AI is stupid because people insist on wanting to be able to solo everything, as every class.
If you designed an AI that was smart, you would have NPC's traveling in packs, where they would work together to surround the player, placing melee into LOS between the player and their ranged attackers, ensuring that some of them were always flanking. Basically, solo players would die like dogs, and then go whining to the forums about how horrible the game is.
So, if you want smart AI, you need to convince the dev studios that there's a good sized market for a group-only MMO.
Not the textures themselves, but all of the objects they make up.
I'll break it down for you simpletons (though this is by no means a real-world example)
You have 500 bytes in 1995 to make a mob. 300 is allocated to the actual rendering of said mob, 200 to their behavior, etc.
You have 1,000,000,000 bytes in 2015 to make a mob. Now that the standard for graphics has dramatically increased, so has the requirement for rendering. So even though you have a crazy amount more to work with, it gets eaten up by "progress" unless you're not worried about looks.
Now, not only do you have to make sure that the rendering of the character itself has what it needs to be up to snuff, it also needs to have effects, physics, etc that apply to it that did not back in the day. Orrrr you can look forward to people talking about how stupid your animations, etc look (see: tsw).
And that's all you get for free. Any effort taken beyond this point is just an exercise in futility, as you're not likely to understand.
So just gonna try and help you out.
I found one of the most fun ways to learn java programming wars Robocode. It's a game where you program the robots AI.
If you want to check out some cool game specific stuff, Bethesda made the Radiant AI for the Elder Scrolls series. Or you could check out the Skyrim creation kit tutorials and learn papyrus scripting to do your own AI scripts.
If your really serious and want to start to learn programming I recommend Dietel and Deitel programming books which have been around for over a decade since I went to college for programming.
Oh and also something really fun! A.L.I.C.E. chat bots. (This site advertises with digital AI chat bots) it's worth trying out tho and talking to it to see what it says! I mean come on after looking at this how can you not want to code an AIML chat bot!
"You CAN'T buy ships for RL money." - MaxBacon
"classification of games into MMOs is not by rational reasoning" - nariusseldon
A players' health is 100k. a tank's health is 200k. a boss has 650,000,000 hp. Melee attacks for 50k (in tank armor) and 120k in non-tank armor. Abilities that deal 120k (threat-based), ground telegraph abilities that deal 1,000,000 and aoe abilities that deal 30k to all.
Combine the red lines and you can see the issue. If the boss has any intelligence it'll be unkillable.
The reason a raid with 100k hp members can beat a boss dealing millions of damage is by avoiding mechanics. If boss is intelligent, in tab-target environment it's no longer possible.
You'd literarly need action combat where you can dodge a pyroblast or a cleave.
Addresing the Tier 9 Faction Champions: there were tricks that took advantage of the Ai like in any traditional fight:
- mobs targeted low-hp members, so tanks were never topped off.
-mobs would heal their injured companions so a few dps was assigned to constantly damage the pet and sponge heals away from main NPCs.
Probably don't need that logical fallacy to argue against smart AI. To make a fair comparison you have to compare the best possible smart AI design with the best possible dumb AI design. (With the caveat that "best possible" also implies the design is feasible.)
Showing that smart AI wouldn't work with numbers which were balanced for dumb AI is not a compelling argument. Obviously for a fair comparison you'd have to assume the designers weren't complete idiots and actually tried to make things balanced appropriately for a smart AI game (which strongly implies numbers a lot closer to the player's own.)
Personally I haven't seen smart AI in any game which gives me the impression that it would be more fun in MMORPGs than dumb AI. Dumb AI is simply one of the rules of the game. Often it's not even a big part of the rules that establish the "puzzle" of any given encounter; mostly the rules you care about are the specific abilities the boss will use this fight. By keeping the AI simple, you can immediately move on to study the things you know are different each encounter. You don't have to worry about weird random oddities happening.
So I do largely agree with your implication that smart AI isn't going to be an improvement.
Also FWIW real-time dodging is possible and used in several WOW fights (not to mention GW1). A recent example being Brackenspore who fires projectiles at your group. It goes even further than that, because you actually want 1 person intercepting the projectiles sometimes (because they're AOE explosions and if they reach your main group and it isn't paying attention then that's a ton of raid damage.)
"What is truly revealing is his implication that believing something to be true is the same as it being true. [continue]" -John Oliver
AI is stupid because people insist on wanting to be able to solo everything, as every class.
If you designed an AI that was smart, you would have NPC's traveling in packs, where they would work together to surround the player, placing melee into LOS between the player and their ranged attackers, ensuring that some of them were always flanking. Basically, solo players would die like dogs, and then go whining to the forums about how horrible the game is.
So, if you want smart AI, you need to convince the dev studios that there's a good sized market for a group-only MMO.
Good luck!
Why wouldn't the player simply move to a more defensible spot, stay on the move, make a plan to isolate one at a time, or simply beat feet away?
I'm assuming here in this fictional MMO with AI that there's not a lot of static content and this particular encounter might have been with just 1 MOB, or 20. And that the makeup of this MOB group can be quite different on any single encounter.
And that, I think, is a big part of why many gamers want better AI. So that the encounters can be something much less predictable than today's games, both in makeup and in actions. And with that last part, "in actions", I think a randomness needs to be introduced to the MOB's actions. Rather than an AI that always does "Perfect-10%" it needs to vary, not just on the encounter, but per rounds of action.
Every journey starts with the first step. Sitting at home in couch wont get you anywhere. And thats just what most MMOs have been doing for past 15 years.
I didnt even mention its possible to do in entirety atm, but that doesnt mean it cant be improved upon. Oh yeah, it woulod take a lot of hard work, so what. Lot of things take a lot of hard work.
Yes, you can. But why bother?
Games are already challenging. So what if the NPCs act a bit wooden. You are asking for challenges, not for them to pass a turing test.
why, bother with anything, caves server a purpose, and walking is perfectly good way of travelling and electricity is just convenience.
Because electricity makes a HUGE difference, and better AI does not?
How would you know, how many games with smart AI have you played?
I can tell you right off the bat that chess is pretty awesome against smart AI and pretty bad against dumb AI.
Blizz already tried smart AI ... got QQ a lot, and dropped it.
hmm ... actually chess is pretty dumb AI, just alpha-beta pruning will beat 99.9999% of the population. Only Kasporov needs special treatment.
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
There are certain queer times and occasions in this strange mixed affair we call life when a man takes this whole universe for a vast practical joke, though the wit thereof he but dimly discerns, and more than suspects that the joke is at nobody's expense but his own. -- Herman Melville
AI is stupid because people insist on wanting to be able to solo everything, as every class.
If you designed an AI that was smart, you would have NPC's traveling in packs, where they would work together to surround the player, placing melee into LOS between the player and their ranged attackers, ensuring that some of them were always flanking. Basically, solo players would die like dogs, and then go whining to the forums about how horrible the game is.
So, if you want smart AI, you need to convince the dev studios that there's a good sized market for a group-only MMO.
Good luck!
Why wouldn't the player simply move to a more defensible spot, stay on the move, make a plan to isolate one at a time, or simply beat feet away?
I'm assuming here in this fictional MMO with AI that there's not a lot of static content and this particular encounter might have been with just 1 MOB, or 20. And that the makeup of this MOB group can be quite different on any single encounter.
And that, I think, is a big part of why many gamers want better AI. So that the encounters can be something much less predictable than today's games, both in makeup and in actions. And with that last part, "in actions", I think a randomness needs to be introduced to the MOB's actions. Rather than an AI that always does "Perfect-10%" it needs to vary, not just on the encounter, but per rounds of action.
Combat system would obviously need rethinking too, this is not change that affect just mobs.
And people cant get away from seeing it too narrow, its not JUST about challenge, its about upgrade of quality of game as a whole, where dog would act like a dog and elite mercenary would act as elite mercenary. Challenge on both of those can be tuned to whatever design goals are.
Along that, it would remove highly repetitve encounters, just go through WoW for example and go to some old raid. NOTHING has changed, that raid from 3 years ago is exactly the same, its throwaway content today. WoW can afford to do that but other MMOs (PvE ones at least) cant. Devs cannot churn out content nearly fast enough.
And lastly, it would allow for AI to make actual changes to game world, some of that ideas that EQN was supposed to bring on (although that game is highly questionable). You wiped out those mercenaries last time? No problems, wildlife has moved into their base. Thats lab you stormed last time? well, their experiments seems to came to life....
Oh yeah, the ultimate goal is having real AI, but until then at least some improvements can be done. Yeah, its not simple, its not something that will just come to life out of ether, but yeah, MMOs need something new, let those devs earn their money
Every journey starts with the first step. Sitting at home in couch wont get you anywhere. And thats just what most MMOs have been doing for past 15 years.
I didnt even mention its possible to do in entirety atm, but that doesnt mean it cant be improved upon. Oh yeah, it woulod take a lot of hard work, so what. Lot of things take a lot of hard work.
Yes, you can. But why bother?
Games are already challenging. So what if the NPCs act a bit wooden. You are asking for challenges, not for them to pass a turing test.
why, bother with anything, caves server a purpose, and walking is perfectly good way of travelling and electricity is just convenience.
Because electricity makes a HUGE difference, and better AI does not?
How would you know, how many games with smart AI have you played?
I can tell you right off the bat that chess is pretty awesome against smart AI and pretty bad against dumb AI.
Blizz already tried smart AI ... got QQ a lot, and dropped it.
hmm ... actually chess is pretty dumb AI, just alpha-beta pruning will beat 99.9999% of the population. Only Kasporov needs special treatment.
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
Strong != smart. Again, a simple algorithm like alpha-beta pruning can beat 99.999% of humans just because the computer can look ahead more. That is an very old very simple algorithm.
The point is .. making an AI that is challenging is easy in games. It is not necessarily to be desirable because where is the fun if the computer wins 99.99999% of the time?
AI is stupid because people insist on wanting to be able to solo everything, as every class.
If you designed an AI that was smart, you would have NPC's traveling in packs, where they would work together to surround the player, placing melee into LOS between the player and their ranged attackers, ensuring that some of them were always flanking. Basically, solo players would die like dogs, and then go whining to the forums about how horrible the game is.
So, if you want smart AI, you need to convince the dev studios that there's a good sized market for a group-only MMO.
Good luck!
Why wouldn't the player simply move to a more defensible spot, stay on the move, make a plan to isolate one at a time, or simply beat feet away?
I'm assuming here in this fictional MMO with AI that there's not a lot of static content and this particular encounter might have been with just 1 MOB, or 20. And that the makeup of this MOB group can be quite different on any single encounter.
And that, I think, is a big part of why many gamers want better AI. So that the encounters can be something much less predictable than today's games, both in makeup and in actions. And with that last part, "in actions", I think a randomness needs to be introduced to the MOB's actions. Rather than an AI that always does "Perfect-10%" it needs to vary, not just on the encounter, but per rounds of action.
Combat system would obviously need rethinking too, this is not change that affect just mobs.
And people cant get away from seeing it too narrow, its not JUST about challenge, its about upgrade of quality of game as a whole, where dog would act like a dog and elite mercenary would act as elite mercenary. Challenge on both of those can be tuned to whatever design goals are.
Along that, it would remove highly repetitve encounters, just go through WoW for example and go to some old raid. NOTHING has changed, that raid from 3 years ago is exactly the same, its throwaway content today. WoW can afford to do that but other MMOs (PvE ones at least) cant. Devs cannot churn out content nearly fast enough.
And lastly, it would allow for AI to make actual changes to game world, some of that ideas that EQN was supposed to bring on (although that game is highly questionable). You wiped out those mercenaries last time? No problems, wildlife has moved into their base. Thats lab you stormed last time? well, their experiments seems to came to life....
Oh yeah, the ultimate goal is having real AI, but until then at least some improvements can be done. Yeah, its not simple, its not something that will just come to life out of ether, but yeah, MMOs need something new, let those devs earn their money
I agree that combat would change. It wouldn't just be tank and nuke with support. Those elements would still be there, but there'd be added decisions based on unpredictable "tactics". Those tactics, the way I think of it in my idea of AI, would actually be changes based on current actions and results but with randomness modified by a "morale" factor for the MOBs.
This totally removes predictable MOB behavior. It makes it seem like MOBs are thinking because they are changing up according to what's happening, yet they aren't always reacting the same way. The "morale" modifier I mentioned would probably be the defining difference between "elite" and regular MOBs.
The way I see AI being built is on a tier. Ranging from lowest basic survival instincts all the way up to superior intelligence. Each level adds actionable tactics as "choices", and socialized MOBs work as a group with more options. And all modified by "morale". Also, a socialized individual could lose all "morale" and break out of that group order and into basic individual "choice", usually total fear and the choice of just running away. Add the possibility of extreme plus "morale" where a socialized individual breaks from group command and decides to act on that MOB's own, usually trying to take over and lead that group of MOBs (leading an attack, taking over as leader of the group, leaving that group-part of MOB expansion, etc.). See, I think it's possible to create an AI where MOBs are constantly doing things in the world without players being involved. Seeking and building shelter and defense, moving in search of better food and shelter conditions, seeking resources, and generally advancing themselves as individuals or socialized groups. Think of it as making what appears to be advanced thinking AI but really simply a range of modified random actions based on circumstances.
On top of this AI would be a world that adds to choices. Locations (small and large) that are tagged for possible uses. So, as an example, in a dungeon there can be areas tagged for construction of defensive items such as picket fences, stonework walls, traps of various kinds, etc. There can also be areas tagged for a variety of uses such as choke points (for missile traps or concentrated fire) and other things. And each race or MOB type would have their own sorts of uses in each of these tagged areas.
All of this would be expensive to make and require lots of extra computer power. But I think it would create a much more interesting world. An ever changing world. A world that feels alive, dangerous, and full of adventure.
Every journey starts with the first step. Sitting at home in couch wont get you anywhere. And thats just what most MMOs have been doing for past 15 years.
I didnt even mention its possible to do in entirety atm, but that doesnt mean it cant be improved upon. Oh yeah, it woulod take a lot of hard work, so what. Lot of things take a lot of hard work.
Yes, you can. But why bother?
Games are already challenging. So what if the NPCs act a bit wooden. You are asking for challenges, not for them to pass a turing test.
why, bother with anything, caves server a purpose, and walking is perfectly good way of travelling and electricity is just convenience.
Because electricity makes a HUGE difference, and better AI does not?
How would you know, how many games with smart AI have you played?
I can tell you right off the bat that chess is pretty awesome against smart AI and pretty bad against dumb AI.
Blizz already tried smart AI ... got QQ a lot, and dropped it.
hmm ... actually chess is pretty dumb AI, just alpha-beta pruning will beat 99.9999% of the population. Only Kasporov needs special treatment.
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
Strong != smart. Again, a simple algorithm like alpha-beta pruning can beat 99.999% of humans just because the computer can look ahead more. That is an very old very simple algorithm.
The point is .. making an AI that is challenging is easy in games. It is not necessarily to be desirable because where is the fun if the computer wins 99.99999% of the time?
And that's why you need a randomness factor so that the computer can screw up and give you an opening now and again, if you see it and take advantage of it.
If you miss that one, there'll be another one coming for various effect soon enough. Maybe.
Serious question (based on the above statement) : How long does it take to actually implement new content (new quests, new characters, new stories, program it it all and get it in the world)? How long does it take for putting it into the world (I would guess the coding end) once the stories etc are already worked out?
As far as the AI goes... Instead of dumbing down content, wouldn't it be possible to set up tutorials to build up the players to more challenging content? (ie... dumb down the tutorials instead?)
Serious question (based on the above statement) : How long does it take to actually implement new content (new quests, new characters, new stories, program it it all and get it in the world)? How long does it take for putting it into the world (I would guess the coding end) once the stories etc are already worked out?
As far as the AI goes... Instead of dumbing down content, wouldn't it be possible to set up tutorials to build up the players to more challenging content? (ie... dumb down the tutorials instead?)
Tough question to answer, but it's clear that content just can't be created fast enough. No one can do it, and everyone who does it says they see a problem here.
That would be a huge benefit if you could create AI that allows for the game world to add content on it's own by changing the world. But I don't see this ever being up to the needed standard but rather being a big help if done right. GMs setting MOB course-of-actions and building eventful plans up along with this sort of AI could be very powerful, in my opinion.
That last part, I'm hoping you'd like to have that one back.
Serious question (based on the above statement) : How long does it take to actually implement new content (new quests, new characters, new stories, program it it all and get it in the world)? How long does it take for putting it into the world (I would guess the coding end) once the stories etc are already worked out?
Depends on way too many factors.
Generic quests using existing assets and destined for existing world environments can probably be pumped out rather quick. There's a lot more to it than that. Just for the creation of the quest text, items and basic gameplay alone you have Continuity, Adherence to lore, Game balance, Reward itemization among other factors to consider. If it is getting its own assets or its own worldspace, you're now greatly increasing the number of people involved. Voiceovers or cutscenes? That's more people and money, too.
The smarter the mobs, the more scripted the quest, the more elaborate each scene is that has to be created... the more teams need to be involved.
So, yeah, generic quests with complete disregard for economy, gameplay, brand, continuity or lore could probably be pumped out by the dozens every day, even then you're probably bound by build/release schedules. Beyond that, it all depends on the game's features, the quest's impact on existing gameplay, dev tools, finite resources, and how crazy you want to get with it.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein "Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
That would be a huge benefit if you could create AI that allows for the game world to add content on it's own by changing the world. But I don't see this ever being up to the needed standard but rather being a big help if done right. GMs setting MOB course-of-actions and building eventful plans up along with this sort of AI could be very powerful, in my opinion.
The scary part of that is that for every dev team of 3-300 devs that works to create an AI to do that, there's a playerbase of hundreds of thousands of players that will do everything in their power to master it and game the crap out of it. Not saying it shouldn't be done, rather that is a very good reason why we haven't seen one released.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein "Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
Elementary, my dear Watson. MMO AI is and has been the same for a long time because that is the norm and it is easier to just keep it that way then take chances on evolving it. That and it was/has been difficult for dev teams, tech wise, to make it possible.
The first MMO I ever played, Nexustk, had about the same mob AI as most games these days do. Except that it was 2d and the year 1998. Everquest 1 had the same thing, with one difference. Mobs would run away at a certain health % and bring back other mobs to wreck your ass. And pretty much every other MMO post-EQ1 used the same AI. WoW has the same AI, and every WoW clone in the last decade does too.
EverQuest 2 made a bold claim during its development, saying how it was going to introduce smart AI. The example they used was if an Orc spotted you across a valley it would then process the threat level. If it thought it could kill you it would attack and kill you. If the Orc thought you were to strong and it would die in a fight it would then run off and get help and then you would get swarmed by Orcs. I remember reading this with my buddies while at sea and thinking how fucking cool it sounded. And how it would revolutionize the MMO gameplay. EQ2 launched, and the AI was the same fucking thing it had always been. But it was actually worse, now mobs would reset at a certain distance away from their spawn point or their travel path. Making the world feel even less real and the danger level even less severe. You wouldn't have to run for your life and pray someone helps you or you get to a zone line before the Orc catches up to you and mercs your ass. Just run a few feet away and it resets back to full health and it is like nothing had ever even happened.
Honestly I think AI has stayed so stagnant because it is to risky to create something new and improved. If it fucks up and doesn't work then your entire game is screwed and hundreds of millions of dollars are flushed down the toilet. It is safer for investors and corporations to just keep pumping out the same old shit with a new skin and reaping as much profit from the traveling nomadic MMO players.
Then again the market is shifting and those fat cat companies are moving on and forgetting about the MMO genre. SOE is now Daybreak a much smaller version of itself. Blizzard gave up on Titan and is now going to MOBA town. Funcom will probably end soon. The asian market is steering more towards Sandparks. Cryptic, Turbine, etc.. are no longer relevant. Smaller companies are starting to kickstart things and try new shit.
And yet again! Daybreak is saying that EQN's emergent AI is going to change the game! Mobs will now actually have thought patterns and behavior. They will migrate and look for environments that suit their needs/wants. If attacked they will either pick up camp and move, or they will build up and retaliate. The days of a mob standing still until you get up in their face will be over and they will never keep spawning in the same place. But this is all words and knowing SOE/Daybreak's history it will never actually happen.
The most I expect to see out of EQN and other games in the near future is a mix of Rift and GW2. Which is the same old shit, mobs on a scripted path doing the same shit day in, day out.
Playing: Smite, Marvel Heroes Played: Nexus:Kingdom of the Winds, Everquest, DAoC, Everquest 2, WoW, Matrix Online, Vangaurd, SWG, DDO, EVE, Fallen Earth, LoTRo, CoX, Champions Online, WAR, Darkfall, Mortal Online, Guild Wars, Rift, Tera, Aion, AoC, Gods and Heroes, DCUO, FF14, TSW, SWTOR, GW2, Wildstar, ESO, ArcheAge Waiting On: Nothing. Mmorpg's are dead.
Every journey starts with the first step. Sitting at home in couch wont get you anywhere. And thats just what most MMOs have been doing for past 15 years.
I didnt even mention its possible to do in entirety atm, but that doesnt mean it cant be improved upon. Oh yeah, it woulod take a lot of hard work, so what. Lot of things take a lot of hard work.
Yes, you can. But why bother?
Games are already challenging. So what if the NPCs act a bit wooden. You are asking for challenges, not for them to pass a turing test.
why, bother with anything, caves server a purpose, and walking is perfectly good way of travelling and electricity is just convenience.
Because electricity makes a HUGE difference, and better AI does not?
How would you know, how many games with smart AI have you played?
I can tell you right off the bat that chess is pretty awesome against smart AI and pretty bad against dumb AI.
Blizz already tried smart AI ... got QQ a lot, and dropped it.
hmm ... actually chess is pretty dumb AI, just alpha-beta pruning will beat 99.9999% of the population. Only Kasporov needs special treatment.
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
Strong != smart. Again, a simple algorithm like alpha-beta pruning can beat 99.999% of humans just because the computer can look ahead more. That is an very old very simple algorithm.
The point is .. making an AI that is challenging is easy in games. It is not necessarily to be desirable because where is the fun if the computer wins 99.99999% of the time?
And that's why you need a randomness factor so that the computer can screw up and give you an opening now and again, if you see it and take advantage of it.
If you miss that one, there'll be another one coming for various effect soon enough. Maybe.
That is even easier .. which is in all chess programs now. When you are doing the alpha-beta pruning, just put in RNG .. and occasionally choose a non-optimized option.
But again, the point is that being strong is trivial .. it is about making the game fun. That is much harder.
That is even easier .. which is in all chess programs now. When you are doing the alpha-beta pruning, just put in RNG .. and occasionally choose a non-optimized option.
But again, the point is that being strong is trivial .. it is about making the game fun. That is much harder.
I normally don't agree with Nari due to their solo console game mentality.. (Don't get me wrong, I like soloing on console games too, but I don't mix MMO's with solo games and expect the same experience).. Anyways.. lol
I did want to highlight the two areas that I agree with Nari on that apply to any computer game.. I've always suggested that AI's be smart and optimal (100%) .. However, do not treat all mobs the same in regards to their AI.. Some mobs will be almost expert in the terms of AI skill, and other mobs will make repeated mistakes, just like chess programs.. The trick is to interject a variety of mobs together so that it gives the player a unexpected and fun time.. Which is what playing a game is all about.. FUN..
I've noticed that in every mmo I've ever played the AI has been noticeably dumber then in most single player games. With the exception of bosses the standard mob will behave in the same simplistic fashion in almost all mmos. They should be attacking in formations, avoiding especially powerful attacks, finding cover against ranged players or just generally doing anything other then running in a straight line and randomly executing whatever skills are off cooldown. I know this may be the compromise developers have to make to keep mechanics like tanking working or reduce server lag but smart AI is one of the most satisfying ways to make a game challenging and maybe it's about time developers stopped making that compromise.
Challenging AI
= risky AI for mostly for niche gamers, smaller community = nich game, less $$$$$
Simplistic AI =
Little risk AI for majority easy to satisfy 'follow the carrot' community = more $$$$$$
"going into arguments with idiots is a lost cause, it requires you to stoop down to their level and you can't win"
Every journey starts with the first step. Sitting at home in couch wont get you anywhere. And thats just what most MMOs have been doing for past 15 years.
I didnt even mention its possible to do in entirety atm, but that doesnt mean it cant be improved upon. Oh yeah, it woulod take a lot of hard work, so what. Lot of things take a lot of hard work.
Yes, you can. But why bother?
Games are already challenging. So what if the NPCs act a bit wooden. You are asking for challenges, not for them to pass a turing test.
why, bother with anything, caves server a purpose, and walking is perfectly good way of travelling and electricity is just convenience.
Because electricity makes a HUGE difference, and better AI does not?
How would you know, how many games with smart AI have you played?
I can tell you right off the bat that chess is pretty awesome against smart AI and pretty bad against dumb AI.
Blizz already tried smart AI ... got QQ a lot, and dropped it.
hmm ... actually chess is pretty dumb AI, just alpha-beta pruning will beat 99.9999% of the population. Only Kasporov needs special treatment.
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
Strong != smart. Again, a simple algorithm like alpha-beta pruning can beat 99.999% of humans just because the computer can look ahead more. That is an very old very simple algorithm.
The point is .. making an AI that is challenging is easy in games. It is not necessarily to be desirable because where is the fun if the computer wins 99.99999% of the time?
And that's why you need a randomness factor so that the computer can screw up and give you an opening now and again, if you see it and take advantage of it.
If you miss that one, there'll be another one coming for various effect soon enough. Maybe.
That is even easier .. which is in all chess programs now. When you are doing the alpha-beta pruning, just put in RNG .. and occasionally choose a non-optimized option.
But again, the point is that being strong is trivial .. it is about making the game fun. That is much harder.
Of course that's true. That will always be true with any game. And it takes a special knack to be that creative.
But "fun" comes in many forms. And there's more to making games than just the fun factor. Especially persistent worlds. If you want them to really persist, you need it to change all the time in some way that adds "content" for the players. As an ongoing need for content, AI that functions to accomplish change and create new content is the only way you can keep up with those needs of the players. And if you're going to do that, it adds even more to the game as far as "interesting" if that AI follows the game world's lore and history, adding to that too. The AI design can do that on a basic level, and GMs can do that on a very big level if they have tools built to use that AI with it.
Have you played an MMO in the past decade? If they actually improved the AI to make fights more challenging the forums for that game would be filled with the endless tears of entitled crybabies begging for nerfs because it's too hard. God forbid you would actually have to learn to play your class properly and think a little bit while playing instead of just mindlessly mashing the same 3 or 4 buttons over and over again for hundreds of hours because every fight plays out exactly the same.
Have you played an MMO in the past decade? If they actually improved the AI to make fights more challenging the forums for that game would be filled with the endless tears of entitled crybabies begging for nerfs because it's too hard. God forbid you would actually have to learn to play your class properly and think a little bit while playing instead of just mindlessly mashing the same 3 or 4 buttons over and over again for hundreds of hours because every fight plays out exactly the same.
But... how do you really feel, kaiser?
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein "Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
There's a few major points to take into consideration when it comes to computer AI in a video game, the most important being that a computer never tires and never sleeps, and that a computer never makes calculation errors and always aims correctly even after dozens of hours of playing.
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
There's a few major points to take into consideration when it comes to computer AI in a video game, the most important being that a computer never tires and never sleeps, and that a computer never makes calculation errors and always aims correctly even after dozens of hours of playing.
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
Wait... wut? I think you are confusing pattern with precision/perfection.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein "Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
Originally posted by Jean-Luc_Picard There's a few major points to take into consideration when it comes to computer AI in a video game, the most important being that a computer never tires and never sleeps, and that a computer never makes calculation errors and always aims correctly even after dozens of hours of playing.
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
It actually easy. You can just throw a dice with math.random() and it will stop beeing predictable.
But does the majority really want it? wow has addons for players that announce predictable moves. Most use them. why? Cause they can't do it without. If you made it all random they would get wrecked hard and likely ragequit.
Same shit with AI, whenever games put out a really though AI the rivers of tears open. It's not that the tech is hard, it's just that a large part of the customers don't want it while everyone is still willing to play against stupid monsters.
tldr: stupid predictable monsters are simply more profitable.
Originally posted by Aison2 It actually easy. You can just throw a dice with math.random() ...
Now there's some sound game design!
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein "Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
There's a few major points to take into consideration when it comes to computer AI in a video game, the most important being that a computer never tires and never sleeps, and that a computer never makes calculation errors and always aims correctly even after dozens of hours of playing.
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
Wait... wut? I think you are confusing pattern with precision/perfection.
Well, I think there is a definite pattern to perfection.
Comments
AI is stupid because people insist on wanting to be able to solo everything, as every class.
If you designed an AI that was smart, you would have NPC's traveling in packs, where they would work together to surround the player, placing melee into LOS between the player and their ranged attackers, ensuring that some of them were always flanking. Basically, solo players would die like dogs, and then go whining to the forums about how horrible the game is.
So, if you want smart AI, you need to convince the dev studios that there's a good sized market for a group-only MMO.
Good luck!
So just gonna try and help you out.
I found one of the most fun ways to learn java programming wars Robocode. It's a game where you program the robots AI.
Also check out:
Designing artificial intelligence for games
Introduction to AI programming for games
If you want to check out some cool game specific stuff, Bethesda made the Radiant AI for the Elder Scrolls series. Or you could check out the Skyrim creation kit tutorials and learn papyrus scripting to do your own AI scripts.
If your really serious and want to start to learn programming I recommend Dietel and Deitel programming books which have been around for over a decade since I went to college for programming.
And Artificial Intelligence resources which has a ton of stuff.
Oh and also something really fun! A.L.I.C.E. chat bots. (This site advertises with digital AI chat bots) it's worth trying out tho and talking to it to see what it says! I mean come on after looking at this how can you not want to code an AIML chat bot!
"classification of games into MMOs is not by rational reasoning" - nariusseldon
Love Minecraft. And check out my Youtube channel OhCanadaGamer
Try a MUD today at http://www.mudconnect.com/Probably don't need that logical fallacy to argue against smart AI. To make a fair comparison you have to compare the best possible smart AI design with the best possible dumb AI design. (With the caveat that "best possible" also implies the design is feasible.)
Showing that smart AI wouldn't work with numbers which were balanced for dumb AI is not a compelling argument. Obviously for a fair comparison you'd have to assume the designers weren't complete idiots and actually tried to make things balanced appropriately for a smart AI game (which strongly implies numbers a lot closer to the player's own.)
Personally I haven't seen smart AI in any game which gives me the impression that it would be more fun in MMORPGs than dumb AI. Dumb AI is simply one of the rules of the game. Often it's not even a big part of the rules that establish the "puzzle" of any given encounter; mostly the rules you care about are the specific abilities the boss will use this fight. By keeping the AI simple, you can immediately move on to study the things you know are different each encounter. You don't have to worry about weird random oddities happening.
So I do largely agree with your implication that smart AI isn't going to be an improvement.
Also FWIW real-time dodging is possible and used in several WOW fights (not to mention GW1). A recent example being Brackenspore who fires projectiles at your group. It goes even further than that, because you actually want 1 person intercepting the projectiles sometimes (because they're AOE explosions and if they reach your main group and it isn't paying attention then that's a ton of raid damage.)
"What is truly revealing is his implication that believing something to be true is the same as it being true. [continue]" -John Oliver
Why wouldn't the player simply move to a more defensible spot, stay on the move, make a plan to isolate one at a time, or simply beat feet away?
I'm assuming here in this fictional MMO with AI that there's not a lot of static content and this particular encounter might have been with just 1 MOB, or 20. And that the makeup of this MOB group can be quite different on any single encounter.
And that, I think, is a big part of why many gamers want better AI. So that the encounters can be something much less predictable than today's games, both in makeup and in actions. And with that last part, "in actions", I think a randomness needs to be introduced to the MOB's actions. Rather than an AI that always does "Perfect-10%" it needs to vary, not just on the encounter, but per rounds of action.
Once upon a time....
I wouldn't call it dumb. Computer chess games are tactically very strong. They have all the common openings and end games down to a science. They are strategically weak though. Strange positioning can sometimes lead to the computer to make silly moves.
I frequently play Fritz Chess and get stomped pretty much every game. Once in a blue moon I do get to capitalize on an odd ball move that I doubt an average human player would make (let alone a Garry Kasparov).
There are certain queer times and occasions in this strange mixed affair we call life when a man takes this whole universe for a vast practical joke, though the wit thereof he but dimly discerns, and more than suspects that the joke is at nobody's expense but his own.
-- Herman Melville
Combat system would obviously need rethinking too, this is not change that affect just mobs.
And people cant get away from seeing it too narrow, its not JUST about challenge, its about upgrade of quality of game as a whole, where dog would act like a dog and elite mercenary would act as elite mercenary. Challenge on both of those can be tuned to whatever design goals are.
Along that, it would remove highly repetitve encounters, just go through WoW for example and go to some old raid. NOTHING has changed, that raid from 3 years ago is exactly the same, its throwaway content today. WoW can afford to do that but other MMOs (PvE ones at least) cant. Devs cannot churn out content nearly fast enough.
And lastly, it would allow for AI to make actual changes to game world, some of that ideas that EQN was supposed to bring on (although that game is highly questionable). You wiped out those mercenaries last time? No problems, wildlife has moved into their base. Thats lab you stormed last time? well, their experiments seems to came to life....
Oh yeah, the ultimate goal is having real AI, but until then at least some improvements can be done. Yeah, its not simple, its not something that will just come to life out of ether, but yeah, MMOs need something new, let those devs earn their money
Strong != smart. Again, a simple algorithm like alpha-beta pruning can beat 99.999% of humans just because the computer can look ahead more. That is an very old very simple algorithm.
The point is .. making an AI that is challenging is easy in games. It is not necessarily to be desirable because where is the fun if the computer wins 99.99999% of the time?
I agree that combat would change. It wouldn't just be tank and nuke with support. Those elements would still be there, but there'd be added decisions based on unpredictable "tactics". Those tactics, the way I think of it in my idea of AI, would actually be changes based on current actions and results but with randomness modified by a "morale" factor for the MOBs.
This totally removes predictable MOB behavior. It makes it seem like MOBs are thinking because they are changing up according to what's happening, yet they aren't always reacting the same way. The "morale" modifier I mentioned would probably be the defining difference between "elite" and regular MOBs.
The way I see AI being built is on a tier. Ranging from lowest basic survival instincts all the way up to superior intelligence. Each level adds actionable tactics as "choices", and socialized MOBs work as a group with more options. And all modified by "morale". Also, a socialized individual could lose all "morale" and break out of that group order and into basic individual "choice", usually total fear and the choice of just running away. Add the possibility of extreme plus "morale" where a socialized individual breaks from group command and decides to act on that MOB's own, usually trying to take over and lead that group of MOBs (leading an attack, taking over as leader of the group, leaving that group-part of MOB expansion, etc.). See, I think it's possible to create an AI where MOBs are constantly doing things in the world without players being involved. Seeking and building shelter and defense, moving in search of better food and shelter conditions, seeking resources, and generally advancing themselves as individuals or socialized groups. Think of it as making what appears to be advanced thinking AI but really simply a range of modified random actions based on circumstances.
On top of this AI would be a world that adds to choices. Locations (small and large) that are tagged for possible uses. So, as an example, in a dungeon there can be areas tagged for construction of defensive items such as picket fences, stonework walls, traps of various kinds, etc. There can also be areas tagged for a variety of uses such as choke points (for missile traps or concentrated fire) and other things. And each race or MOB type would have their own sorts of uses in each of these tagged areas.
All of this would be expensive to make and require lots of extra computer power. But I think it would create a much more interesting world. An ever changing world. A world that feels alive, dangerous, and full of adventure.
Once upon a time....
And that's why you need a randomness factor so that the computer can screw up and give you an opening now and again, if you see it and take advantage of it.
If you miss that one, there'll be another one coming for various effect soon enough. Maybe.
Once upon a time....
Serious question (based on the above statement) : How long does it take to actually implement new content (new quests, new characters, new stories, program it it all and get it in the world)? How long does it take for putting it into the world (I would guess the coding end) once the stories etc are already worked out?
As far as the AI goes... Instead of dumbing down content, wouldn't it be possible to set up tutorials to build up the players to more challenging content? (ie... dumb down the tutorials instead?)
Tough question to answer, but it's clear that content just can't be created fast enough. No one can do it, and everyone who does it says they see a problem here.
That would be a huge benefit if you could create AI that allows for the game world to add content on it's own by changing the world. But I don't see this ever being up to the needed standard but rather being a big help if done right. GMs setting MOB course-of-actions and building eventful plans up along with this sort of AI could be very powerful, in my opinion.
That last part, I'm hoping you'd like to have that one back.
Once upon a time....
Depends on way too many factors.
Generic quests using existing assets and destined for existing world environments can probably be pumped out rather quick. There's a lot more to it than that. Just for the creation of the quest text, items and basic gameplay alone you have Continuity, Adherence to lore, Game balance, Reward itemization among other factors to consider. If it is getting its own assets or its own worldspace, you're now greatly increasing the number of people involved. Voiceovers or cutscenes? That's more people and money, too.
The smarter the mobs, the more scripted the quest, the more elaborate each scene is that has to be created... the more teams need to be involved.
So, yeah, generic quests with complete disregard for economy, gameplay, brand, continuity or lore could probably be pumped out by the dozens every day, even then you're probably bound by build/release schedules. Beyond that, it all depends on the game's features, the quest's impact on existing gameplay, dev tools, finite resources, and how crazy you want to get with it.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein
"Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
The scary part of that is that for every dev team of 3-300 devs that works to create an AI to do that, there's a playerbase of hundreds of thousands of players that will do everything in their power to master it and game the crap out of it. Not saying it shouldn't be done, rather that is a very good reason why we haven't seen one released.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein
"Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
Elementary, my dear Watson. MMO AI is and has been the same for a long time because that is the norm and it is easier to just keep it that way then take chances on evolving it. That and it was/has been difficult for dev teams, tech wise, to make it possible.
The first MMO I ever played, Nexustk, had about the same mob AI as most games these days do. Except that it was 2d and the year 1998. Everquest 1 had the same thing, with one difference. Mobs would run away at a certain health % and bring back other mobs to wreck your ass. And pretty much every other MMO post-EQ1 used the same AI. WoW has the same AI, and every WoW clone in the last decade does too.
EverQuest 2 made a bold claim during its development, saying how it was going to introduce smart AI. The example they used was if an Orc spotted you across a valley it would then process the threat level. If it thought it could kill you it would attack and kill you. If the Orc thought you were to strong and it would die in a fight it would then run off and get help and then you would get swarmed by Orcs. I remember reading this with my buddies while at sea and thinking how fucking cool it sounded. And how it would revolutionize the MMO gameplay. EQ2 launched, and the AI was the same fucking thing it had always been. But it was actually worse, now mobs would reset at a certain distance away from their spawn point or their travel path. Making the world feel even less real and the danger level even less severe. You wouldn't have to run for your life and pray someone helps you or you get to a zone line before the Orc catches up to you and mercs your ass. Just run a few feet away and it resets back to full health and it is like nothing had ever even happened.
Honestly I think AI has stayed so stagnant because it is to risky to create something new and improved. If it fucks up and doesn't work then your entire game is screwed and hundreds of millions of dollars are flushed down the toilet. It is safer for investors and corporations to just keep pumping out the same old shit with a new skin and reaping as much profit from the traveling nomadic MMO players.
Then again the market is shifting and those fat cat companies are moving on and forgetting about the MMO genre. SOE is now Daybreak a much smaller version of itself. Blizzard gave up on Titan and is now going to MOBA town. Funcom will probably end soon. The asian market is steering more towards Sandparks. Cryptic, Turbine, etc.. are no longer relevant. Smaller companies are starting to kickstart things and try new shit.
And yet again! Daybreak is saying that EQN's emergent AI is going to change the game! Mobs will now actually have thought patterns and behavior. They will migrate and look for environments that suit their needs/wants. If attacked they will either pick up camp and move, or they will build up and retaliate. The days of a mob standing still until you get up in their face will be over and they will never keep spawning in the same place. But this is all words and knowing SOE/Daybreak's history it will never actually happen.
The most I expect to see out of EQN and other games in the near future is a mix of Rift and GW2. Which is the same old shit, mobs on a scripted path doing the same shit day in, day out.
Playing: Smite, Marvel Heroes
Played: Nexus:Kingdom of the Winds, Everquest, DAoC, Everquest 2, WoW, Matrix Online, Vangaurd, SWG, DDO, EVE, Fallen Earth, LoTRo, CoX, Champions Online, WAR, Darkfall, Mortal Online, Guild Wars, Rift, Tera, Aion, AoC, Gods and Heroes, DCUO, FF14, TSW, SWTOR, GW2, Wildstar, ESO, ArcheAge
Waiting On: Nothing. Mmorpg's are dead.
That is even easier .. which is in all chess programs now. When you are doing the alpha-beta pruning, just put in RNG .. and occasionally choose a non-optimized option.
But again, the point is that being strong is trivial .. it is about making the game fun. That is much harder.
I normally don't agree with Nari due to their solo console game mentality.. (Don't get me wrong, I like soloing on console games too, but I don't mix MMO's with solo games and expect the same experience).. Anyways.. lol
I did want to highlight the two areas that I agree with Nari on that apply to any computer game.. I've always suggested that AI's be smart and optimal (100%) .. However, do not treat all mobs the same in regards to their AI.. Some mobs will be almost expert in the terms of AI skill, and other mobs will make repeated mistakes, just like chess programs.. The trick is to interject a variety of mobs together so that it gives the player a unexpected and fun time.. Which is what playing a game is all about.. FUN..
Challenging AI
= risky AI for mostly for niche gamers, smaller community = nich game, less $$$$$
Simplistic AI =
Little risk AI for majority easy to satisfy 'follow the carrot' community = more $$$$$$
"going into arguments with idiots is a lost cause, it requires you to stoop down to their level and you can't win"
Of course that's true. That will always be true with any game. And it takes a special knack to be that creative.
But "fun" comes in many forms. And there's more to making games than just the fun factor. Especially persistent worlds. If you want them to really persist, you need it to change all the time in some way that adds "content" for the players. As an ongoing need for content, AI that functions to accomplish change and create new content is the only way you can keep up with those needs of the players. And if you're going to do that, it adds even more to the game as far as "interesting" if that AI follows the game world's lore and history, adding to that too. The AI design can do that on a basic level, and GMs can do that on a very big level if they have tools built to use that AI with it.
Once upon a time....
Have you played an MMO in the past decade? If they actually improved the AI to make fights more challenging the forums for that game would be filled with the endless tears of entitled crybabies begging for nerfs because it's too hard. God forbid you would actually have to learn to play your class properly and think a little bit while playing instead of just mindlessly mashing the same 3 or 4 buttons over and over again for hundreds of hours because every fight plays out exactly the same.
But... how do you really feel, kaiser?
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein
"Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
Once upon a time....
Wait... wut? I think you are confusing pattern with precision/perfection.
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein
"Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
The whole point of building any AI of any level is exactly NOT to do that. It's easy to make AI that does one thing the same all the time, harder to make it so that it changes up, does something "wrong", etc. But that's the point here, players are tired of knowing what's going to happen next and just beating the math.
It actually easy. You can just throw a dice with math.random() and it will stop beeing predictable.But does the majority really want it? wow has addons for players that announce predictable moves. Most use them. why? Cause they can't do it without. If you made it all random they would get wrecked hard and likely ragequit.
Same shit with AI, whenever games put out a really though AI the rivers of tears open.
It's not that the tech is hard, it's just that a large part of the customers don't want it while everyone is still willing to play against stupid monsters.
tldr: stupid predictable monsters are simply more profitable.
Pi*1337/100 = 42
Now there's some sound game design!
There isn't a "right" or "wrong" way to play, if you want to use a screwdriver to put nails into wood, have at it, simply don't complain when the guy next to you with the hammer is doing it much better and easier. - Allein
"Graphics are often supplied by Engines that (some) MMORPG's are built in" - Spuffyre
Well, I think there is a definite pattern to perfection.
Once upon a time....