I hope and pray for The AI Patch next

[MeatyThud]

This game, it would be so great, if I don't have to play against an imbecile. Multiplayer may be fun, if someone will spend 50 hour game on my schedule, when I want, when I can. So not gonna happen. I just wish, I truly hope and pray, some proper AI programmer be given this task. In the age of big data, crunch a million games, surely, and let the lessons be learned. Or reason it, step by step, the answer to each challenge. But please, I beg, I wish and hope and pray, or what is this game for, which we have no-one to play with?

 

[NASAROG]

I'm with you.

 

[Galgus]

Personally I'd love to see land-based AIs actually build a navy if they have sea cities, so they aren't just easy pickings.

 

[LordHater]

Lack of decent AI and some major bugs is why I haven't bought the game yet.

 

[Lord Tirian]

In many ways, while I'd like to see an AI patch, I'd much rather see them releasing the DLL. Civ4 and Civ5 had nicely improved AI mods.

Yes, I know, mods shouldn't be the solution to it, but AI mods are usually developing and improving for a long time and incorporate a lot of incremental improvements over a long time singular patches often can't really match.

With the DLL out there, people can start working on it or even port Civ5 AI improvements over to Civ:BE.

 

[PhilBowles]

Ideally, they'll focus on the fundamental gameplay problems above all else. Apart from anything else, if they were to improve the AI (which is unlikely, as they didn't substantially improve the Civ V AI over time), there's no point doing so and then have to recode it again once the game is fixed.

 

[Auren]

I'll be honest. The AI in Beyond Earth/CiV was the main reason I never bothered to even buy Rising Tides. There is no reason for a primarily single player game not to have a challenging AI. If it was too hard to code an AI given the hex/one unit per tile then they should never have used it as a primary game system. I'm frankly very disappointed they've never bothered to improve it.

This is the same [crap] product development choices that drove me completely away from the Total War Series.

 

[Gorbles]

I was under the impression that regardless of the tile-based system, the AI has always been poor in the Civilisation games, and indeed, is a problem across the entire strategy genre.

 

[spfun]

Programming AI is hard, they probably want a better AI but just can't make it better or are limited by the engine.

 

[Ishea]

Hmm.. speaking about big data.. imagine.. a (super) computer hooked up to every civ game, with a neural net, learning how to play civ from all those games.. THEN we'd get a scary opponent methinks.

 

[PhilBowles]

I'll be honest. The AI in Beyond Earth/CiV was the main reason I never bothered to even buy Rising Tides. There is no reason for a primarily single player game not to have a challenging AI. If it was too hard to code an AI given the hex/one unit per tile then they should never have used it as a primary game system. I'm frankly very disappointed they've never bothered to improve it.

This is the same [crap] product development choices that drove me completely away from the Total War Series.

1UPT is not the issue with the AI - the ranged unit system is. Most of the AI's bad decisions, from unit positioning to embarking within range of cities or enemy ranged units, boil down to being unable to appropriately factor in ranged units (and, later in the game, aircraft, as these operate in a similar way albeit with much longer ranges). Ranged units other than artillery should simply be dropped from the game - archers etc. would act like melee units with some kind of attack or defence bonus in much the same way as in Civ IV.

Civ V AIs are capable of dealing with aircraft to some degree - not as well as they should, but far better than BE that appears to have reverted to a pre-G&K form. Civ V AIs always build aircraft of their own and occasionally will respond to air aggression by setting interceptors; BE AIs never even seem to build aircraft. The Civ V AIs are also helped by the existence of anti-aircraft units, something that doesn't exist in Beyond Earth. Of course, aircraft don't use the 1UPT system, but they can be much more devastating in Civ V than in Civ IV.

Personally I wouldn't want to see the aircraft system change, since unlike every previous entry in the series Civ V does a very good job of showing warfare evolving over time (once artillery and later air power arrives, battles just play very differently) instead of just giving units stat upgrades - however the AI does need to be able to counter air power more effectively.

Warfare isn't the main failing of the AI in any event - the AIs are bad at it, but they're also typically bad strategically. Civ V's may be the best, in part because of the simplified tech tree relative to BE or Civ IV, but even so it will commonly get halfway to a spaceship, for instance, and then just stall. The military failings of the AI become a problem only when the AI has no other way to win.

In Civs I-IV, you generally lost to being wiped out by a hostile AI, or you won, since the AIs couldn't achieve any of the other victory conditions reliably (save in Civ IV diplomatic victory, since that was basically a subtype of domination). In Civ V the AI could compete and win non-military victories more consistently, so unless you yourself were going for domination its relative weakness in warfare was less important. BE's back to the bad old pre-expansion Civ V model where the game both can't handle 1UPT as well as Civ V's later incarnations and can't achieve non-military victories - in part because most of them have affinity thresholds the AI is apparently incapable of reaching before the player.

 

[Callonia]

In Civs I-IV, you generally lost to being wiped out by a hostile AI, or you won, since the AIs couldn't achieve any of the other victory conditions reliably (save in Civ IV diplomatic victory, since that was basically a subtype of domination). In Civ V the AI could compete and win non-military victories more consistently, so unless you yourself were going for domination its relative weakness in warfare was less important. BE's back to the bad old pre-expansion Civ V model where the game both can't handle 1UPT as well as Civ V's later incarnations and can't achieve non-military victories - in part because most of them have affinity thresholds the AI is apparently incapable of reaching before the player.

I agree with bad ranged decisions from the AI.

But the hostile AIs tend to have pretty hard time killing me because I generally go overkill on garrison strength. And always build barracks.

The Ai could send three stacks of doom, I can counter with 10+ very easily, but to do so, I would certainly have to mobilize the inactive garrisons sitting in cities. I've been laid siege on same time by five AIs once. They all got destroyed by my superior army strength because I focused on economic base aspect of civ models. Then I go on march and take the attackers out one by one.

1UPT forces me to be alert. And use my units carefully instead of firing them off in suicidal attacks. Because I only have like 10+ doing a siege on city here instead of 100+. And if the AI have decent force of defenders left and they might be able to force me away from the city at least to out of the range of it's bombardment.

 

[Gort]

Bad balance leads to bad AI a lot of the time.

Developers will design choices into their game, so a player picking technology (for example) might have ten choices they can make at a particular moment. The developers will have tried to make each of these choices valid in different ways - the skill of a player is expressed in their ability to make the right choices for their situation.

In Civ 5 and BE, AIs are tagged with "flavours" which determine the kind of choices they make. Hutama might go with choices that give him energy, while Elodie might prioritise culture. All well and good.

HOWEVER, if the game isn't well balanced, many of the choices on offer might be "trap" choices, and many might be no-brainer choices. A decent player can tell the difference, and will pick the no-brainer choices, while the AI has been designed to treat all choices as valid and pick according to its flavour, and fall into a lot of traps along the way.

The result is that the AI ends up looking like it doesn't know how to play the game, when in reality it's just trying to play a better-balanced game than the one the devs made.

In short, if you make a balanced game, making an AI that plays it better gets a lot easier.

 

[nimling]

Ranged units are a major stumbling block for the AI, and 1UPT necessitated broken ranged units (among other design decisions which just make the game less-fun and flawed). So it really does come back to 1UPT.

MUPT would be the obvious solution to make the game better in just about every respect. The existence of flank bonuses and zoc is enough to encourage players to split their stacks, and stack counters as in Civ4 would be helpful to discourage stacking. In Civ4 the problem was that stack counters had to be protected by stacks themselves, at least until air units became a thing, and that happens far sooner in BE.

1UPT adds nothing to the game but a lot of frustration and AI bottlenecks. I am hoping for a quality MUPT patch, as once that is cleared up Civ5 could be made into a much better game.

 

[PhilBowles]

Ranged units are a major stumbling block for the AI, and 1UPT necessitated broken ranged units (among other design decisions which just make the game less-fun and flawed). So it really does come back to 1UPT.

There's no logical reason for ranged units to be necessary in a 1UPT system - most 1UPT games do perfectly well without them. So no, it's not fundamentally an issue with 1UPT.

MUPT would be the obvious solution to make the game better in just about every respect. The existence of flank bonuses and zoc is enough to encourage players to split their stacks, and stack counters as in Civ4 would be helpful to discourage stacking.

MUPT has the same issue: ranged units need to go. Otherwise you either restrict stacks to melee units and civilians, or set a damage rule that ranged units take damage in preference to melee units on the same tile, as the alternative is protecting the ranged units that deal the damage - the same thing that can make city defence so easy in Civ V. Since it's not by any means clear that 1UPT is an issue without ranged units, simply removing this feature is the best place to start.

In Civ4 the problem was that stack counters had to be protected by stacks themselves, at least until air units became a thing, and that happens far sooner in BE.

And what would prevent the same happening in Civ VI or a modded/patched Civ V, if this system were implemented there? As for BE, it has trouble dealing with air units as it is as it seems unable to use its own as interceptors and unlike Civ V there are no AA units. It doesn't seem great design to change the system to one that further benefits a human who can exploit it better than the AI.

 

[leif erikson]

Moderator Action: Please return this discussion to the subject of the thread. Straying into another 1 UPT debate is not productive.
Please read the forum rules: http://forums.civfanatics.com/showthread.php?t=422889

 

[Acken]

I was under the impression that regardless of the tile-based system, the AI has always been poor in the Civilisation games, and indeed, is a problem across the entire strategy genre.

That's why the AI gets bonuses. The problem is when people find the AI poor when it has already huge bonuses.

Programming AI is hard, they probably want a better AI but just can't make it better or are limited by the engine.

Making an AI play "better" is really not that difficult (which is not the same thing as having an AI play well). Especially when it comes to economy. 1UPT AI combat and army management is the biggest difficulty and which is why I think it should at least be AI-friendly so more melee focused and less range focused.

For the economy experience should end up in improving the code. Making the AI play efficiently on day 1 is difficult but once good habits are identified they should be implemented in the next patch. That's basically what most AI mods do... they recognize that science is important for example so increase the likelihood of the AI having a sound science strategy. An AI randomly building stuff is bound to not be very efficient.

Add AI friendly mechanics/balance in both combat and economy and you'll have a tougher challenge. Guaranteed.

 

[Westwall]

The AI in BE/BERT is actually worse than Civ 5. Some people will claim it's a problem with the whole genre, but they're wrong.

Amplitude for example has a team specifically dedicated to developing and improving AI opponents, which is a far cry from what we're seeing at Firaxis, specifically the BE team.

 

[Acken]

Erh... the AI in endless space is really poor.

 

[Sherlock]

Lack of decent AI and some major bugs is why I haven't bought the game yet.

And Endless Legend.