To me, SOE attempting to make a better scripted AI is a great thing. It's a step in the right direction and if they succeed at even half of what they are wanting, then other game companies will see this and follow suit.
I'm not expecting really intelligent AI. I have no knowledge of how difficult this is to program as compared to maybe single player games, but I would think there will be limitations with so many players playing at once.
depends really, you can script so much into monster ai, its very easy to overpower , or make a enemy ai nearly unbeatable. this is just an exert from 4GAI, the 4th gen aAI in Unangband, from 5 years ago, just think whatt can be done now.
A target (x,y) which the monster is heading towards.
A minimum range that the monster wants to stay away from the target.
A best range, that the monster wants to ideally be at.
A set of flags which contain some AI state information: in particular, did the player attacked me the last turn.
A set of smart flags indicating what resistances the monster has learned about the player
Monsters talk to each other, conveying information about the player position, resistances or lack of resistances.
Monsters wake up their fellows nearby when they are under attack.
Monsters calculate their moral, weather in group or single.
Single encounter monsters have exit strategies ( ie root and scoot)
monster s retain memories of player interaction if not killed.( they come after you)
Archers will share ammunition with each other, so that each monster in the group will usually have ammunition available.
Monsters will eat when they are hungry, to help them recover from injuries.
If a group of monsters is fighting the player and the player can be flanked, some will sneak off and take the back route to the player.
Added friendly monsters, which intelligently move around the player and get out of their way, and attack targets, all done without adding a single 'order a monster to do something' command.
When fighting a group of players, or group of monsters, archers and spell casters will stay at the back while warriors will move up and form a front line.
Faster monsters like wolves will attack isolated enemies, and the edges of the front line, dashing in for a bite and then retreating back.
The emergent AI they discussed doesn't seem that hard to do, the question is whether it will make the game better or not. We will use the scenario that Dave gave us about the orcs they let loose into the world.
Currently, in most games, we would be given a quest to go to the forest and kill 10 orcs. We would then go to the forest and find a bunch of orcs standing around, maybe even moving a few steps one way or another. We would then begin to kill 1 or 2 at a time until we have our 10 kills. Player #2 could also be doing the same quest and this would go on for every new character created. Each player would go to the same spot and find the same orcs in the same spot acting the same way.
With the emergent AI described, this group of orcs would leave the forest after X number of players came to kill them. In the description given, they would look for another spot similar to the first because that is the type of area they like. They would then move again and again each time they die enough times.
My question is what happens to the players as the orcs move further and further away from their starting point? You are sent to the woods to kill orcs, only to find there aren't any there? You check the next closest woods only to find there are none there either. You have to move further and further out to finish your quest. Eventually you find some orcs, but unfortunately the trolls in that area are much higher level than you and you die trying to kill the orcs. QUEST FAILED.
I know, I know. They game didn't just let 1 group of orcs out into the world. After a while the game would spawn another group near the forest to take the place of the first group. The only problem here is that even though player #1 fought a different group of orcs in the woods than player #300,593 to each player it seems like the same static spawn of orcs. Even though emergent AI worked, the end result was still the same. Every player still went to the same spot in the woods and killed 10 orcs. GAME IMPROVEMENT FAILED.
This only gets cool if after moving 5 times (and eventually being left alone by the players because they are far enough away from the quest area) the orcs are able to settle in. Over the course of a few weeks more and more orcs arrive until there are hundreds or thousands of orcs. They all start to talk to each other and realize the players from the city keep coming out and killing them. They realize that there aren't that many players at any one time, and they currently outnumber the players 100 to 1. Now 100 to 1 odds are just high enough for the orcs to overcome their fear of the guards in the city, and in one fell swoop they travel back into the woods killing every player they find. After they clear out the woods they then advance on the city and either destroy it or are wiped out in the process. GAME IMPROVEMENT SUCCESS.
They would not be "looking" for a new spot, although it might appear that way. They will move to a spot based on the parameters of the situation. This spot will already be pre-programmed, or in other words will be scripted to happen. That is why I've said countless times that what they are calling AI, isn't actually AI.
They are using, hopefully a complex script, and when I mean script, I mean scripting of events. The complex script would then give off the illusion of intelligent Orcs and would also give off the illusion that their placement is random.
I think a combination of player intelligence and AI would be best. Create tools for staff to coordinate the actions of groups of mobs. For example the staff person sees some orcs getting their butts handed to them by players and he can decide to send some reinforcements from a nearby mountain. Human intelligence makes things unpredictable and a lot more fun.
Maybe they can and maybe they can't, all I can say as a CS major is that AI tech in other industries is much better than what we had in mmo or just general games. Anyway it doesn't matter because these are all just estimations. At the end of the day, the one thing we'll remember is the final product.
Glass half full here.
As a recently graduated CS major I'll second this one. I've only taken one class on AI programming... and we only just touched on adaptive AI (with genetic and neural net algorithms)... but modern AI tech is WAY beyond the level that most games implement.
In a finite space, a 'simple' tree search algorithm can give you a surprisingly hard opponent to beat. And if they're using multiple layers of decision making algorithms... well. The problem might be tuning it down, so it doesn't kick our collective arses in casual, small-group play.
To me, SOE attempting to make a better scripted AI is a great thing. It's a step in the right direction and if they succeed at even half of what they are wanting, then other game companies will see this and follow suit.
I'm not expecting really intelligent AI. I have no knowledge of how difficult this is to program as compared to maybe single player games, but I would think there will be limitations with so many players playing at once.
depends really, you can script so much into monster ai, its very easy to overpower , or make a enemy ai nearly unbeatable. this is just an exert from 4GAI, the 4th gen aAI in Unangband, from 5 years ago, just think whatt can be done now.
A target (x,y) which the monster is heading towards.
A minimum range that the monster wants to stay away from the target.
A best range, that the monster wants to ideally be at.
A set of flags which contain some AI state information: in particular, did the player attacked me the last turn.
A set of smart flags indicating what resistances the monster has learned about the player
Monsters talk to each other, conveying information about the player position, resistances or lack of resistances.
Monsters wake up their fellows nearby when they are under attack.
Monsters calculate their moral, weather in group or single.
Single encounter monsters have exit strategies ( ie root and scoot)
monster s retain memories of player interaction if not killed.( they come after you)
Archers will share ammunition with each other, so that each monster in the group will usually have ammunition available.
Monsters will eat when they are hungry, to help them recover from injuries.
If a group of monsters is fighting the player and the player can be flanked, some will sneak off and take the back route to the player.
Added friendly monsters, which intelligently move around the player and get out of their way, and attack targets, all done without adding a single 'order a monster to do something' command.
When fighting a group of players, or group of monsters, archers and spell casters will stay at the back while warriors will move up and form a front line.
Faster monsters like wolves will attack isolated enemies, and the edges of the front line, dashing in for a bite and then retreating back.
Thanks for posting this.
"if u forcefully insert foriegn objects into my? body, i will die"
The emergent AI they discussed doesn't seem that hard to do, the question is whether it will make the game better or not. We will use the scenario that Dave gave us about the orcs they let loose into the world.
BIG SNIP
They would not be "looking" for a new spot, although it might appear that way. They will move to a spot based on the parameters of the situation. This spot will already be pre-programmed, or in other words will be scripted to happen. That is why I've said countless times that what they are calling AI, isn't actually AI.
They are using, hopefully a complex script, and when I mean script, I mean scripting of events. The complex script would then give off the illusion of intelligent Orcs and would also give off the illusion that their placement is random.
You are very likely correct. But let's suspend disbelief for a minute. Let's assume that they ARE given a simple AI and do randomly move to a new location based off of it meeting certain criteria. The location wouldn't have to be on a predetermined list because they are smart enough to find a new location on their own. This could be very similar to a herd of animals looking for a water source when their current one dries up.
The problem I was trying to point out is that even if they gave the orcs enough AI to migrate, it wouldn't make the game any more realistic to an individual player. As you said, the original spawn points are going to be set (how else would they be able to generate quests around their existence if they weren't). So even though players are actually encountering different groups of mobs, to the players it may as well be the same exact group since it is in the same exact spot.
I'd just like to point out that SOE is falsely using the word AI. It isn't AI at all. The poster above me said scripted AI, and he was partially right. They've made nothing more than a more advanced "script", something no other game developer has tried their hand at to this extent. The "script" makes the monsters look more "intelligent", that's the idea. It is good game design, but good game design does not equate to artificial intelligence. A script does not equal AI.
Every AI is scripted, even human intelligence. Learning from your mistakes is scripted, its like a script that tells you to create your own script from an outcome. Its like extremely advanced if then else statements that branches out into trillions of branches through more experience the artificial intelligence goes through. Basically like you get hurt from touching something and than your brain learns from that and makes more neurons from that new experience, and in this case our neurons are like new formed scripts are brains have written down.
Weve actually seen 5 minutes of footage of a game not even close to beta,how about we wait until its 6 months until release to jump to conclusions.Oh and btw this is coming from a EQ vet that actually wanted a shiny new old version of EQ.
Originally posted by truce12 Weve actually seen 5 minutes of footage of a game not even close to beta,how about we wait until its 6 months until release to jump to conclusions.Oh and btw this is coming from a EQ vet that actually wanted a shiny new old version of EQ.
We're talking about the realistic approach taken to create "intelligent AI" whether it being a repeated script tweeked a bit or entirely new system(script) that combines loads of different variables and parameters to depict a intelligent AI.
This is done in the early planning and development stages and is not left "6 months until release" and they are going on about their "improved, intelligent AI" so it's right to talk about it now.
"if u forcefully insert foriegn objects into my? body, i will die"
The problem I was trying to point out is that even if they gave the orcs enough AI to migrate, it wouldn't make the game any more realistic to an individual player. As you said, the original spawn points are going to be set (how else would they be able to generate quests around their existence if they weren't). So even though players are actually encountering different groups of mobs, to the players it may as well be the same exact group since it is in the same exact spot.
That's because they can tag each location with values that may or may not be attractive to the mob's AI... and NPC's AI can read those values to generate quests as well. AND those values can shift over time due to player or mob activity. In other words, if a strong group of players clear out all the wandering groups looking for food or loot, the orcs at the camp may hunker down and fortify... or call in reinforcements. If the camp is destroyed there are obviously strong players hunting in the area... so the value of the area changes and the orcs are less likely to return. Until the players stop hunting the area so aggressively and the values in the area start to return to normal, the orcs will stay away.
But maybe the gnolls or bandits will start expanding into the area first next time... they may have different environmental threshholds than orcs.
You should really watch the debut video again, where Dave describes the emergent behavior of the orcs and the village.
Originally posted by truce12 Weve actually seen 5 minutes of footage of a game not even close to beta,how about we wait until its 6 months until release to jump to conclusions.Oh and btw this is coming from a EQ vet that actually wanted a shiny new old version of EQ.
We're talking about the realistic approach taken to create "intelligent AI" whether it being a repeated script tweeked a bit or entirely new system(script) that combines loads of different variables and parameters to depict a intelligent AI.
This is done in the early planning stages and is not left "6 months until release" and they are going on about their "improved, intelligent AI" so it's right to talk about it now.
Get away from the 'script' idea. That's not how AI works, unless by 'script' you're referring to the agent's preferences. And Storybricks has been in development for quite awhile now.
Originally posted by truce12 Weve actually seen 5 minutes of footage of a game not even close to beta,how about we wait until its 6 months until release to jump to conclusions.Oh and btw this is coming from a EQ vet that actually wanted a shiny new old version of EQ.
We're talking about the realistic approach taken to create "intelligent AI" whether it being a repeated script tweeked a bit or entirely new system(script) that combines loads of different variables and parameters to depict a intelligent AI.
This is done in the early planning stages and is not left "6 months until release" and they are going on about their "improved, intelligent AI" so it's right to talk about it now.
Get away from the 'script' idea. That's not how AI works, unless by 'script' you're referring to the agent's preferences. And Storybricks has been in development for quite awhile now.
I was, but I have limited knowleage in the area only done some simple game development projects with a few mates making small games. I need to do more research on storybricks, I should of before I made this thread actually.
Do you know where I could get more information on them and what they are doing?
"if u forcefully insert foriegn objects into my? body, i will die"
To the op - we don't know what to expect from the AI. I think you're right to be skeptical, but you could just as easily be underestimating as well.
But he's right. The moment you have a smart AI that makes thing really challenging is when you have people raging and QQing on the forums demanding nerfs. MMOs have come to point where soloists and faceroll content reigns supreme since the game will be designed where a below avg player can play.
People will say they want smart AIs but when put in the situation they will hate it. Much like have everyone praised GW2 loss of the trinity or no gear grind but when they played it they sang a completely different tune after a while.
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
You'll pardon me for questioning your qualifications to pontificate on the matter when you confuse actual AI, that is Artificial Intelligence, which hasn't been developed anywhere yet despite decades of intensive research, with mob behavior in MMORPGs, which we gamers call "AI" but which is nothing of the kind. It is instead simple scripting.
Nor has SOE or any sensible third party said that EQN will have "player level intelligence." What we in fact might see, and what they are shooting for, is nothing more than much better and more sophisticated scripting. SOE is calling it "AI," but they clearly mean Gamer AI rather than Computer Science AI, which is what you're talking about.
To some extent this is proven technology (Storybricks,) but not on the scale EQN is going to try to deploy it. Which is IMO the single largest question regarding the whole project. If it works then Rallying Calls will work as well.
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
You'll pardon me for questioning your qualifications to pontificate on the matter when you confuse actual AI, that is Artificial Intelligence, which hasn't been developed anywhere yet despite decades of intensive research, with mob behavior in MMORPGs, which we gamers call "AI" but which is nothing of the kind. It is instead simple scripting.
Nor has SOE or any sensible third party said that EQN will have "player level intelligence." What we in fact might see, and what they are shooting for, is nothing more than much better and more sophisticated scripting. SOE is calling it "AI," but they clearly mean Gamer AI rather than Computer Science AI, which is what you're talking about.
To some extent this is proven technology (Storybricks,) but not on the scale EQN is going to try to deploy it. Which is IMO the single largest question regarding the whole project. If it works then Rallying Calls will work as well.
"Small scripting" is a form of AI. Anything that, on some level, is responding to something else, be it players or environment, constitutes AI. Even if its very simple. Quite simply its "a system that perceives its environment and takes actions that maximize its chances of success."
Even mob aggro is a form of Artificial intelligence. Though its only assigning threat as a numeric value to each and every action, be it the amount of damage inflicted by players, proximity and location of the players, or the health of each of the players, those things are then calculated and determine a course of action that the mob will take.
Go, go, semantics debates.
Just a side note, Everquest had more variables involved in its aggro mechanics and AI than any other game I've played. Much of this aggro bashing is really the fault of developers who fail to even meet the standard set by EQ1. The nature of AI and aggro is such that, the more variables you add to it, the smarter it becomes. This is why I laugh at people who write off threat management as stupid, especially developers (I'm looking at you Darrin McPherson), because its only as stupid as you make it. In EQ mobs would chase players low health regardless of the taunts or threat generated by tanks. They would also go out of their way to attack players sitting. Should the tank leave melee range, they would lose a certain amount of proximity threat which would send mobs chasing the casters and healers. Those are just a few of the dozens of mechanics associated with aggro in EQ1 (14 year old MMO) that this brilliant gaming industry failed to reproduce.
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
A game with really intelligent AI boss will already sense adventurers intruding his/her lair and will set up traps to one hit KO them even before they get to see the AI boss. Problem is this is not fun. That is why AI is always dumb, so players can have fun.
The possibility of the universe collapsing into a singularity is higher than the birth of a perfect MMORPG.
Originally posted by Dullahan "Small scripting" is a form of AI. Anything that, on some level, is responding to something else, be it players or environment, constitutes AI. Even if its very simple. Quite simply its "a system that perceives its environment and takes actions that maximize its chances of success."Even mob aggro is a form of Artificial intelligence. Though its only assigning threat as a numeric value to each and every action, be it the amount of damage inflicted by players, proximity and location of the players, or the health of each of the players, those things are then calculated and determine a course of action that the mob will take.Go, go, semantics debates.
Sure, but when you start asking the question of whether said AI is human-level or funamentally unpredictable, that's someting many orders of magnitude different from what we see in games. The point is that no one said EQN AI was going to be human level.
Just a side note, Everquest had more variables involved in its aggro mechanics and AI than any other game I've played. Much of this aggro bashing is really the fault of developers who fail to even meet the standard set by EQ1. The nature of AI and aggro is such that, the more variables you add to it, the smarter it becomes. This is why I laugh at people who write off threat management as stupid, especially developers (I'm looking at you Darrin McPherson), because its only as stupid as you make it. In EQ mobs would chase players low health regardless of the taunts or threat generated by tanks. They would also go out of their way to attack players sitting. Should the tank leave melee range, they would lose a certain amount of proximity threat which would send mobs chasing the casters and healers. Those are just a few of the dozens of mechanics associated with aggro in EQ1 (14 year old MMO) that this brilliant gaming industry failed to reproduce.
This is a good point; there's a lot that could be done even within the context of traditional threat-style AI to make encounters more dynamic and interesting. It's too bad so few development teams have tried to do that.
In EQN's case the idea is that mob AI is just one corner of a fundamentally different AI system (Storybricks) that changes the game dynamic even outside of how mobs act.
Originally posted by zwei2 A game with really intelligent AI boss will already sense adventurers intruding his/her lair and will set up traps to one hit KO them even before they get to see the AI boss. Problem is this is not fun. That is why AI is always dumb, so players can have fun.
Oddly enough this exact thing works great in DDO. But there, players have ways to detect and circumvent such traps. This is a problem with designer (and player) tunnel vision rather than with better AI.
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
You'll pardon me for questioning your qualifications to pontificate on the matter when you confuse actual AI, that is Artificial Intelligence, which hasn't been developed anywhere yet despite decades of intensive research, with mob behavior in MMORPGs, which we gamers call "AI" but which is nothing of the kind. It is instead simple scripting.
Nor has SOE or any sensible third party said that EQN will have "player level intelligence." What we in fact might see, and what they are shooting for, is nothing more than much better and more sophisticated scripting. SOE is calling it "AI," but they clearly mean Gamer AI rather than Computer Science AI, which is what you're talking about.
To some extent this is proven technology (Storybricks,) but not on the scale EQN is going to try to deploy it. Which is IMO the single largest question regarding the whole project. If it works then Rallying Calls will work as well.
You're not completely wrong... simple script & trigger AI technically falls under the category of 'model-based reflex agents'. They're a bit more complex than say, an automated Pong opponent (simple reflex) in that they maintain an internal "model of their world" in the form of the threat-list of all the players around them. So, although it kinda pains me, I have to admit that they can be considered 'intelligent agents' too.
(There is a serious drawback to e-textbook rental... I was looking for specific examples, but I've lost access to the damn book. Never again...)
It sounds to me like EQN wants to use learning agent AI (at least for tactical AI). That's where the AI grades its performance and alters its own decision making processes to try to get a better score. It sounds exotic, but its actually pretty common these days, especially in pattern recognition software - OCR and Voice recognition, as well as language translation software.
When you say that AI "hasn't been developed anywhere yet", I think you're referring to the branch they called 'cybernetic' AI research, which tried to create decision making algorithms by duplication actual brain functions and structures. That's been pretty much abandoned these days. Computer Scientists these days mostly focus on 'behavioral' AI research - the "if it walks like a duck" approach.
(Man... it torques me off that I can't go back and look at that book. )
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
I agree. EA made a big thing about mob AI in SWTOR and yet we had exactly the same thing as we always get: four mobs standing in place every ten yards just waiting to get killed and seeing nothing and hearing nothing till you got in aggro range.
Just a side note, Everquest had more variables involved in its aggro mechanics and AI than any other game I've played. Much of this aggro bashing is really the fault of developers who fail to even meet the standard set by EQ1. The nature of AI and aggro is such that, the more variables you add to it, the smarter it becomes. This is why I laugh at people who write off threat management as stupid, especially developers (I'm looking at you Darrin McPherson), because its only as stupid as you make it. In EQ mobs would chase players low health regardless of the taunts or threat generated by tanks. They would also go out of their way to attack players sitting. Should the tank leave melee range, they would lose a certain amount of proximity threat which would send mobs chasing the casters and healers. Those are just a few of the dozens of mechanics associated with aggro in EQ1 (14 year old MMO) that this brilliant gaming industry failed to reproduce.
I couldn't agree more , and have been saying this for weeks, but no one is listening.. LOL If you over de-buffed the mob you got agro, if you over healed the mob, you got agro.. And yes.. The AI mobs only seem to get dumber because the DEVS made it that way.. It's not the fault of the mechanics as some like to point fingers at.. It's amazing how technology and programming got better, but the devs failed to implement them correctly.. EQ1's early AI was the best that I"ve experienced..
You are all overestimating how intelligent and organic AI can be created in our current state of game development as of 2013.
A player level intelligence, that is random and unpredictable is something we are incapable of creating in an MMO as of now, Unless Sony has some hidden groundbreaking system which sets a complete new standard for game/MMO AI, we will not get the amazing, intelligent, groundbreaking AI you all seem to think is possible.
Anyone who has basic knowleage of AI, or has taken simple lessons in game development, computer science or just has a general interest in the area would know, that what Sony is promising is unrealistic and likely not going to come to fruition.
What are other people's thoughts on this matter?
Bingo.
Ignorance is bliss, but unfortunately as both a 25 year game veteran, 15 year mmo veteran and a computer programmer, I can't help but be disappointed at these lofty promises of emergent AI.
Their best bet is just building on the threat management system. No one has recreated it so that it was challenging and engaging as it was in classic Everquest, and no one has fully realized the potential of making the system much more dynamic. Most if not all of what they are promising could be done while still maintaining threat management and the class role system.
So because you say something can't be done then why even bother iterating on the process. Glad major industry didn't take this same line back during the industrial revolution or else we'd never have internet, modern medicine or anything else technology has given mankind.
Random Forum Poster: I want an MMO that is different, original and fun.
Me: So you want something like EQN
Them: Nah dude, I want a Holy Trinity, Tab Target combat, Instanced Raiding, and Rigid classes.
Join a pvp match check the good and bad players. fight the bad player and win fight the good player and get roflstomped
Join a WvW match in GW2 and its run run dodge run run kite run run dodge run run kite.
Let asume bosses scale to the amount of players like in GW2: 100 people hitting boss, random AoE, random targeting, random CC. evrything will be random, maybe AI always targets the lowest health, thats the msartest decision AI can make. Expect to DIE...to RUN....to ZERG This is how GW2 AI is made. And EQN will not be much diffrent. Smart AI will always target the almost weakest link or it isnt smart AI but dumb AI. Again Expect to DIE alot !
MMO's with no healers is a mess, no matter how you look at it. It creates chaos mobs and bosses running around people on low health sprinting away.
I dont understand what people like in mmo's without healers.....or tanks..... it creates a flow of battle. With healers and tanks mobs can still wipe your party no ? but then it will be due to bad decisions of the players, not due to mobs training 1 target.
But people will see how awfull it is, GW2 players already know how it works, the people who dint play GW2 will find out once they witness the great advanced AI mobs training players to death.
Comments
depends really, you can script so much into monster ai, its very easy to overpower , or make a enemy ai nearly unbeatable. this is just an exert from 4GAI, the 4th gen aAI in Unangband, from 5 years ago, just think whatt can be done now.
over 20 years of mmorpg's and counting...
They would not be "looking" for a new spot, although it might appear that way. They will move to a spot based on the parameters of the situation. This spot will already be pre-programmed, or in other words will be scripted to happen. That is why I've said countless times that what they are calling AI, isn't actually AI.
They are using, hopefully a complex script, and when I mean script, I mean scripting of events. The complex script would then give off the illusion of intelligent Orcs and would also give off the illusion that their placement is random.
I think a combination of player intelligence and AI would be best. Create tools for staff to coordinate the actions of groups of mobs. For example the staff person sees some orcs getting their butts handed to them by players and he can decide to send some reinforcements from a nearby mountain. Human intelligence makes things unpredictable and a lot more fun.
As a recently graduated CS major I'll second this one. I've only taken one class on AI programming... and we only just touched on adaptive AI (with genetic and neural net algorithms)... but modern AI tech is WAY beyond the level that most games implement.
In a finite space, a 'simple' tree search algorithm can give you a surprisingly hard opponent to beat. And if they're using multiple layers of decision making algorithms... well. The problem might be tuning it down, so it doesn't kick our collective arses in casual, small-group play.
Thanks for posting this.
"if u forcefully insert foriegn objects into my? body, i will die"
You are very likely correct. But let's suspend disbelief for a minute. Let's assume that they ARE given a simple AI and do randomly move to a new location based off of it meeting certain criteria. The location wouldn't have to be on a predetermined list because they are smart enough to find a new location on their own. This could be very similar to a herd of animals looking for a water source when their current one dries up.
The problem I was trying to point out is that even if they gave the orcs enough AI to migrate, it wouldn't make the game any more realistic to an individual player. As you said, the original spawn points are going to be set (how else would they be able to generate quests around their existence if they weren't). So even though players are actually encountering different groups of mobs, to the players it may as well be the same exact group since it is in the same exact spot.
Every AI is scripted, even human intelligence. Learning from your mistakes is scripted, its like a script that tells you to create your own script from an outcome. Its like extremely advanced if then else statements that branches out into trillions of branches through more experience the artificial intelligence goes through. Basically like you get hurt from touching something and than your brain learns from that and makes more neurons from that new experience, and in this case our neurons are like new formed scripts are brains have written down.
We're talking about the realistic approach taken to create "intelligent AI" whether it being a repeated script tweeked a bit or entirely new system(script) that combines loads of different variables and parameters to depict a intelligent AI.
This is done in the early planning and development stages and is not left "6 months until release" and they are going on about their "improved, intelligent AI" so it's right to talk about it now.
"if u forcefully insert foriegn objects into my? body, i will die"
That's because they can tag each location with values that may or may not be attractive to the mob's AI... and NPC's AI can read those values to generate quests as well. AND those values can shift over time due to player or mob activity. In other words, if a strong group of players clear out all the wandering groups looking for food or loot, the orcs at the camp may hunker down and fortify... or call in reinforcements. If the camp is destroyed there are obviously strong players hunting in the area... so the value of the area changes and the orcs are less likely to return. Until the players stop hunting the area so aggressively and the values in the area start to return to normal, the orcs will stay away.
But maybe the gnolls or bandits will start expanding into the area first next time... they may have different environmental threshholds than orcs.
You should really watch the debut video again, where Dave describes the emergent behavior of the orcs and the village.
Get away from the 'script' idea. That's not how AI works, unless by 'script' you're referring to the agent's preferences. And Storybricks has been in development for quite awhile now.
I was, but I have limited knowleage in the area only done some simple game development projects with a few mates making small games. I need to do more research on storybricks, I should of before I made this thread actually.
Do you know where I could get more information on them and what they are doing?
"if u forcefully insert foriegn objects into my? body, i will die"
But he's right. The moment you have a smart AI that makes thing really challenging is when you have people raging and QQing on the forums demanding nerfs. MMOs have come to point where soloists and faceroll content reigns supreme since the game will be designed where a below avg player can play.
People will say they want smart AIs but when put in the situation they will hate it. Much like have everyone praised GW2 loss of the trinity or no gear grind but when they played it they sang a completely different tune after a while.
You'll pardon me for questioning your qualifications to pontificate on the matter when you confuse actual AI, that is Artificial Intelligence, which hasn't been developed anywhere yet despite decades of intensive research, with mob behavior in MMORPGs, which we gamers call "AI" but which is nothing of the kind. It is instead simple scripting.
Nor has SOE or any sensible third party said that EQN will have "player level intelligence." What we in fact might see, and what they are shooting for, is nothing more than much better and more sophisticated scripting. SOE is calling it "AI," but they clearly mean Gamer AI rather than Computer Science AI, which is what you're talking about.
To some extent this is proven technology (Storybricks,) but not on the scale EQN is going to try to deploy it. Which is IMO the single largest question regarding the whole project. If it works then Rallying Calls will work as well.
Ardwulf's Lair | Ardwulf's YouTube Channel | Ardwulf on Twitter
"Small scripting" is a form of AI. Anything that, on some level, is responding to something else, be it players or environment, constitutes AI. Even if its very simple. Quite simply its "a system that perceives its environment and takes actions that maximize its chances of success."
Even mob aggro is a form of Artificial intelligence. Though its only assigning threat as a numeric value to each and every action, be it the amount of damage inflicted by players, proximity and location of the players, or the health of each of the players, those things are then calculated and determine a course of action that the mob will take.
Go, go, semantics debates.
Just a side note, Everquest had more variables involved in its aggro mechanics and AI than any other game I've played. Much of this aggro bashing is really the fault of developers who fail to even meet the standard set by EQ1. The nature of AI and aggro is such that, the more variables you add to it, the smarter it becomes. This is why I laugh at people who write off threat management as stupid, especially developers (I'm looking at you Darrin McPherson), because its only as stupid as you make it. In EQ mobs would chase players low health regardless of the taunts or threat generated by tanks. They would also go out of their way to attack players sitting. Should the tank leave melee range, they would lose a certain amount of proximity threat which would send mobs chasing the casters and healers. Those are just a few of the dozens of mechanics associated with aggro in EQ1 (14 year old MMO) that this brilliant gaming industry failed to reproduce.
A game with really intelligent AI boss will already sense adventurers intruding his/her lair and will set up traps to one hit KO them even before they get to see the AI boss. Problem is this is not fun. That is why AI is always dumb, so players can have fun.
The possibility of the universe collapsing into a singularity is higher than the birth of a perfect MMORPG.
Sure, but when you start asking the question of whether said AI is human-level or funamentally unpredictable, that's someting many orders of magnitude different from what we see in games. The point is that no one said EQN AI was going to be human level.
This is a good point; there's a lot that could be done even within the context of traditional threat-style AI to make encounters more dynamic and interesting. It's too bad so few development teams have tried to do that.
In EQN's case the idea is that mob AI is just one corner of a fundamentally different AI system (Storybricks) that changes the game dynamic even outside of how mobs act.
Ardwulf's Lair | Ardwulf's YouTube Channel | Ardwulf on Twitter
Oddly enough this exact thing works great in DDO. But there, players have ways to detect and circumvent such traps. This is a problem with designer (and player) tunnel vision rather than with better AI.
Ardwulf's Lair | Ardwulf's YouTube Channel | Ardwulf on Twitter
You're not completely wrong... simple script & trigger AI technically falls under the category of 'model-based reflex agents'. They're a bit more complex than say, an automated Pong opponent (simple reflex) in that they maintain an internal "model of their world" in the form of the threat-list of all the players around them. So, although it kinda pains me, I have to admit that they can be considered 'intelligent agents' too.
(There is a serious drawback to e-textbook rental... I was looking for specific examples, but I've lost access to the damn book. Never again...)
It sounds to me like EQN wants to use learning agent AI (at least for tactical AI). That's where the AI grades its performance and alters its own decision making processes to try to get a better score. It sounds exotic, but its actually pretty common these days, especially in pattern recognition software - OCR and Voice recognition, as well as language translation software.
When you say that AI "hasn't been developed anywhere yet", I think you're referring to the branch they called 'cybernetic' AI research, which tried to create decision making algorithms by duplication actual brain functions and structures. That's been pretty much abandoned these days. Computer Scientists these days mostly focus on 'behavioral' AI research - the "if it walks like a duck" approach.
(Man... it torques me off that I can't go back and look at that book. )
I agree. EA made a big thing about mob AI in SWTOR and yet we had exactly the same thing as we always get: four mobs standing in place every ten yards just waiting to get killed and seeing nothing and hearing nothing till you got in aggro range.
I couldn't agree more , and have been saying this for weeks, but no one is listening.. LOL If you over de-buffed the mob you got agro, if you over healed the mob, you got agro.. And yes.. The AI mobs only seem to get dumber because the DEVS made it that way.. It's not the fault of the mechanics as some like to point fingers at.. It's amazing how technology and programming got better, but the devs failed to implement them correctly.. EQ1's early AI was the best that I"ve experienced..
So because you say something can't be done then why even bother iterating on the process. Glad major industry didn't take this same line back during the industrial revolution or else we'd never have internet, modern medicine or anything else technology has given mankind.
Random Forum Poster: I want an MMO that is different, original and fun.
Me: So you want something like EQN
Them: Nah dude, I want a Holy Trinity, Tab Target combat, Instanced Raiding, and Rigid classes.
Me: Double Facepalm.
I would like to see threat work in PvP .
I think that would make the game much more tactical .
Then CC would really be useful in PvP to mez healers to kill tanks etc .
Atlantica Online kinda had that .
Join a pvp match
check the good and bad players.
fight the bad player and win
fight the good player and get roflstomped
Join a WvW match in GW2 and its run run dodge run run kite run run dodge run run kite.
Let asume bosses scale to the amount of players like in GW2:
100 people hitting boss, random AoE, random targeting, random CC.
evrything will be random, maybe AI always targets the lowest health, thats the msartest decision AI can make.
Expect to DIE...to RUN....to ZERG
This is how GW2 AI is made.
And EQN will not be much diffrent.
Smart AI will always target the almost weakest link or it isnt smart AI but dumb AI.
Again Expect to DIE alot !
MMO's with no healers is a mess, no matter how you look at it.
It creates chaos mobs and bosses running around people on low health sprinting away.
I dont understand what people like in mmo's without healers.....or tanks.....
it creates a flow of battle.
With healers and tanks mobs can still wipe your party no ?
but then it will be due to bad decisions of the players, not due to mobs training 1 target.
But people will see how awfull it is, GW2 players already know how it works, the people who dint play GW2 will find out once they witness the great advanced AI mobs training players to death.