ever end?

After wasting quite a bit of time talking… Rsrsrs.

We finally got to the point.

You said:

And that’s the point: “it’s a chain reaction problem!”. You’re right. Regarding IA, that’s definitely it.

I play (along with friends and family), we’ve been playing against the AI ​​since 2021 until now, and we’ve witnessed enormous changes in the AI’s behavior during those years. Some patches in particular broke the AI ​​to a point where they couldn’t implement certain types of behaviors back that were easily visible.


I’ll tell you some facts that contribute to it being a problem, a real chain reaction problem:

NOTE: you can search here on the forum and find old topics reporting some of these facts.

1st fact:

The AI ​​was very different in the early years than it is now. I don’t remember exactly if its best year was 2022 or 2023.

At that time there were some serious bugs, for example:

  • The AI ​​wouldn’t undid formation to escape or retreat. So the player could easily eliminate it in some situations, because it retreated slowly and sometimes didn’t fight back.

  • Generally, the AI ​​ignored the sacred point and didn’t bother to recover it.

  • However, in combat, when the AI ​​decided to attack you, it was quite strong. (especially on difficulty 4)

2nd fact:

I don’t remember which patch it was. But it must have occurred between 2022 or 2023. I think it must have been 2022.
The AI ​​gained the ability to split the army, usually a small group of horses, which circled your army and attacked your rear to destroy your siege units. This happened easily, it wasn’t difficult to see this behavior.

3rd fact:

In trying to fix the bugs from “1st fact”, they managed to break the AI.
The AI ​​on difficulty 4 became very easy.
Many players reported this on the forum and asked the developers to fix it.
The AI ​​on difficulty 4 was ridiculously weaker. And it no longer made large armies.
I don’t remember if it still divided the army here in this phase of stupidity. Rsrsrsrs

4th fact:

When they tried to fix it to make it like before… the AI ​​got more and more lost.
—> At this point the AI ​​was never the same again. They never managed to return to the same level of aggression they had initially (fact 1 and fact 2).
Players continued to complain that they wanted something more difficult. Because obviously even having “fixed” it, it wasn’t really like before. It had apparently been fixed, but it was noticeable that it wasn’t the same as before.

5th fact:

Responding to requests for a more difficult AI… difficulty levels 5, 6 and 7 were created. However, these new AIs cheat.

6th fact:

They broke the AI ​​again… regarding the walls…
Now the AI ​​no longer makes walls.

Those who played a lot before know that the AI ​​used to make walls. Now it doesn’t anymore.

Note: I don’t remember exactly when this event occurred. Perhaps it corresponds to fact 5. Maybe they stopped building the wall just when they released the other difficulty levels 5, 6, and 7.

7th fact

They kept updating and fixing bugs in the AI.

The gameplay is less buggy, but it’s lost that great sense of surprise… it seems the AI ​​plays better with certain civilizations… Some civilizations can survive and defeat a higher-level AI, for example: a level 4 AI can defeat a level 5 AI, depending on the civilization. I believe that in a 1v1, this shouldn’t happen… but I play a lot in the “battle royale” style (every man for himself), so I see many situations where an inferior AI beats a superior AI.

In short…
The AI ​​has changed a lot.

And many behaviors are harder to see…
Becoming rarer or practically nonexistent…

Probably yes.

The main reason is that with each patch, the AI ​​has changed some behaviors, but overall it’s losing some capabilities it previously had, while improving in other aspects. Overall, it’s losing its unpredictability factor, but it appears to have fewer bugs.


Suggestions for developers to adjust and thus bring back some behaviors that were quite fun and easily observed.

What I and other PvE players expect… is:

Developers please —> create custom AIs for players to choose from, where each AI will have its own personality and behavior.

Example:

  • Defensive AI (can build walls, and attacks and defends only sacred points and can build the wonder)
  • Guerrilla AI (attacks quickly and focuses mainly on cheap units, and uses any remaining gold to upgrade units and build sieges)
  • Rusher AI (attacks quickly, but produces varied units. Likes to divide the army into more attack groups for systemic attacks)
  • Tactics AI (AI specialized in dividing armies and attacking the same target, with the aim of encircling the enemy’s rear)
  • Conqueror AI (uses the most efficient means to evolve and have the most powerful army to destroy the enemy with a large army)
  • Turtle AI (looks very similar to the Defensive AI, because it also builds walls, but it can attack enemy bases, but it takes a long time to attack, focusing first on strengthening itself considerably).

If they do this… it will make a good improvement.