Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Rolan7]

Pac-man.

Surprisingly good, yeah.  Four distinct AIs which interact well together.  They don't play optimally, but the game's more fun than it might have been.

[BurnedToast]

Galactic Civilication II has a good AI and does not cheat on standard levels - in contrast of the AI of Civilization..

GalCiv2's AI is actually pretty terrible.

It's not the AI's fault though, the game is just horribly designed. The AI tries to follow the "correct" way to play, but that's actually a really terrible way to do things.

I'm not sure how good the AI would be if the game were properly designed.

(I would explain why buy it would turn into a big offtopic rant)

[nyulzsiraf]

CS:GO has the best, proof https://www.youtube.com/watch?v=MDxQ4NU47vA

[Leatra]

Spec Ops: The Line is a pretty underrated game that has a good AI. Assault guys provide cover for Juggernauts and if not, you can hear Juggernauts screaming "Where is my damn covering fire?" since they are pretty much helpless without support and they know it. If they have a good number superiority Juggernauts will usually try to get you pinned down with their suppressing fire and shotgunners will finish you off from both sides.

If they got you really pinned down, they'll throw a grenade and you'll have to run away and usually end up dead in higher difficulties.

They use cover system well, moving from cover to cover quickly and not lifting their heads above their cover like an idiot all the time (like how it is in most cover-based games).

Your teammates hold the flanks very well, so you are unlikely to get suppressed too much. There are a few moments you'll be on your own, and it becomes really difficult to not get pinned down. Your teammates will notice if you are getting pinned down and act accordingly. You can order them to throw stun grenades to get out of such situations. You can order Lugo to take out far-away targets with his sniper and Adams to suppress the enemy or throw grenades at pinned down targets.

[Kolbur]

Galactic Civilication II has a good AI and does not cheat on standard levels - in contrast of the AI of Civilization..

GalCiv2's AI is actually pretty terrible.

It's not the AI's fault though, the game is just horribly designed. The AI tries to follow the "correct" way to play, but that's actually a really terrible way to do things.

I'm not sure how good the AI would be if the game were properly designed.

(I would explain why buy it would turn into a big offtopic rant)

I guess you are talking about the economy and how the spending sliders work. I'd agree about that but the AI is terrible for other reasons as well, mainly being a complete failure at diplomacy and fleet management.

[a1s]

It's not the AI's fault though, the game is just horribly designed. The AI tries to follow the "correct" way to play, but that's actually a really terrible way to do things.

I'd like to read that rant. If you're worried about distraction, just spoiler it.

[Vendayn]

Ultima 7 has better (at least better scripted) npcs than any modern game. It has a better living world with better AI than what Bethesda puts in their games.

[BurnedToast]

It's not the AI's fault though, the game is just horribly designed. The AI tries to follow the "correct" way to play, but that's actually a really terrible way to do things.

I'd like to read that rant. If you're worried about distraction, just spoiler it.

Not sure how familiar you are with the game, and I love hearing myself talk (so to speak) so you get the really long version

Spoiler: long rant about galciv2 (click to show/hide)

Your economy is split into factories and labs. They don't produce a flat amount per, they convert dollars into beakers and hammers... so more labs or more factories lets you convert more dollars into beakers or hammers. So far, no problem.

The problem is, for some reason they decided it would be a good idea to force you to choose between the two. You can run factories at 100%, or you can run labs at 100%, or you can run them both at 50% (or whatever combination you want, it's a slider). This means if you want to build a spaceship fast on one planet, you have to literally shut off your research completely for your whole empire... and if you want to research something quickly you need to stop upgrading buildings on your planets, even the ones with no labs. It also means you're always wasting slots on planets (and planet slots are very precious) since you can't ever run everything at 100%. Note - this is completely arbitrary, it's quite possible to have enough money to run both at 100%, and it's not an issue of population (since how many people you have does not influence production) it's just the way the game is.

HOWEVER - you can press a button on a planet that converts beakers to hammers, or hammers to beakers. Yeah, that's right - you can somehow make your laboratories build spaceships, or your factories do high tech research. It's not a perfect conversion, they produce less and there's some waste... but it's a LOT better then the alternative of letting building sit around half used.

So that leads to the "all X" strategy. You either build all factories, or all laboratories (there's advantages and disadvantages to both), and then just convert to hammers or beakers as needed. It also has the huge advantage of only requiring one tree worth of research (no need to upgrade factories if you're going all labs) which saves a ton of beakers too. Since the AI does not do that, you can easily outproduce him in beakers and/or hammers by a truly massive amount.

Second... the combat system is designed such that ships fight to the death, and there's never any ties. However, both ships fire at the same time which means they can possibly destroy each other... but there's not allowed to be a tie. So how does it decide? very simple, the ship with the most guns (well, highest offensive rating) wins. That means if you have a ship with enough guns to completely destroy the enemy ship in one round, as long as the enemy ship does not have a higher offensive rating that ship is immortal, and will always win any combat with 1 hp left. Combine that with the fact that every weapon type is good against two armors, but each armor is only good vs one weapon... and the obvious strategy is to just stack as many guns as possible, completely ignore all defense. The AI does not realize that, and tries to build balanced ships, which means sometimes you can destroy his whole armada with a single 1 hp ship.

Third - fleets. Fleets are supposed to be a bonus, in fact you need special researches to make bigger fleets. However, again because every ship gets to fire in a combat round that means you are better off not grouping your ships in fleets. Why? take this example - you have 10 ships that are all guns. The enemy also has 10 ships. Your ships can kill two of his per volley since they are all guns, and his can kill two of yours since you have no armor. You send them in a fleet vs his fleet, the end result is you win after one round, with a single ship with 1 HP left (since you have more offensive rating).

Send them in one at a time however and what happens? You one ship fires, kill two of his. His 10 ships fire and massively overkill your ship. He now has 8 ships, you have 9. Send in the next ship, he's got 6 ships you have 8... keep repeating it and eventually you have 6 ships left, 5 undamaged and one with 1 HP left. The enemy has zero ships left.

In short, fleets are completely terrible and actually harmful to you. And... you guessed it... the AI loves fleets and makes them all the time.

Well, anyway, sorry if there's any small mistakes in that long rant I haven't played in a very long time (the sliders make me mad just thinking about it) so I might be misremembering a little bit.

Edit: that's not even touching on diplomacy, as kolbur mentioned, which is also completely and hilariously broken to the point that with a big enough diplomacy bonus (from research and/or racial picks) you can basically make the AI empires do whatever you want for pocket change.

[Shadowgandor]

Hah. I never knew about all those exploits. Whenever I played the game, I basically played similar to the AI. They were pretty damn hard when the player is using the same mechanics as the AI

[Rez]

You're using the same mechanics regardless.

[Shadowgandor]

Lets phrase it differently; The AI is challenging when you're taking a similar approach (going both research and production, using fleets and balanced ships, etc)

[BurnedToast]

Lets phrase it differently; The AI is challenging when you're taking a similar approach (going both research and production, using fleets and balanced ships, etc)

But that says nothing about the strength of the AI.

For example in chess you will sometimes sacrifice a weak piece to capture a strong piece.

If the AI is too stupid to do that, then of course you can make the game more challenging if you don't do it either, but that just means you're both playing sub-optimally, not that the AI is good.

[jhxmt]

Lets phrase it differently; The AI is challenging when you're taking a similar approach (going both research and production, using fleets and balanced ships, etc)

But that says nothing about the strength of the AI.

For example in chess you will sometimes sacrifice a weak piece to capture a strong piece.

If the AI is too stupid to do that, then of course you can make the game more challenging if you don't do it either, but that just means you're both playing sub-optimally, not that the AI is good.

To some extent, though, playing in such a way as to take advantage of known flaws in the AI is counterintuitive - it's a strategy that only works when you know that you're playing against an AI.  An approach that excels against the AI may well not be effective against an experienced human opponent (obviously).  To some extent, the AI represents the basic, inexperienced human player e.g. as a beginner in any 4x strategy game, I'd probably start by building balanced fleets, balanced planet compositions and I'd fleet up as much as possible, because it initially appears like the best way to play.  Which I think is basically what your rant said - it's not that the AI's bad, it's that any AI (or inexperienced human) strategy can be countered by someone who knows the quirks of the mechanics.   

A challenging AI in this instance, then, would presumably be based on it being adaptable - being able to try a variety of different approaches within the (quirky) mechanics so that it's not as predictable.

I'd use the example of AI War: Fleet Command here, but that's almost too obvious.    I do remember reading some of Arcen's devblog about their goals behind developing the AIs for that game, and the benefits that building them around probable outcomes (rather than clearly defined outcomes) gave, in terms of keeping the player on their toes.  Sure, a certain tactic might work well 95% of the time, but 5% of the time the AI would just do something entirely unexpected and non-optimal.  Often, that would result in it losing a bunch of ships, but every so often it would trash your carefully-built defensive plan simply because it was unpredictable.  That type of chaotic adaptability (or, rather, the ability to play suboptimally at random moments) actually works remarkably well in that type of game.

[a1s]

Lets phrase it differently; The AI is challenging when you're taking a similar approach (going both research and production, using fleets and balanced ships, etc)

But that says nothing about the strength of the AI.

For example in chess you will sometimes sacrifice a weak piece to capture a strong piece.

If the AI is too stupid to do that, then of course you can make the game more challenging if you don't do it either, but that just means you're both playing sub-optimally, not that the AI is good.

The situation isn't like sacrificing a weak piece for a strong piece, or even sacrificing it for "the initiative" (which AIs have problems with). It's like, one day you sit down, read the FIDE handbook and discover that as long as you don't directly touch the pieces, you can make an opponent move with the piece you want, by physically taking him by the hand and making him touch that piece. No one else does that, and of course they lose to you (at least people who are slightly above your level do.) Are they stupid because of it? Or should all chess matches become arm wrestling matches?

[Rolan7]

Those game mechanics sound broken, and should be fixed.  The AI isn't broken for not exploiting the mechanics.  I'm slightly disappointed the AI doesn't detect those optimal strategies dynamically and use them, but only because I have a high opinion of GalCiv AI from the OS/2 days.