Does the 'AI' have a clue what to research? (It is a rhetoric question.)

[zup]

What's wrong with this game. Turn 521, epic speed, I checked on Capria's progress in the field of technology. It turns out I could have sold her ancient chants, not that she could give me anything for it. I find it odd that on prince the AI does not get education by turn 500.

Now the only explanation I can think of is that because of her 'impressive' conquest based expansion, her empire spanned half the world. So her city maintenance has to be horrible. Sure she had fielded some axemen, but that hardly scares Tasunke's iron wielding knights. Probably she could not afford to spend commerce on research and without education, could not get more commerce.

And with her world shaking score of 1100 and something she ranked the second in the world. No other AI built cottages either. I did see some citadels.

The map was small great plains with 2 extra civs. There was lots of conflict early on, if that matters. I was constantly at war from the moment I had horsemen. I never attacked the Bannor though.

Education should be significantly higher in priority for the AI.

 

[vivictius]

I see that quite a bit, the Amurites, on price or monarch rarely seem to study magic while the Luchurip get construction but would rather build warriors instead of golems.

 

[Soam]

Pretty easy fix, move up in difficulty.

 

[Sarisin]

That doesn't always work.

I am playing at Monarch. It is an Epic game around turn 500. Varn Gosam is at war with the Hippus.

I opened the World Builder. and in the 6 Malakhim cities they are building, tada, Lightbringers!

I don't have to tell you how the war is going for Varn Gosam.

 

[FireBlaze]

That doesn't always work.

I am playing at Monarch. It is an Epic game around turn 500. Varn Gosam is at war with the Hippus.

I opened the World Builder. and in the 6 Malakhim cities they are building, tada, Lightbringers!

I don't have to tell you how the war is going for Varn Gosam.

I actually had (when playing mercurian) a lightbringer take out 3 lvl 5 angels. They were at max health, and they had been defending against champions.

Still trying to figure that one out >.>

 

[Ghosts]

I actually had (when playing mercurian) a lightbringer take out 3 lvl 5 angels. They were at max health, and they had been defending against champions.

Still trying to figure that one out >.>

Extremely lucky with those odds? hehe

 

[WarKirby]

I'm in an epic game just now. Can't remember exactly how many turns in, but it's probably a few hundred. I've just survived the apocalypse, defeated the four horsemen, and the Avatar of Wrath.

And ashen veil has never even been founded....
I'm not sure what the sheiam are doing, but it doesn't seem to be pursuing world destruction. They are kind of fighting back and forward with the mercurians atm, but you'd think they'd find some time to research Corruption of Spirit somewhere in there..

 

[Sarisin]

That is kinda strange. You are at least Armageddon Counter 90 and AV has not even been founded yet?

How did the counter get raised so high only a few hundred turns in without AV? Epic has 990 turns by the way.

 

[snarko]

The AI has no idea when to take cities and when to raze them. Seen Tasunke take over so many cities in such a short time he was losing money on 0% science. He lost a lot of units to desertions with most of his cities building new units only to lose them again to desertions.


The AI does not understand how important commerce is in the early game. Particuarly those forced to beeline something which there's little reason to beeline, like bronze working. Warriors are superb fighters in the early game, you don't need anything else.

 

[zup]

I have seen AI do better on settler than this game on prince. This game has too many variables to consider to make the AI always go for Bronze Working immediately. It is a viable strategy if you settle near two gold hills though. However, since education is so critical, the AI should pursue it more aggressively. Beelining education is usually a good choice.

Had Capria razed those cities she might have done better too. Perhaps some kind of mechanism should be developed to make AI raze cities it really cannot afford to keep. I can understand Falamar failing to develop his economy because my horsemen (flanking III, drill IV, blitz) kept him replacing units all the time.

 

[WarKirby]

That is kinda strange. You are at least Armageddon Counter 90 and AV has not even been founded yet?

How did the counter get raised so high only a few hundred turns in without AV? Epic has 990 turns by the way.

Well, I turned Last Days on, that probably helped, I imagine.

I did a little city razing in the beginning. It was maybe 13 or so when I settled down to grow peacefully. From there, it steadily rose over a looong time, though I haven't a clue why. Maybe the AIs were razing cities in their own little wars.

Started going up a bit faster after 40, due to Stephanos razing cities. And it was pretty much all downhill from 50, because Buboes increases it every time he kills anything that's alive. The gap from 50-100 was a hell of a lot shorter than what came before.

Edit: I've just checked the time. It was exactly turn 600 when I killed the Avatar of Wrath. AC dropped back from 100 to 97 on his death

 

[[to_xp]Gekko]

yeah, AI controlled civs beeline way too much. they should be thaught that a decent economy comes FIRST.

 

[Ajidica]

Then you run into the problem where the AI civs are complete pushovers because they have weak troops but nice land. At least now they may still be pushovers, but their land isnt as attractive because you have to build it up.

 

[Sarisin]

The AI has no idea when to take cities and when to raze them. Seen Tasunke take over so many cities in such a short time he was losing money on 0% science. He lost a lot of units to desertions with most of his cities building new units only to lose them again to desertions.

When have you seen an AI civ or the barbarians raze a city (pop. 2 or more) as opposed to capturing it? I just haven't seen it since several FFH versions back. The AI captures every city and, as you said, it creates a logistics nightmare for them. Basium is especially good at this.

I'd like to see something that lets barbs and AI capture oh maybe only 50% of the time and raze the rest. Of course, this would advance the AC higher in games.

I most of my games I find the capturing of cities a little out of control, especially with the barbs, who you would think would raze more than they capture.

 

[zup]

Well I'd expect the AI to get at least education on prince. Its requirements are agriculture and ancient chants, two extremely important technologies in every game. I don't do warrior rushes so AI not having str 5 units by turn 50 is not an issue.

agriculture => farms => food + 3 health resources (wheat, rice, corn)
agriculture => calendar => 4 happiness resources (silk, cotton, dye, sugar) and at least 4 commerce resources (silk, cotton, dye, incense; not sure about sugar)
agriculture => animal husbandry => horses (hammers + commerce), cow (food + hammers), pig (food), sheep (food + commerce); cow, pig and sheep give health too
agriculture => education => cottages = a reliable, effective source of commerce, can be used almost anywhere
ancient chants => mysticism => elder councils, a specialist slot and extra beakers, a cheap building.
ancient chants => education => cottages (see above)
ancient chants => monuments => culture

Furthermore, education is required for technologies like warfare and code of laws (which make lots of sense for Bannor)

Now lets add some numbers to illustrate my point further, particularly research costs.

Bronze working beeline for Bannor = 725 (modified by handicap and game speed of course)
crafting = 125
mining = 200
bronze working = 400

Education beeline = 500 (again handicap and game speed modified)
agriculture = 125
ancient chants = 125
education = 250

Bronze working is 45% more expensive than education for Bannor to research. Now one might argue that in the time it takes to get education, you can get about 67% to bronze working for some axeman rushing. However as one builds cottages and those grow (cottage=>hamlet=>village=>town), commerce rate starts to go up. More commerce = more beakers + more gold for unit and city maintenance. I don't have time to experiment if it is actually possible to increase one's commerce so quickly after getting education that one could get both education and bronze working in the same number of turns it takes to beeline bronze working.

Obviously if the Bannor started near lots of gold or gems, they could mine those for commerce. In this case beelining Bronze working is a means to get some gold/gem mines for happiness and commerce, as well as axemen for early warfare. However, the AI beelines regardless of whether it has gold or gems its cities can work.

Furthermore, though the AI might get those axemen, it is not guaranteed to have access to copper. And str 4 axemen will not bust a city guarded by str 3 warriors easily. If a human player could not locate accessible copper when he finished researching mining, he would abandon the beeline gamble here. The AI just pushes on. Perhaps even a small check on game start if there's copper around will do.
IF find copper
THEN beeline bronze working
ELSE IF find gold
THEN beeline bronze working
ELSE IF find gems
THEN beeline bronze working
ELSE beeline education
Since the AI already cheats, it does not matter if it knows where unrevealed resources (copper) are from the beginning. It probably already does. Only it does not change its play style accordingly.

As I have said before, education is a critical technology but so are its prerequisites. There is no reason not to get education asap. It is viable even if you have gold in your capital bfc. Without commerce to pay for cities and units, axemen for early conquest are worthless. And conquest is the only way the AI wages war afaik.

 

[Fafnir13]

Hmmm...I used to beeline for education more, until I found out I really wasn't building many cottages anyways (unless I've got plenty of floodplains around). Tend to spam farms more, what with all the mines and specialists I try to support.
These days, I rush for God King, then head up to mining and use a priest to fetch RoK (usually just before or after turn 100). Then I start to think about the other techs based on what resources and how many cities I have. Education doesn't usually become a priority until I have about three cities or so (from conquest or building).
Not sure how effective I'm really being. Seems to be working out, but I don't play anything higher then monarch.

 

[smusebaer]

could it be, that the AI in multiplayergames is even worse? We had a darkelf, with just one city, with farms and cottages, but just a citysize of 2 in turn 400. And there was allways plenty of room to expand...

 

[Anathema]

The AI does seem very.. inconsistent. I usually play on emperor, and there are AIs that do amazingly well - expanding rapidly, conquering other AIs, and simultaneously surpassing me on research. But there are still AIs that do terribly, they build a couple small cities and wind up with a stagnated economy and hardly any research. It isn't always the same civs either, I usually go with 8-10 random AIs and each game different AIs excell and different ones fail horribly.. although there are some exceptions, the Luchuirp are always terrible, they like to spam warriors and chariots and never do build many golems.

 

[zup]

Luchuirp are not very good at leveling and keeping alive Barnaxus so they are better off with Chariots.

 

[MagisterCultuum]

Now that difficulty levels let the AI cheat by giving them free xp, the Luchuirp whose main armies cannot use xp are at a bit of a disadvantage. (This change was awesome for the Mercurians though.)