AI improvements "Brave new world"

[RohirrimElf]

Hi all,

Ive been away for a while from playing civilization 5. I`ve played gods and kings for hundreds of hours and the basic game for even more hours. I liked what the developers have done on gods and kings when it came down to the AI`s intelligence in battles. But some things about the game kept me for playing more. Like the AI needing a vast number more troops to take me on 1 on 1 effectively. And runaway civs which gobbled up everything in their paths when they become overpowered (i felt rushed to do the same)

My question is: What progress have been made to the AI on Brave new world? Can anyone give me a link of what the AI`s competence will be compared to the latest patch on gods and kings. Or if the developers have given up hope on creating a better AI.

I like the improvements on diplomacy and the ability to play more peacefully and still succesfull. My fear is that this might have something to do that the developers have no clue on how to improve the competence of the AI in battle. And are trying to devert our attention from some of the worst parts of the game to a more peacefull game in which warmongering is no longer the most important thing. It is clever but does not fix the problems i had with playing G&K and i would loose interest at some point.

Can someone give me a link on what is said about the changes to the ai?

I am tempting to buy the game on steam for 10% less but my fear remains that not enough has been done on making the AI more effective with using less cheats to do so.

Thanx for reading my post and i hope a Civ fanatic can make me buy Brave new worlds.

 

[Circuit]

I too would love to know what effort is being made at improving the AI. The current cheating AI is the biggest drag on the game, in my opinion. It prevents players from fully utilizing many aspects of the current game (namely AI city spamming, in which insane bonuses to happiness have the potential, when combined with the right policies or beliefs, to create mathematically divergent systems where MORE cities lead to MORE happiness; players become forced to go tall in order to compete), or it creates vulnerabilities that players then easily exploit (I die inside every time the AI moves a great general right by my units in war, completely exposed with no friendly units in sight; "Well, I guess I'm obligated to kill you"). AI wonderspamming is also very annoying (I have to roll my eyes every time I find Petra was built in a city with only one desert tile nearby).

I believe Firaxis has the resources to create a truly intelligent, non-cheating AI. I hope that the only reason we haven't seen one appear yet is because they want to release a final iteration of Civ and then fix the AI in one fell swoop.

 

[bcaiko]

Yeah, hopefully someone will ask the devs about what they're doing on this. The AI has made progress since Civ5's initial release, but they still make some silly decisions.

Also, if the AI is improved further, the AI's rediculous bonuses (as nicely described above) will have to be toned down. I'm really tired of seeing runaway Rome on another continent.

 

[KrikkitTwo]

I believe Firaxis has the resources to create a truly intelligent, non-cheating AI. I hope that the only reason we haven't seen one appear yet is because they want to release a final iteration of Civ and then fix the AI in one fell swoop.

They do not have the resources for an AI intelligent enough to not have bonuses. They can
1. Make the AI better
2. Give the AI different bonuses (less distorting)
3. Change game mechanics that humans can exploit (ranged combat)

 

[RohirrimElf]

Thanx guys,

Do i have not seen any info in this treath about the AI in the next expansion i`m glad more people feel the same. It just feels wrong to start next to another civilization and see him quickly pop 4 citys with a decent army while you only have created one city beyond your capital. The thing is you do know that you will beat your AI neighbour if you dont fall behind to much. If you have reached a certain point with your civilization you just know you can steamroll over you opponents. The game (Gods and kings) just does not feel right at this point. I`m playing at immortal difficulty. You can tell me why i`m not playing on deity but then the AI will not get more competent. It just cheats more.

I`m thinking of waiting for the reviews to pop up around release date and checking the forums from time to time on this issue. I`ve played the game since CIV2 and have played civ 5 for hundreds of hours but when Brave new worlds is not an improvement compared to G&K ill rather stick to company of heroes 2 for the time being.

I`m hanging around till i have more info on this till release. Please dont take my last 2 posts as a cheap way of trolling the forums. I`m not posting this treath to make people not buy this game. I can understand that a lot of people do like to play this game. It`s just that after playing the game for a lot of time the bad parts are taking over the good. And i cant play any more till i know i will get challenged or entertained in a good way on civ 5.

 

[Gamewizard]

What you are asking for is tactical AI improvements...an AI that can conduct war. I personally am not expecting improvements in this area as I think Firaxis would have touted that feature already, because it is what fans want the most. Firaxis wants this game to be accessible by the masses, not just by those who own supercomputers that would be able to process a complex tactical AI in a reasonable amount of time. Those running the game on older machines would see turn times increase to the point of the game no longer being fun.

However, as has been mentioned in other threads, there are certain things I think they could fix without massively increasing processing time, such as:
- AI embarking units when an enemy vessel is nearby.
- Great Generals wandering on their own
- Suicidal air units

 

[Talcove]

What you are asking for is tactical AI improvements...an AI that can conduct war. I personally am not expecting improvements in this area as I think Firaxis would have touted that feature already, because it is what fans want the most. Firaxis wants this game to be accessible by the masses, not just by those who own supercomputers that would be able to process a complex tactical AI in a reasonable amount of time. Those running the game on older machines would see turn times increase to the point of the game no longer being fun.

However, as has been mentioned in other threads, there are certain things I think they could fix without massively increasing processing time, such as:
- AI embarking units when an enemy vessel is nearby.
- Great Generals wandering on their own
- Suicidal air units

By far, the worst is the AI's lack of understanding of siege units. Even when my city has no shot blocking terrain around it, the AI seems to think it necessary to zip their cannons right up against my city before setting it up.

 

[bcaiko]

- Suicidal air units

To be fair, even the player can't always accurately guage when an air attack is suicidal, since the combat prediction doesn't show how much damage your air unit will likely take. I don't mind that the AI sometimes gets its air units destroyed. Heck, they probably got those for free, anyway.

I suppose you could be talking about the AI's infrequent tendancy to move its air units onto random Carriers or into cities that are about to be taken by an opposing force, though...

 

[SnowKomodo]

The game is even more complicated than before. I have a tough time believing the AI has seen significant improvements

 

[Gali]

You need to remember the best AI from Civ 4 was community driven. Since Firaxis waited too long with the DLL they don't have that advantage this time.

As for the AI. I recently had an invasion of Persia as Polynesia. All naval units and Persia built a dozen subs. Their subs did not fire a single shot at me in the war. They wandered up to me and got blown out of the water. With an AI that broken after an expansion focusing on naval combat just lower your expectations.

 

[SnowKomodo]

As for the AI. I recently had an invasion of Persia as Polynesia. All naval units and Persia built a dozen subs. Their subs did not fire a single shot at me in the war. They wandered up to me and got blown out of the water. With an AI that broken after an expansion focusing on naval combat just lower your expectations.

I very rarely have to worry about naval invasions. I mean, G&K improved on that pretty well but it was still a non-issue overall. Fingers crossed they improve it. Because that would give archipelago maps much more depth and increase its replay value

 

[Gali]

No No I was invading him. Please naval invasions what are those. Playing on water maps is pretty much an always peace game even on King. It is so sad to me that conquest is easier in CiV than in civ 4 even though I play 2 difficulties higher. I'm probably going to favor Pangaea Plus maps in BNW for sea trade routes with actual AI competition.

People knock on the sea AI, but the land AI is just as bad. The thing that saves the land AI is terrain and city defense. It was far harder to take one Ottoman city because it was screened by rough terrain, than to destroy a much more advanced Persian Empire. Of course artillery destroys that difficulty and bombers make it completely laughable. Empires twice my size with tech parity, and a military 3 times my size. With artillery and/or bombers they go down just as easily as a backwards civ, just takes longer.

The combat model on this game is so broken. It has to make city defense and terrain overpowered to shield the AI. Of course that makes a tall defense way too easy, but that's the price. It also makes ranged and siege units hideously overpowered and makes melee units a joke. Then that dependence makes late game war, which removes city defense and terrain, a complete joke.

For those who think they can fix it, look at this. Horse-Melee-Iron- Ranged. Those are the unit types that have been overpowered in CiV. Every time they fix something, something else breaks.

 

[headcase]

They do not have the resources for an AI intelligent enough to not have bonuses. They can
1. Make the AI better
2. Give the AI different bonuses (less distorting)
3. Change game mechanics that humans can exploit (ranged combat)

2) would be sweet, especially with the new discontentedness feature with ideologies.

- Far less happiness than the current situation on all Prince+ difficulties
- Less gold bonuses, in favor of a GPT "floor" on higher difficulties, where if the AI is at less than x net gold per turn, they ignore their actual GPT and get x gold for that turn instead. Dumb, but less dumb IMO than +a million GPT. The AI would not be actively aware of this, and so wouldn't use this floor strategically. Either that, or scale the bonuses so they are less when the AI is doing well for GPT.

- more science catch-up effect (and faster stealing)
- more culture (at least on King it feels not quite high enough compared to other bonuses)
- faster workers on high difficulties (if not already)
- more GP production (although more GEs would be annoying for wonder races).
- on the highest difficulties, certain buildings (e.g. monuments) are given for FREE upon settling a city.

 

[Karmah]

If you really want to help the devs make something out of those 'AI strat sucks' threads, please indicate simply how many % increase of calculation time for each AI engaged in a war you'd feel would be acceptable.
They can make a very good AI with 15 minutes to think each turn for each CIV , and a very stupid one with 5s...there is a middle ground for which they set obviously. If you want to change that setting , indicate what you are willing to give

EIDT : actually the best would be to have a difficulty setting where you accept the calculations to go to the roof in exchange for a challenging opponent in war.

EDIT: oh god , i m so unclear , can someone rewrite my tough in english please , i m too tired

 

[BAshley]

If you really want to help the devs make something out of those 'AI strat sucks' threads, please indicate simply how many % increase of calculation time for each AI engaged in a war you'd feel would be acceptable.

I normally wait around 20 seconds per turn on all standard settings. I'd be completely fine with 45-50 second waiting times if the AI would make a couple fewer barmy decisions per game.

 

[AriochIV]

My question is: What progress have been made to the AI on Brave new world? Can anyone give me a link of what the AI`s competence will be compared to the latest patch on gods and kings. Or if the developers have given up hope on creating a better AI.

The developers are often asked about improvements to the AI, but they've said very little about it. Dennis Shirk has made a few vague comments about how the diplomatic AI should be "less schizophrenic" and more amenable to long-term friendships, but nothing concrete. Whenever they are asked about the tactical AI, they usually say that they felt it was a problem in the vanilla release of Civ V, but that they've improved it a lot since then and that they're satisfied with where it is now.

I would expect that most of the AI work in BNW has been to teach the AI how to use the new features. I would not expect significant improvements to the existing problems. I do not expect that until a later patch, and not until after there has been a lot of screaming about it, because I don't think that they believe it is a problem.

The AI can always be improved. Those saying that it's not possible to improve AI without lengthening turn processing times are ignoring history; AI has improved a lot since vanilla without lengthening turn times. Algorithms can always be improved and optimized and made more efficient, especially when they are as complex as these must be.

 

[Kimuyama]

One thing with the AI that bothers me is that the other leaders, seemingly, go by their own feelings and not by what is best for their empire:
One game I played as Japan and conquered a lot of stuff. This of course annoys the other leaders, as it should, but the Ottomans, whom I had saved from the brink of extinction be the empire formerly known as Maya, plus that in good will I provided for 70% of their economy, traded them pearls and only wanted fur in exchange. They were not in a position to denounce me, yet they did so. There should be some algorithm that decides that, even though Player X is a douche, they should try to stay on his good side in such a situation.

 

[Balerune]

All this AI talk always seems to center on battles to the exclusion of other aspects of the game.

Years ago I played a game of Civ IV. Everything was under control, I was well ahead in score and my empire was stable and protected. Around 30 turns from the end there was a noticeable shift in the score. Other AIs were dropping points and Mongolia was steadily accumulating points. I sent a expedition over to one of the other continents and found that the Mongols were doing (ingame) what they had done historically: they were destroying every civilization they could get their hands on. I had to whip up a force (I was quite peaceful) and destroy them before they gained enough points in order to win my game. It is a game I will never forget, mostly because the endgame was unexpected.

Its not all about war battles. It is about management, flexibility and being able to overcome the unexpected. That is what I always enjoyed about Civ games.

 

[Karmah]

The AI can always be improved. Those saying that it's not possible to improve AI without lengthening turn processing times are ignoring history; AI has improved a lot since vanilla without lengthening turn times. Algorithms can always be improved and optimized and made more efficient, especially when they are as complex as these must be.

OK fair enough but there is a limit (information theory wise) to how much you can do with the same calculation time (at the end it s all only data compression in a sense).
Especially since most of the calculation time gained comes from cleaning inefficient code or rewriting it , there is only so much you can do once it s cleaned
Yet if you want the discussion with a dev to be fruitfull , tell him which margin he gets.
If you say none , then don't be surprised if only a few tweaks occurs.
My point is , we all have a list very long of "minor tweaks" in minds, but it s a very long list, yet noone has ever put itself in the dev position.

 

[JtW]

Regarding turn times: I think it would be possible to make AI use the time during the player's turn for thinking. Sure, some units might get repositioned and the pure tactics would have to occur during the AI's turn, but couldn't the whole strategic layer be moved to the player's turn (which is taking several minutes more often than not).
Also "peaceful" decisions like which tiles to improve, which cities to convert etc. could be handled during that time.

Wouldn't this be a possible solution? (It's a genuine question. I don't know much about programming, but this is the first thing that comes to mind when I read about how turn times pose a limit on the AI quality).