AIs and the Art of War

Thank you thank you thank you Koshling.

Now AI is very dangerous with huge stacks of death.... by the way i lost Thebes :D
 
Two problems with AI I keep meaning to mention.

1) Hunter units on auto hunt should not be exploring goody huts.

2) Ships on auto pirate seem more interested in exploring and hunting animals rather than pirating enemy nations.

2a) Ships on auto pirate should leave your vassals and colony ships and improvements alone.
 
1) Hunter units on auto hunt should not be exploring goody huts.

also are they still Suicidal with Terrain damage?
i've had that issue they would just Wander into the Desert go ouch ouch walk back out Heal and walk back into Terrain damage and then Repeat at infinitum :lol:
doesn't mean the Exact same spot but usually they go like Ok i am healed Back into Danger >.>
 
also are they still Suicidal with Terrain damage?
i've had that issue they would just Wander into the Desert go ouch ouch walk back out Heal and walk back into Terrain damage and then Repeat at infinitum :lol:
doesn't mean the Exact same spot but usually they go like Ok i am healed Back into Danger >.>

They are a bit dumb in regard to terrain damage, but not suicidal like they used to be. As you have seen they do know enough to back out and heal periodically, but if the best explore/hunt spots they know of are still in damaging terrain they will go back in briefly and do so repeatedly.

The cycle will be broken by one of a few possibilities:
1) A hunting target shows up within visibility range in some other direction (so they go after that)
2) You gain visibility on more terrain just beyond the damaging area so it is no longer so attractive in term of need to explore
3) They gain a promotion, in which case they are very likely to choose the relevant terrain damage immunity

It does need more work, but it's better than it used to be.
 
3) They gain a promotion, in which case they are very likely to choose the relevant terrain damage immunity

Well usually the promotion that solves it is the promotions for Deserts and Tundra :lol: but those promotions are only Secundairy on the list because i want the hunter promotions First so they are more usefull
 
Koshling, if you want to improve the AI on Terrain damage, I'd be happy to help you!

- I think the AI should be more foresighted. If there is a big desert to cross to get to a fertile new land (or it's enemy), the AI should promote a few units with the correct promotion, send them along with workers in the desert and should build some forts on the way (and defend them).

- Same if the AI want to siege a city in the desert. They don't need to promote all their units with terrain immunity, but a fort close to that city would be really helpfull.

While writing this, I recognized that it isn't really about terrain damage but more on the advantages of Forts.

Do you want to adress this next or is there something more pressing (beside bugfixes and crash fixes) like the "AI and SoD" or "AI and using ships as shortcut (Carthago to Italy or Denmark to Sweden for example)" things?
 
Koshling, if you want to improve the AI on Terrain damage, I'd be happy to help you!

- I think the AI should be more foresighted. If there is a big desert to cross to get to a fertile new land (or it's enemy), the AI should promote a few units with the correct promotion, send them along with workers in the desert and should build some forts on the way (and defend them).

- Same if the AI want to siege a city in the desert. They don't need to promote all their units with terrain immunity, but a fort close to that city would be really helpfull.

While writing this, I recognized that it isn't really about terrain damage but more on the advantages of Forts.

Do you want to adress this next or is there something more pressing (beside bugfixes and crash fixes) like the "AI and SoD" or "AI and using ships as shortcut (Carthago to Italy or Denmark to Sweden for example)" things?

I don't plan to focus on AI for a while, so not just yet I think. I want to continue with performance next (parallelize unit processing), which will be the next cycle, then work on memory usage the cycle after that. Once those are done I'll spend a cycle on AI. In the meanwhile I just plan to address critical AI issues (like you're totally blocked attack stack one a few weeks ago)
 
After a certain point in time, say 1100BC - a hunter + an atalist is not sufficient to defend a leveled up great commander. Its far too risky to go roaming around with him like this when horseman and other dangerous units are out. This is especially true if the AI has neighbours in close proximity. A leveled up great commander should be valued very highly, its as valuable as like a key position city for example.
 
After a certain point in time, say 1100BC - a hunter + an atalist is not sufficient to defend a leveled up great commander. Its far too risky to go roaming around with him like this when horseman and other dangerous units are out. This is especially true if the AI has neighbours in close proximity. A leveled up great commander should be valued very highly, its as valuable as like a key position city for example.

I agree. I need to do more work in that area at some point.
 
A bit Off topic, but why is the AI falling behind in the techrace? I'm in the early industrial era right now and there were no wars for over 100 or 200 turns. But the AI seems to fall behind. Take them 15-25 turns to research a new tech, while I need ~9 turns per tech. Eventhough they have almost twice as much cities than I have!
 
A bit Off topic, but why is the AI falling behind in the techrace? I'm in the early industrial era right now and there were no wars for over 100 or 200 turns. But the AI seems to fall behind. Take them 15-25 turns to research a new tech, while I need ~9 turns per tech. Eventhough they have almost twice as much cities than I have!

Look at their stats in the BBAI log - that should give you a lot of clues since it shows there research output and so on. If you compare an AIs stats with yours you may be able to pick out a pattern
 
A bit Off topic, but why is the AI falling behind in the techrace? I'm in the early industrial era right now and there were no wars for over 100 or 200 turns. But the AI seems to fall behind. Take them 15-25 turns to research a new tech, while I need ~9 turns per tech. Even though they have almost twice as much cities than I have!

Yeah when i was playing in the Industrial era i was starting to get bored:shake: as nothing was going on at all, but when i made it to the Modern era, all hell broke lose around the Stealth Tech, so go figure??:dunno:

It all has to do with the Civics that were chosen, as far as i can tell??
 
How do I turn on/look at the BBAI Log?

SO, when it depends on the Civics, the question is, WHY does the AI choose those civics?

No simple answer to that one. That would require a detailed analysis of the choice process in the context for your game.

To turn on the logging:

1) Enable logging generally (in the INI file) if you have not already (if you haven't you won't be getting any logs, such as Python logs etc., so if you are getting any that has already been done)

2) In the BUG screen go to the 'logging' tab and change BBAI logging level to level 3. That will cause detailed AI logging to BBAI.log
 
How do I turn on/look at the BBAI Log?

SO, when it depends on the Civics, the question is, WHY does the AI choose those civics?

It does so by who is the neighbor, are they at war, is the closest neighbor getting troops near their border, it has alot to contend with, so it picks the most drastic one, is all i am "guessing":dunno:
 
It does so by who is the neighbor, are they at war, is the closest neighbor getting troops near their border, it has alot to contend with, so it picks the most drastic one, is all i am "guessing":dunno:

Nope. Being at war will influence civic choice somewhat, but proximity won't.
 
Locust units have a NotUnitAI of UNITAI_ATTACK_CITY so they are not allowed to attack cities. I just lost a city to 6 locusts!

Not having a unit ai of UNITAI_ATTACK_CITY does or mean they cannot attack cities. It just means that their AI won't be focussed on attacking cities. However, if a city is an easy kill the standard UNITAI_ATTACK (and a few others potentially) will take out a city quite happily. If you want units incapable of attacking cities you would need a tag for that, which I don't think exists.
 
I think all animals shuld be able to cross border but not to attack cities.

Wild animals avoid big people stacks like villages or towns but it dosent mean that the do not cross inhabitaed border area at all.

This change will make animals more dangerous and give new depth to early gameplay. SO please do this tag.
 
Top Bottom