We don't need an AI that is hard to beat...

Thread: We don't need an AI that is hard to beat...

  1. ray243's Avatar

    ray243 said:

    Default We don't need an AI that is hard to beat...

    We need an AI that is able to behave in a logical or rational manner.

    I mean, it is easy to create an AI that is hard to beat by giving the AI some bonus and "cheats", but that does not make a game fun when you have an AI that does not even behave in a rational manner. I mean, in online battles, I do not enjoy a game where I am forced to defeat cheaters after cheaters.

    I don't necessary need an AI that is near impossible to defeat due to the AI having something like morale bonus and all that, nor do I enjoy having a hard time on the campaign map due to the AI being extremely aggressive. Making the AI aggressive makes the game tougher to win, but that does not mean the game is more enjoyable.


    I don't mind having an AI that might be easier to beat, as long as it can "think" in a logical manner.
     
  2. uanime5's Avatar

    uanime5 said:

    Default Re: We don't need an AI that is hard to beat...

    Creating an AI that acts logically and rationally is very difficult and probably unhistorical (many rulers were neither logical or rational). By contrast making an AI that has bonuses or is more aggressive is far easier to make, so that's probably why it is so common.

    Technically speaking a logical and rational AI would play defensively until it had an overwhelming technical and numerical advantage; then it would attack the closest, weakest faction; then once they've been conquered go back on the defence. Only an irrational AI would repeated attack the player (unless the player was much weaker than the AI).
    Morning Sun (adds Korea and China to the Shogun 2 map)
    http://www.twcenter.net/forums/forum...28-Morning-Sun

    Expanded Japan mod (97 new regions and 101 new factions)
    http://www.twcenter.net/forums/showt...ew-factions%29

    How to split a region in TWS2
    http://www.twcenter.net/forums/showt...split-a-region

    Eras Total Conquest 2.3 (12 campaigns from 970-1547)
     
  3. snuggans's Avatar

    snuggans said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by uanime5 View Post
    Creating an AI that acts logically and rationally is very difficult and probably unhistorical (many rulers were neither logical or rational).
    that has to be the worst defense for all the ridiculous things i've seen the AI do.

    the unhistorical part, that is.
     
  4. kekeke said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by uanime5 View Post
    Creating an AI that acts logically and rationally is very difficult and probably unhistorical (many rulers were neither logical or rational)

    Are you a lawyer? God damn that is pathetic argumentation.
     
  5. priam11's Avatar

    priam11 said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by kekeke View Post
    Are you a lawyer? God damn that is pathetic argumentation.
    Actually if he is then it is perfect statement.

    Son..you are going places in the world of public relations.
     
  6. Humble Warrior's Avatar

    Humble Warrior said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by uanime5 View Post

    Technically speaking a logical and rational AI would play defensively until it had an overwhelming technical and numerical advantage; then it would attack the closest, weakest faction; then once they've been conquered go back on the defence. Only an irrational AI would repeated attack the player (unless the player was much weaker than the AI).
    Completely and quite utterly wrong. Surely, you must know this even as you write this poppycock! You talk like you know but you obviously haven`t even thought about it.

    A logical AI would have several Logical options depending on the situation.

    1. If the AI sees the Player\rival faction committed to other wars, it may elect to attack straightaway while the Player is committed and weakened on other fronts. If it attacks hard and fast, the Player will not have time or enough troops to defend. The attacking AI will be able to focus ALL its troops on one objective and win.

    2. If it can create an Alliance with two or three other nations against its Rival then it can attack straightaway, not needing to wait to build up numbers because it has support from allies.

    3. It may see a threat (perhaps the Player) heading straight for his country and failing diplomacy, decide to attack and fight straightaway rather than wait for the player to get to it. Attack is often the best defence - throw off the Player.

    There are many other possible logical options.

    Sitting back and building up troops until it is strong enough is only ONE possible logical option and only good if the player and rival factions are not also sitting back and building up mighty armies. Otherwise all you`ll get is the same situation, just with bigger armies.

    In fact just sitting back and hoping to build itself up would set up for an immediate attack by anyone who actually thinks aggressively and is not afraid to take a risk and attack, especially if they get allies too.


    The main point is a good CAI would have MANY Logical choices depending on the situation. Not just what you think it would do.
     
  7. uanime5's Avatar

    uanime5 said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by Humble Warrior View Post
    1. If the AI sees the Player\rival faction committed to other wars, it may elect to attack straightaway while the Player is committed and weakened on other fronts. If it attacks hard and fast, the Player will not have time or enough troops to defend. The attacking AI will be able to focus ALL its troops on one objective and win.
    This isn't logical. Unless the player only has a few provinces they will not be defeated from a sudden attack and by playing defensively they can counter the assault with a prolonged siege war to buy some time to produce more soldiers. Then the AI is locked into a long war with the player over several cities (though it may be possible to win the war).

    Also if the player is much larger than the AI after the initial attack the player will be able to rebuild their army and overwhelm the AI (such as Napoleon vs the Ottoman Empire). I seem to recall a lot of complaints by players when small factions were attacking their empires even though they had no chance of winning. In short this strategy is only useful against factions of a similar size but useless if there is a large difference between them.

    2. If it can create an Alliance with two or three other nations against its Rival then it can attack straightaway, not needing to wait to build up numbers because it has support from allies.
    The success / failure of this will depend upon the diplomacy options in NTW but while you probably will be able to gather support if the Rival attacks you your allies may be less willing to attack the Rival because they don't want to fight a long war which may not be beneficial for them. Assuming your allies will join you in a war and won't be bought off by your Rival is very often not logical.

    3. It may see a threat (perhaps the Player) heading straight for his country and failing diplomacy, decide to attack and fight straightaway rather than wait for the player to get to it. Attack is often the best defence - throw off the Player.
    Alternatively they could play defensively because the invader won't be able to conquer their land without taking their cities and by forcing the invader into a long siege war (expensive for the invader) they can buy time to train reinforcements.

    Sitting back and building up troops until it is strong enough is only ONE possible logical option and only good if the player and rival factions are not also sitting back and building up mighty armies. Otherwise all you`ll get is the same situation, just with bigger armies.
    This problem only occurs if the factions are all the same size and start at the same development level. If one is larger or starts with a more modern army they will be able to build up a defensive army far more rapidly than the other and begin conquering the weaker nations.

    In fact just sitting back and hoping to build itself up would set up for an immediate attack by anyone who actually thinks aggressively and is not afraid to take a risk and attack, especially if they get allies too.
    If you can get allies who are willing to attack another player then this will work but one faction using some of its army to attack another faction that has built up its defence usually ends badly for the attacker.

    The main point is a good CAI would have MANY Logical choices depending on the situation. Not just what you think it would do.
    The most logical choice is often defend you cities and don't engage in pointless wars. The CAI should only attack a rival faction if they have a good chance of winning when fighting against a rival faction and the rival faction's allies, otherwise defensive is the best strategy.


    Personally I'd like to see an AI that's based upon the abilities of the faction leader and their generals. For example if the King is a moron and the general have bought their ranks their AI decisions will be poor because the decision makers are idiots. By contrast an AI faction with better leaders and generals will make better decisions.
    Morning Sun (adds Korea and China to the Shogun 2 map)
    http://www.twcenter.net/forums/forum...28-Morning-Sun

    Expanded Japan mod (97 new regions and 101 new factions)
    http://www.twcenter.net/forums/showt...ew-factions%29

    How to split a region in TWS2
    http://www.twcenter.net/forums/showt...split-a-region

    Eras Total Conquest 2.3 (12 campaigns from 970-1547)
     
  8. Humble Warrior's Avatar

    Humble Warrior said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by uanime5 View Post
    This isn't logical. Unless the player only has a few provinces they will not be defeated from a sudden attack and by playing defensively they can counter the assault with a prolonged siege war to buy some time to produce more soldiers. Then the AI is locked into a long war with the player over several cities (though it may be possible to win the war).

    Also if the player is much larger than the AI after the initial attack the player will be able to rebuild their army and overwhelm the AI (such as Napoleon vs the Ottoman Empire). I seem to recall a lot of complaints by players when small factions were attacking their empires even though they had no chance of winning. In short this strategy is only useful against factions of a similar size but useless if there is a large difference between them.



    The success / failure of this will depend upon the diplomacy options in NTW but while you probably will be able to gather support if the Rival attacks you your allies may be less willing to attack the Rival because they don't want to fight a long war which may not be beneficial for them. Assuming your allies will join you in a war and won't be bought off by your Rival is very often not logical.



    Alternatively they could play defensively because the invader won't be able to conquer their land without taking their cities and by forcing the invader into a long siege war (expensive for the invader) they can buy time to train reinforcements.



    This problem only occurs if the factions are all the same size and start at the same development level. If one is larger or starts with a more modern army they will be able to build up a defensive army far more rapidly than the other and begin conquering the weaker nations.



    If you can get allies who are willing to attack another player then this will work but one faction using some of its army to attack another faction that has built up its defence usually ends badly for the attacker.



    The most logical choice is often defend you cities and don't engage in pointless wars. The CAI should only attack a rival faction if they have a good chance of winning when fighting against a rival faction and the rival faction's allies, otherwise defensive is the best strategy.


    Personally I'd like to see an AI that's based upon the abilities of the faction leader and their generals. For example if the King is a moron and the general have bought their ranks their AI decisions will be poor because the decision makers are idiots. By contrast an AI faction with better leaders and generals will make better decisions.
    Unbelievable. You somehow managed to respond to everything I said and THEN still ended up saying that the most logical choice is to defend when it is not as I described, it depends on the SITUATION presented. You looked, you saw, you respond, yet you still came out with with the same ridiculous conclusion.

    I`m finished discussing with you as it is pointless.
     
  9. ray243's Avatar

    ray243 said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by Humble Warrior View Post
    Unbelievable. You somehow managed to respond to everything I said and THEN still ended up saying that the most logical choice is to defend when it is not as I described, it depends on the SITUATION presented. You looked, you saw, you respond, yet you still came out with with the same ridiculous conclusion.

    I`m finished discussing with you as it is pointless.
    I think what he is trying to say is that it is a logical conclusion that war should be avoided unless all other means to a problem has been exhausted.

    However, the mistake he did made was confusing the difference between making a decision to declare war and make a decision during a war.
     
  10. eregost's Avatar

    eregost said:

    Default Re: We don't need an AI that is hard to beat...

    How much did CA line your pockets for you to post this Ray?
    Rep me and leave your name and I promise on pain of death to rep you back!

    JOIN THE RTW ALL HUMAN MULTIPLAYER HOTSEAT CAMPAIGN HERE!
     
  11. Abspara's Avatar

    Abspara said:

    Default Re: We don't need an AI that is hard to beat...

    I just want to face an AI which makes me believe that it's operating in a logical manner. It doesn't have to be overpowering or soaked with bonuses to compensate for it's deficiencies.
    The extra development time will allow us to finalize and polish Empire, making it the most accomplished and epic of the Total War series." said Kieran Brigden, Studio Communications Manager at The Creative Assembly. "There is a great deal of anticipation around Empire: Total War and we want to ensure that it is the benchmark for strategy games upon its release.
     
  12. Humble Warrior's Avatar

    Humble Warrior said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by Abspara View Post
    I just want to face an AI which makes me believe that it's operating in a logical manner. It doesn't have to be overpowering or soaked with bonuses to compensate for it's deficiencies.
    Same.

    In fact a realistic, Human like AI would be easier to fight against anyone who actually knows battle strategy and hard for those who don`t. All that would be needed is to throw in an added counter option for the AI so it can recognise traps and counter attack and it would make the game at least immersive.

    But wait, this is what CA lied to us all before release would be in ETW- and it wasn`t, at all. Ever.

    They had one programmer who ran away, so they lied to cover up so they could still sell a half-complete game.
     
  13. Kuarlos said:

    Default Re: We don't need an AI that is hard to beat...

    Alls it needs to do is get a ton of good units and attack you with them.

    Simples.

    ETW ai was capable of neither.
    89% of people have suddenly decided it is cool to make an emo statment about the world, tag on some made up stats and ask other people to copy and paste it in what is clearly an effort to really annoy me. If you are one of the 11% of people who does this, please dont copy and paste this into your profile, just stop it.

    And if you are one of these Magic the Gathering people, who think a silly litle colour test is a relevant thing to tell the world; I am going to find your plains backside, swamp your eyes till islands come out and you go forest and are sick then beat you till there are mountains all over the shop.
     
  14. Ulan1990 said:

    Default Re: We don't need an AI that is hard to beat...

    well i don't expect an ai which is hard to beat just an ai who is capable of using basic tactics and reacts to your moves properly plus an artillery ai which is at least capable of not shooting canister shot in your own ranks, hitting the enemy and only shoot when the enemy is within range
     
  15. Greve Af Göteborg's Avatar

    Greve Af Göteborg said:

    Default Re: We don't need an AI that is hard to beat...

    I would like to see logical AI over hard AI and the way the AI acts should depend on who's leading the nation.
    If the Leader is financial & defensive, then he would build up his infrastructure and economy and put his military efforts into defending his lines.
    And if the leader is agressive & expansive he would invade weaker nations and colonize to expand his own nation.

    Kind of like the AI in Civilization IV.
     
  16. eleftherios said:

    Default Re: We don't need an AI that is hard to beat...

    The best CAI & BAI was the AI of M1TW:
    Every nation's decisions toward your faction had to do:
    1)With your Leader's influence (agents worldwide, prestige through military/technological achievements, trade routes, allies, e.t.c.).
    2)Your religion (Orthodox,Catholic,Muslim), especially with those Jihads/Crusades.
    3)You honoring of treaties (e.g. no execution of prisoners/no wars without just cause like being attacked or taking the land from rebels or crusades/Jihad).
    4)Possible acts of espionage especially framing from the opposition to "cultivate" a coalition against you.
    5)Their Generals level of stars (in management, loyalty, military dread or pity since the letter two can't coexist).

    I believe that in order to make a good AI one must have many parameter to work with. I also believe that the battle AI started having issues when with the introduction of certain mechanics like the fact that units with many stars in M1TW such as generals were almost immortals and in RTW cavalry and other units to a degree, could leave combat at will plus the fact that they started giving more range to missile cavalry than missile troops. My opinions coincide with the creator of Darthmod on this matter. I mean How can the AI weigh it's decisions properly if the mechanics favor such a unrealistic behavior?

    Extra effective enemy generals that might do one of the following:a) More income b) Interesting tactical options through their unique abilities c) Different diplomatic options depending on their faction to force the player to make a unique approach to each situation.

    So to make my answer clear: GIVE ME A HARD AI IN THOSE PARAMETERS NOT IN CHEATING, AS THIS WONT GIVE THE GAME THE REALISTIC IMMERSION IT DESERVES.
     
  17. ray243's Avatar

    ray243 said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by eleftherios View Post
    The best CAI & BAI was the AI of M1TW:
    Every nation's decisions toward your faction had to do:
    1)With your Leader's influence (agents worldwide, prestige through military/technological achievements, trade routes, allies, e.t.c.).
    2)Your religion (Orthodox,Catholic,Muslim), especially with those Jihads/Crusades.
    3)You honoring of treaties (e.g. no execution of prisoners/no wars without just cause like being attacked or taking the land from rebels or crusades/Jihad).
    4)Possible acts of espionage especially framing from the opposition to "cultivate" a coalition against you.
    5)Their Generals level of stars (in management, loyalty, military dread or pity since the letter two can't coexist).

    I believe that in order to make a good AI one must have many parameter to work with. I also believe that the battle AI started having issues when with the introduction of certain mechanics like the fact that units with many stars in M1TW such as generals were almost immortals and in RTW cavalry and other units to a degree, could leave combat at will plus the fact that they started giving more range to missile cavalry than missile troops. My opinions coincide with the creator of Darthmod on this matter. I mean How can the AI weigh it's decisions properly if the mechanics favor such a unrealistic behavior?

    Extra effective enemy generals that might do one of the following:a) More income b) Interesting tactical options through their unique abilities c) Different diplomatic options depending on their faction to force the player to make a unique approach to each situation.

    So to make my answer clear: GIVE ME A HARD AI IN THOSE PARAMETERS NOT IN CHEATING, AS THIS WONT GIVE THE GAME THE REALISTIC IMMERSION IT DESERVES.

    If this is the case, then we cannot expect CA to create a proper AI. If the people who wrote the AI understand less about the mechanics of the game than the fans themselves, how can they even write a decent AI?
     
  18. uanime5's Avatar

    uanime5 said:

    Default Re: We don't need an AI that is hard to beat...

    Quote Originally Posted by Humble Warrior View Post
    Unbelievable. You somehow managed to respond to everything I said and THEN still ended up saying that the most logical choice is to defend when it is not as I described, it depends on the SITUATION presented. You looked, you saw, you respond, yet you still came out with with the same ridiculous conclusion.

    I`m finished discussing with you as it is pointless.
    Pointing out the flaws in your argument is part of a debate and throwing a hissy fit because someone disagreed with you is not a mature way to respond to it.

    Quote Originally Posted by ray243 View Post
    I think what he is trying to say is that it is a logical conclusion that war should be avoided unless all other means to a problem has been exhausted.

    However, the mistake he did made was confusing the difference between making a decision to declare war and make a decision during a war.
    I'm saying that the AI should avoid wars unless they can quickly win because a long war will be expensive and weaken the AI faction. The AI should not start wars it has no chance of winning.

    What is the difference between a decision to declare war and make a decision during a war?
    Morning Sun (adds Korea and China to the Shogun 2 map)
    http://www.twcenter.net/forums/forum...28-Morning-Sun

    Expanded Japan mod (97 new regions and 101 new factions)
    http://www.twcenter.net/forums/showt...ew-factions%29

    How to split a region in TWS2
    http://www.twcenter.net/forums/showt...split-a-region

    Eras Total Conquest 2.3 (12 campaigns from 970-1547)
     
  19. Evan MF's Avatar

    Evan MF said:

    Default Re: We don't need an AI that is hard to beat...

    I blame SEGA, they don't let CA work on the AI. All they care about is the graphics,,,

    Evan
     
  20. ♔ST0MPA♔'s Avatar

    ♔ST0MPA♔ said:

    Default Re: We don't need an AI that is hard to beat...

    Ita a pity we cant choose the AI type at the start. Imagine being able to choose an AI for a faction
    1. Agressive:will seek to expand its territory at the cost of properlly securing others
    2. Defensive: will gaurd all territory but be reluctant to attack unless provoked
    3.Comercial: will establish a commercial empire, use spys to cripple other nations and gentlemen to steal commercial techs
    4. Aggresive Commercial: Attack trade routes and profitable colonies to try and bankrupt nations bringing them to their knees to gain their territories through peace treaties or protectorates (would be good to be able to adjust the amount of tax they pay wich would go into your coffers rather than their own)
    5. Defensive Commercial: Protect their trade routes. ports and prize colonies through diplomatic or military means.
    "Yeah tho I walk through the valley of the shadow of death; I shall fear no evil... for I'm the meanest sonofa in the valley."