Fight club (AI's)
Moderators: hoijui, Moderators
- 1v0ry_k1ng
- Posts: 4656
- Joined: 10 Mar 2006, 10:24
Fight club (AI's)
NTAI, AAI, KAI. which is superior, at what mods, and why? my experience;
KAI currently owns due to very agressive behavior, and wins via and early game push
AAI isnt very hot early game but if allowd to get upto speed, develops an economy which by the 20 minutes mark is producing 2-3x the metal and E of KAI, and wins (despite lack of unit management or radarhack) through sheer numberage.
NTAI is currently a bit of an underdog, compatible with most the mods and bundled free but no match for KAI or AAI atm
In short KAI>AAI>NTAI is what Ive observed so far..
KAI currently owns due to very agressive behavior, and wins via and early game push
AAI isnt very hot early game but if allowd to get upto speed, develops an economy which by the 20 minutes mark is producing 2-3x the metal and E of KAI, and wins (despite lack of unit management or radarhack) through sheer numberage.
NTAI is currently a bit of an underdog, compatible with most the mods and bundled free but no match for KAI or AAI atm
In short KAI>AAI>NTAI is what Ive observed so far..
- Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
- 1v0ry_k1ng
- Posts: 4656
- Joined: 10 Mar 2006, 10:24
yeah fair enough KAI is haxtastic with its abailty to radar-see the whole map, but that cant be compared to NTAI's cheat mode in any way cuz NTAI's cheatmode is like terminator let loose in the creche. AAI is good for a non-cheating, but even if it has full map coverage, it wont be as a good as KAI atm because of KAIs defence placement and unit micro.. AAI is fascinating though.. learning ftw. NTAI is broke atm and isnt making gramps proud.. im testing using XTA, SWTA and EE
-
- Posts: 665
- Joined: 06 Jun 2006, 19:49
Anyway to have a seperate command other than .aicheat i.e .aifightclub
So that all ai's could implentment the same cheat i.e See everything
At very least be nice to see how ai's stack against each other on fair fighting terms. Might help show ai dev's where there ai excels at & where it could do with some work.
So that all ai's could implentment the same cheat i.e See everything
At very least be nice to see how ai's stack against each other on fair fighting terms. Might help show ai dev's where there ai excels at & where it could do with some work.
It might be fun, but isn't strictly necessary. I mean, the AI's are designed with different things in mind. KAI is designed with difficult gameplay #1, over fairness. Does this decrease it's merit as an AI? Maybe, if you define an AI as an interesting project in remote intelligence (which, IMO, is a valid definition if you want to go that route). But not at all if you define an AI as something that beats you at Spring.
Besides, if we had a perfectly even way to compare them, all the newbs would just download whatever "AI of the month" was ahead that month and never try the rest. As we have already learned with mods, that kind of monopoly takes more work to break out of than it does to get into.
For example, I'd hate to see player support for NTai dwindle while the nanostall mechanism is screwed up. NTai is not currently the "best" AI at playing a particular mod, but it has a boatload of unique features that I would hate to see lost.
For another, I'd hate to see KAI become the only thing played. Because what would that do to work going toward better scouting algorithms? KAI doesn't need it, but that doesn't mean it isn't worth working on.
Note that the reverse is also true: I'd hate to see KAI dissapear, since it has such nice defense placement and troop micro algorithms.
Besides, if we had a perfectly even way to compare them, all the newbs would just download whatever "AI of the month" was ahead that month and never try the rest. As we have already learned with mods, that kind of monopoly takes more work to break out of than it does to get into.
For example, I'd hate to see player support for NTai dwindle while the nanostall mechanism is screwed up. NTai is not currently the "best" AI at playing a particular mod, but it has a boatload of unique features that I would hate to see lost.
For another, I'd hate to see KAI become the only thing played. Because what would that do to work going toward better scouting algorithms? KAI doesn't need it, but that doesn't mean it isn't worth working on.
Note that the reverse is also true: I'd hate to see KAI dissapear, since it has such nice defense placement and troop micro algorithms.
KAI style competition is an extremely stifling aspect of this AI community.
That style of comeptition takes new AI projects by the throat with hydraulic clamps and crushes them into tiny cubes.
I am aware of a few small AI projects that have the potential to become large but darent put themselves out into the AI forum because they'd get trampled.
I've seen more enthusiasm in my Epic think tank about new AI projects in the last week than I have in this forum in the last half year.
That style of comeptition takes new AI projects by the throat with hydraulic clamps and crushes them into tiny cubes.
I am aware of a few small AI projects that have the potential to become large but darent put themselves out into the AI forum because they'd get trampled.
I've seen more enthusiasm in my Epic think tank about new AI projects in the last week than I have in this forum in the last half year.
recently played some ffa's with ntai, kai and aai
i spect 3 games with xta mod.
first game aai got seriously ownd by kai, but ntai won this game by outnumbering kai and air superiority...
for the second game i changed the config for aai for xta a little bit, so that he prefers metal more...
and then he was very good... he spammed out samsons and jethros and defeated kai. aai also built some stunners and destroyed very successfully troop concentrations of kai. ntai started with an aircraft plant... that was not a good choice cuz of the mass-samsons
third game kai went crazy. he built mass radars!! maybe the radar-hack didnt worked :D
(i have a screen from this game, but i cant upload it cuz the forum says im not allowed to...)
seriously kai won this game by micro managing 4 (!!) Raven's (heavy rocket kbot).
currently i think its just luck which one of the ai's wins... but kai got an unfair advantage...
i spect 3 games with xta mod.
first game aai got seriously ownd by kai, but ntai won this game by outnumbering kai and air superiority...
for the second game i changed the config for aai for xta a little bit, so that he prefers metal more...
and then he was very good... he spammed out samsons and jethros and defeated kai. aai also built some stunners and destroyed very successfully troop concentrations of kai. ntai started with an aircraft plant... that was not a good choice cuz of the mass-samsons
third game kai went crazy. he built mass radars!! maybe the radar-hack didnt worked :D
(i have a screen from this game, but i cant upload it cuz the forum says im not allowed to...)
seriously kai won this game by micro managing 4 (!!) Raven's (heavy rocket kbot).
currently i think its just luck which one of the ai's wins... but kai got an unfair advantage...
- Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
Its just a debug thing to show that the builder wants to make a building it cant (eg farks cant make factories, so they do that). It is known and more or less fixed nowCrampman wrote:I've seen once this radar-loving bug. Very strange, indeed. I didn't take screenies, but I may still have the replay...
- 1v0ry_k1ng
- Posts: 4656
- Joined: 10 Mar 2006, 10:24
AF, this topic is meant in a good natured way, just seeing which AI fights best AI vs AI. Its not entirely fair as you say with KAI cheating, but then KAI's map hack dosnt improve its combat abilities THAT much.. it cant fire accuratly at distant targets etc..
AAI's main fault (XTA) is that it takes so long to get upto speed in comparison to KAI.. KAI whacks out units from the moment go and AAI focus's on const units and eco, which is cool for big maps but most the time KAI vs AAI means that KAI steals 3/4 of the territory at the start and keeps AAI squashed and eventually wins through bombardment (very cool usage of frontline popups and merls).
KAI, as well as an agressive frontline push, also builds defences on its frontline, as opposed to just round its base AAI style, meaning AAI cant really make an effective push until late game where its too late..
AAI has an awesome economy system though, probs the best around at the moment and techs up its whole base as the game progresses. one main prob with KAI is it usualy only uses 2-3 factories.. AAI builds THEM ALL by the end, pumping out dirty hordes of units.
its not like wanting the most competetive AI is crushing the little guy at all.. sure, after seeing KAI those programmers who decided their AI would never catch up gave up, but the programmers who beleive their AI will own in the future will still be going.. KAI is good but with improved unit control and intelligence in attacking for AAI and a more efficent economy and defence placement, it isnt unassailable..
Incidently, a good game for SP xta is player and 2 aais (30%handicap) VS 3 kais (50% handicap) on a map like CC.. not impossible, but entertaining.. pushing foward and curshing one AI may work, but the AAIs will be owned by the two other KAI's while you do so and you will be swampwed.. the key is to help protect your allies until their economys gwet to speed.
so yeah. AI fights FTW
AAI's main fault (XTA) is that it takes so long to get upto speed in comparison to KAI.. KAI whacks out units from the moment go and AAI focus's on const units and eco, which is cool for big maps but most the time KAI vs AAI means that KAI steals 3/4 of the territory at the start and keeps AAI squashed and eventually wins through bombardment (very cool usage of frontline popups and merls).
KAI, as well as an agressive frontline push, also builds defences on its frontline, as opposed to just round its base AAI style, meaning AAI cant really make an effective push until late game where its too late..
AAI has an awesome economy system though, probs the best around at the moment and techs up its whole base as the game progresses. one main prob with KAI is it usualy only uses 2-3 factories.. AAI builds THEM ALL by the end, pumping out dirty hordes of units.
its not like wanting the most competetive AI is crushing the little guy at all.. sure, after seeing KAI those programmers who decided their AI would never catch up gave up, but the programmers who beleive their AI will own in the future will still be going.. KAI is good but with improved unit control and intelligence in attacking for AAI and a more efficent economy and defence placement, it isnt unassailable..
Incidently, a good game for SP xta is player and 2 aais (30%handicap) VS 3 kais (50% handicap) on a map like CC.. not impossible, but entertaining.. pushing foward and curshing one AI may work, but the AAIs will be owned by the two other KAI's while you do so and you will be swampwed.. the key is to help protect your allies until their economys gwet to speed.
so yeah. AI fights FTW
AF, don't take things personal here. You've done a hell of a job, and we're thankfull for that! Now all this thread is about, is comparising AI's, pure by performance. For us users, as fun.
By reading this I get the feeling AAI can perform much better, after it has learned somewhat, and configs are tweaked. Maybe someone wants to volunteer some time and optimize AAI? Then we can have a real fair comparison
By reading this I get the feeling AAI can perform much better, after it has learned somewhat, and configs are tweaked. Maybe someone wants to volunteer some time and optimize AAI? Then we can have a real fair comparison

I think that having all/most AIs competing on levle ground would really help development. Currently, you can't pit the AIs against each other (as has been said) because of the way they can cheat. By making AIs compete on as close a level as humans (ie making the same information availalbe to the AI interface as a human gets) the development of AIs would be driven by competition.
I know some people have some quite strong views on why an AI should be able to cheat vs a human, but I don't think that developing an AI to cheat from the start is a good 'habit' to be in...[/quote]
This is exactly what I would like to see - in this way, all the AI developers are pushing to develop the 'best' AI. When one comes up with an innovotive feature that enables it to regularly beat the other AIs, then the onus would be on the other AIs to find a way to counter that feature, or copy it themselves (and in the process probably improve it). As such, this would drive AIs to creating more and more 'human-like' features and eventually make them human-competetive (without cheating) - which IMHO is the real goal of general AI.Erom wrote:Besides, if we had a perfectly even way to compare them, all the newbs would just download whatever "AI of the month" was ahead that month and never try the rest. As we have already learned with mods, that kind of monopoly takes more work to break out of than it does to get into.
For example, I'd hate to see player support for NTai dwindle while the nanostall mechanism is screwed up. NTai is not currently the "best" AI at playing a particular mod, but it has a boatload of unique features that I would hate to see lost.
I know some people have some quite strong views on why an AI should be able to cheat vs a human, but I don't think that developing an AI to cheat from the start is a good 'habit' to be in...[/quote]
Last edited by Acidd_UK on 20 Jul 2006, 20:12, edited 1 time in total.