New Game AI Poll
Author
Hello Everybody:
I took time off from launching fireworks and working on my Suburban (I don''t *just* do computer stuff) to put up a new poll for July. This one, like last month''s, was inspired by a fascinating round of conversation we had at the GDC 2000 AI roundtables (minutes of which you can read at the site as well).
The topic this month: Would a really great AI make a game a hit? Would a game with killer AI really sell better?
By and large, developers present at the roundtables seemed to think that it might *help*, but that AI by itself probably isn''t a strong enough pull to make a game a hit. What do you think?
--
Steven Woodcock
From the High Mountains of Colorado
ferretman@gameai.com
http://www.gameai.com
Well, the Creatures-series are definitely a hit and they feature nothing but AI. So it is possible to make a hitgame based just on strong AI.
My personal view on the more regular games like FPS and RTS is that people will not look at how good the AI is when they buy a game, because that is one of the things you just can''t know before you''ve actually bought the game. The only info you have then is a review and screenshots. Even if you have a demo it''ll only give you a few basic and easy levels which are not good enough to judge the AI on.
I do think that if the producers aim at making a sequel to the game, they should have good AI in the first edition because then people will be more likely to say: "Hey, game X had crappy AI, so I won''t buy X2, even if the box says (and it always does when the AI sucked in the original) ''Improved AI!''"
******************************
Stefan Baert
On the day we create intelligence and consciousness, mankind becomes God.
On the day we create intelligence and consciousness, mankind becomes obsolete...
******************************
My personal view on the more regular games like FPS and RTS is that people will not look at how good the AI is when they buy a game, because that is one of the things you just can''t know before you''ve actually bought the game. The only info you have then is a review and screenshots. Even if you have a demo it''ll only give you a few basic and easy levels which are not good enough to judge the AI on.
I do think that if the producers aim at making a sequel to the game, they should have good AI in the first edition because then people will be more likely to say: "Hey, game X had crappy AI, so I won''t buy X2, even if the box says (and it always does when the AI sucked in the original) ''Improved AI!''"
******************************
Stefan Baert
On the day we create intelligence and consciousness, mankind becomes God.
On the day we create intelligence and consciousness, mankind becomes obsolete...
******************************
I am currently in a quandry about buying Vampire the Masquerade simply because it''s meant to have terrible AI and DaiKatana is off a lot of people''s ''to buy'' lists partially because of terrible AI.
Fantastic AI may not (although it should, and in the case of certain games, does) sell more copies than merely competent AI but bad AI can destroy a game completely.
When I loaded up Tiberian Sun for the first time and the harvesters still tended to go to a bay half way across the map through enemy territory to drop their harvest rather than wait 20 seconds on a nearby off-load point I turned it off and didn''t play it again.
Mike Ducker
On a soapbox
Somewhere in South London.
Fantastic AI may not (although it should, and in the case of certain games, does) sell more copies than merely competent AI but bad AI can destroy a game completely.
When I loaded up Tiberian Sun for the first time and the harvesters still tended to go to a bay half way across the map through enemy territory to drop their harvest rather than wait 20 seconds on a nearby off-load point I turned it off and didn''t play it again.
Mike Ducker
On a soapbox
Somewhere in South London.
My opinion is that a new brilliant AI for a game would drive a game to hitdom if the game was foundated at the use of AI. That is, the game would have to be very AI orientated and it would come down to how you make use of this brilliant AI.
Examples of making use of the AI would be:
1. Softening the Learning Curve of the game by using ai in a creative and helpful way to the player.
2. Using the ai for more than NPC's. Like using it for Random Encounter Tables, constructing maps, etc
3. Having the game heavily NPC orientated
4. Using it to make the controls of the game more fun.
These are just a few off the top of my head.
Wrap up... Generally what i think is that if you where to say that what makes a game is the fun factor then the AI would have to have a truely significant impact on this (The Fun Factor)!
We are their,
"Children of the Free"
Edited by - Paul Cunningham on July 3, 2000 10:00:03 AM
(Game Designer by Love)
Edited by - Paul Cunningham on July 3, 2000 10:05:54 AM
Examples of making use of the AI would be:
1. Softening the Learning Curve of the game by using ai in a creative and helpful way to the player.
2. Using the ai for more than NPC's. Like using it for Random Encounter Tables, constructing maps, etc
3. Having the game heavily NPC orientated
4. Using it to make the controls of the game more fun.
These are just a few off the top of my head.
Wrap up... Generally what i think is that if you where to say that what makes a game is the fun factor then the AI would have to have a truely significant impact on this (The Fun Factor)!
We are their,
"Children of the Free"
Edited by - Paul Cunningham on July 3, 2000 10:00:03 AM
(Game Designer by Love)
Edited by - Paul Cunningham on July 3, 2000 10:05:54 AM
Hi
I hate to sound the over-optimist but I believe outstanding AI could sell a game, if implemented correctly. Imagine a multiplyer fps where if you treat/train them correctly AI pets/weapons could save your bacon from other players. I''m a big TFC fan and often find myself glowing with pride when my sg mows down half an enemy team.
Take that sensation from building to training, in offline mode and you could really have a hit - look a Pokemon. The game need not revolve round the AI, but with a clever pet watching your back against real opponents I''d say you''d have a big, compulsive hit.
Lord[BRIT]
I hate to sound the over-optimist but I believe outstanding AI could sell a game, if implemented correctly. Imagine a multiplyer fps where if you treat/train them correctly AI pets/weapons could save your bacon from other players. I''m a big TFC fan and often find myself glowing with pride when my sg mows down half an enemy team.
Take that sensation from building to training, in offline mode and you could really have a hit - look a Pokemon. The game need not revolve round the AI, but with a clever pet watching your back against real opponents I''d say you''d have a big, compulsive hit.
Lord[BRIT]
Author
Lord[Brit]: I hate to sound the over-optimist but I believe outstanding AI could sell a game, if implemented correctly. Imagine a multiplyer fps where if you treat/train them correctly AI pets/weapons could save your bacon from other players. I''m a big TFC fan and often find myself glowing with pride when my sg mows down half an enemy team.
We''re in full agreement here. I personally think RPGs are the most fertile ground for us to see some interesting AI, and (probably) the area developers will most likely be able to get away with trying something daring. It''s hard to get much unusual done in a standard RTS game unless you''ve really got the support of the producer, since "everybody knows" what goes into an RTS game.
Ferretman
ferretman@gameai.com
http://www.gameai.com
From the High Mountains of Colorado
We''re in full agreement here. I personally think RPGs are the most fertile ground for us to see some interesting AI, and (probably) the area developers will most likely be able to get away with trying something daring. It''s hard to get much unusual done in a standard RTS game unless you''ve really got the support of the producer, since "everybody knows" what goes into an RTS game.
Ferretman
ferretman@gameai.com
http://www.gameai.com
From the High Mountains of Colorado
i agree too - if we are talking about good application of good AI.
i have long thought the games industry needs more people like peter moulineux (populous, dungeon keeper, black & white) who try things a little different. we need more games that are INDIRECT - you change factors that influence the citizens... C&C and all the other RTS games place you in total control of every aspect of your little empire. quite frankly, it''s dull, but if you could ''grow'' your empire much like simcity... it would be much more interesting...
but i don''t have commercial experience so i don''t know. i just know that i would like to play a game where i interact with ''my people'', not the save/load buttons.
i have long thought the games industry needs more people like peter moulineux (populous, dungeon keeper, black & white) who try things a little different. we need more games that are INDIRECT - you change factors that influence the citizens... C&C and all the other RTS games place you in total control of every aspect of your little empire. quite frankly, it''s dull, but if you could ''grow'' your empire much like simcity... it would be much more interesting...
but i don''t have commercial experience so i don''t know. i just know that i would like to play a game where i interact with ''my people'', not the save/load buttons.
If we could ever accomplish great ai, that would certainly put a halt on the growing population of only online games. Everyone tends to go to straight multiplayer, because playing the regular game tends to get too boring for any good purposes. If we were to make a game so realistic that the characters actually interact with each other based on a growing dictionary of terms, this would make a game historic. What I mean by this, is not only should the game ai be very realistic it should also contain the certain jerks that everyone hate and the one guy that everyone wants to beat. Just like in TF. The reason TF is so popular in my opinion, is because if you couldn''t beat someone you would resort to insulting the shit out of them, which started a heated battle. This is what we need in our ai games. The enemies should be able to hold a somewhat good conversation with the user.
With that, I propose that we have like a little contest to see who could create the best bot to respond in a textual environment. I think I have heard of this before, but I never really seen it. But how it works, is that everyone programs for a certain standard, and we toss all the bots into a room to see how they would interact with the other bots. Sort of like a little private bot chat.
With that, I propose that we have like a little contest to see who could create the best bot to respond in a textual environment. I think I have heard of this before, but I never really seen it. But how it works, is that everyone programs for a certain standard, and we toss all the bots into a room to see how they would interact with the other bots. Sort of like a little private bot chat.
quote:Original post by lxnyce
If we could ever accomplish great ai, that would certainly put a halt on the growing population of only online games. Everyone tends to go to straight multiplayer, because playing the regular game tends to get too boring for any good purposes.
Actually, this is not what ALL game players do.
The overwelmingly majority of computer game players DO NOT play multiplayer games over the internet or in networks. They play single player gaming sessions against an AI.
I know this flies in the face of all those who spend time (like I do) on the internet and in multiplayer games, but it is reality.
Eric
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
