Maniaplanet Forum

I think he is using the argument "humans are better" as a knock down for the AI in general. As you can see in strategical shooters, where you can give commands to you team members (e.g. the rainbow six series) it works out quite well if the AI has it's own limited set of commands (shoot at this, cover me there and and) in a short range around the player. But if you need them to do complex things over a whole map, than simply the actions, which you are needing to program for the AI, so that it can respond to nearly all kind of situations is enormous. This would be the case, if you try to code it for a single map you build (of course, standard routines would be needed to be added by nadeo themself) you only need to add map specific way routines. But think about the example (with map A being only jumping in a well while map B is a nice normal short combat map) in the gamestar article then you'll see how wide the possibility in creating different types of maps is. And as soon as you are able to create game modes with maniascript, the complexity to create according acting bots will increase dramatically.
Therefore i think, nadeo is first going to see how the players are playing the game and maybe add an AI in the late future, if they see that one game mode is dominat (maybe in the esports scene).

I think he is using the argument "humans are better" as a knock down for the AI in general.

And it's not working very well because AI is used for more than just emulating players. Which seems to be what his entire argument is based on.

But if you need them to do complex things over a whole map, than simply the actions, which you are needing to program for the AI, so that it can respond to nearly all kind of situations is enormous.

Yes, this is why Rainbow Six had you plan out the map beforehand, so when those AI squadmembers had to do something important, it was already all set up by you, the planner, in the editor. Much like a map designer would have to do for his map if he was planning on using AI or bots.

And as soon as you are able to create game modes with maniascript, the complexity to create according acting bots will increase dramatically.

And that's why you give the mapmaker and any subsequent modder the ability to control the actions of the bots. It's not like these are going to be dynamically generated AI. Even in the most basic, simplistic of scenarios for AI that I can think of still required a mapmaker or a modder to provide control zones and pathing. I don't see why this would be any different.

@Naimisrepus
Nice, Thank you for your input.

And what do you guys think of my stats idea?

Unit2209 wrote:And what do you guys think of my stats idea?

Hmm, i have totally overseen that part of the post ^^

A nice statistical section for yourself would be pretty nice,i think. Maybe you can make also a ladder with this stuff (e.g. player x had the longest jump with a rocket launcher). Then you guys need to build maps specific for the goal to get first in such statistics

See Halo 1
When you make a custom map you can add AI and they will run fine because of a waypoint and shoot point system. The AI will move to the waypoints but if there is a bad guy and a shoot point they will stand on the shoot point (that the player cannot see) and shoot. If there isn’t a shoot point they will just shoot while moving.

See SPORE GA
If you did play this then you know how easy it was to make a huge base filled with AI patrols and snipers and soldiers.

Why can’t it be as easy as that for this game? All of these games use map editors. (Hopefully Shoot Mania can be a part of that)

You guys don't know how to make a coherent argument. In one sentence you're saying that AI can be used for much broader things than a player-replacement, with you assume is what I'm proposing. And then in the other sentence you basically say the same thing I fear the AI will try to do (replace players), just using different words. Which is it? Is the AI merely going to be used for testing purposes or do you actually want it to be used in a singleplayer campaign? If it's the former, yes, the idea is feasible. You can have some rudimentary commands the AI can run through and therefore test your maps before you put them on the Shootmania Exchange. However, if you want the AI to be an actual substitute for a human player, i.e., in a singleplayer campaign, this is not the game for you. You cannot code the AI to be an adequate challenge while remaining adaptable to the sheer number of user-made maps that are going to come out. Well, yes, you can, but at that point it becomes exceedingly complex and the number of FloPS necessary to make it work becomes ridiculous.

And to address your earlier reply, namus, I'm not the one restricting the game. I'm all for AI--IF it can function in the following manner:

Is scalable (has difficulty levels)
Is adaptable (has behaviors, not just preset routes)
Can "communicate" with AI squadmates, rather than just running around independently
Can function smoothly and completely, no matter what kind of map is thrown at the code.

I don't foresee my request as being too gargantuan...most of those behaviors are present in any modern-day shooter. However, 99.9% of commercial modern shooters do -not- have an editor. The AI's behavior is scripted off of known maps that the company themselves built, tested, and then packaged. However, with SM, that is not going to be the case. 90% of the maps are going to be -user created-, and therefore, if there -is- AI, it's going to have to function under any circumstance, even in circumstances that Nadeo didn't predict--if the craziness of tracks in TM1 is any indication. That's a big hill to climb. Another one is the code itself. If you're going to leave all, most, or even some, of the behaviors to the user to determine for the AI reactions--like when it pops up behind a rock to shoot somebody, or if it uses cover at all, the only efficient way I know how to do that is through scripting. And not everyone can script, or script well. And when there are holes in a script, that's essentially game code, and the CPU has to fudge, or freak out. Usually, it just freaks, with any one of a thousand results. If the error is minor enough, the freaking may go unnoticed by the user. However, if there's a bigger hole in the code, any one of a number of things could happen, from your bots breaking, to the whole game going CTD. And, as a final note, Namus, if I was satisfied with two people standing in a room throwing rocks at each other, I'd've never progressed past doom, thank you very much. If that's all you see in a commercial shooter, what are you even doing here? You fail to see the forest for the trees, and that, I think, is where you fall short.


@Unit's last post: The waypoint and shootpoint system are already integrated into the game...same with Spore's system. I'm saying that I don't think SM will even go that far, especially if you want a flexible AI that can be modified. It'll be up to the user to code behaviors, and then how and where those behaviors work. And that brings me to my previous point of not everyone's a programmer.

AI/Bots would be awesomely fun to play in huge maps like Battlefield 2 maps.
I'm really fond of this kind of playing, in a huge environment.

@trackmaniack

Data, Facts, Info, You need to tell me these because you haven’t giving me very much of them. I was using Halo and Spore as examples to show you that crazy edited maps can still support AI. You also still seem to think that the AI will crash the system and cannot be used. If they do use AI and there is a major coding error that crashes the system don’t you think they will FIX it? And I’m not trying to replace Humans or PVP multiplayer we want an awesome editor with the option to use AI.

I also don’t want a singleplayer...I want to make one.

My data, facts, and info? Playing years of multiplayer games, both with mods and without, and seeing everything break at least once. No one's stopping you from making AI, making bots, or even making an entire singleplayer game once SM comes out. If the maniascript editor is going to be more robust than what we have now (and it will be), there's nothing stopping you from essentially reprogramming SM into an entirely new game. My issue and the main reason I'm arguing with you is that I've seen cases where it doesn't work in games, just as often as seen cases where it does. I don't want you to act like it's the only 'right choice' to put AI into SM, or that you'll be disappointed if you don't...because SM is still going to be a kickass MP game, no matter how you slice it. And my choices of liking the so-called "standard hitscan bullet shooters" doesn't make me any less valid, it simply means I like what I like. You like what you like. Hopefully everyone can be happy with the new SM.

Trackmaniack wrote:You guys don't know how to make a coherent argument. In one sentence you're saying that AI can be used for much broader things than a player-replacement, with you assume is what I'm proposing. And then in the other sentence you basically say the same thing I fear the AI will try to do (replace players), just using different words. Which is it? Is the AI merely going to be used for testing purposes or do you actually want it to be used in a singleplayer campaign? If it's the former, yes, the idea is feasible. You can have some rudimentary commands the AI can run through and therefore test your maps before you put them on the Shootmania Exchange. However, if you want the AI to be an actual substitute for a human player, i.e., in a singleplayer campaign, this is not the game for you. You cannot code the AI to be an adequate challenge while remaining adaptable to the sheer number of user-made maps that are going to come out. Well, yes, you can, but at that point it becomes exceedingly complex and the number of FloPS necessary to make it work becomes ridiculous.

After this paragraph, I'm having trouble believing that you worked with any sort of AI system at all and I'm not even sure where I should begin correcting you.

Trackmaniack wrote:And then in the other sentence you basically say the same thing I fear the AI will try to do (replace players), just using different words. Which is it?

I'm saying you believe this, and you're wrong. which is something you admit to. You seriously think, seriously, seriously seriously think that the AI is in any way a replacement for players. This is wrong. It's a supplement for the map maker. AI can be used in a multitude of ways. They can range from being extraordinarily complex (like AI teammates, AI adversaries) or being incredibly simple (nothing more complex than "see player, kill player). Honestly I'm not sure what you think of when you see the word "AI". They're not replacements for humans. Even AI in strictly-verses multiplayer games is just there to curb boredom or act as a pacifier.

Trackmaniack wrote:However, if you want the AI to be an actual substitute for a human player, i.e., in a singleplayer campaign

No, I don't think anyone said that. They just suggested the tools for mapmakers to implement bots.

Trackmaniack wrote:You cannot code the AI to be an adequate challenge while remaining adaptable to the sheer number of user-made maps that are going to come out.

That's why you give the tools for AI to be successful over to the map makers. I don't think anyone was ever suggesting anything different.

Trackmaniack wrote:but at that point it becomes exceedingly complex and the number of FloPS necessary to make it work becomes ridiculous.

FLOPS has nothing to do with anything when regarding AI. AI is usually made using dynamically generated coordinates and waypoints based on zones and pathing. I don't know where this even came from and has nothing to do with the conversation at hand.

Trackmaniack wrote:And to address your earlier reply, namus, I'm not the one restricting the game. I'm all for AI--IF it can function in the following manner:

And there you go assuming AI has to meet your specific designation.

Trackmaniack wrote:Is scalable (has difficulty levels)

Irrelevant. Should be up to the Mapmaker what kind of AI he wants in his map and how hard it is.

Trackmaniack wrote:Is adaptable (has behaviors, not just preset routes)

Okay, I can agree. It shouldn't be hard to make different AI types. Or at least give people the tools to modify or add on to those types.

Trackmaniack wrote:Can "communicate" with AI squadmates, rather than just running around independently

Irrelevant. Depends on the type of AI. Zombies aren't going to communicate. Soldiers are going to communicate and will likely do so in a way that will not involve any visible communication.

Trackmaniack wrote: Can function smoothly and completely, no matter what kind of map is thrown at the code.

Again, depends on the map maker. A novice map maker will likely have problems making AI that seems smooth and natural.

Trackmaniack wrote:I don't foresee my request as being too gargantuan...most of those behaviors are present in any modern-day shooter. However, 99.9% of commercial modern shooters do -not- have an editor.

This is so incredibly wrong that I'm not even sure what universe it came from. ALL shooters, even most non-shooter games, have an editor of some sort. It is how the heavy work of games are made. It just so happens that the editor is never released to the general public because it would require heavy hacking to make any additions functional anyway.

Trackmaniack wrote:The AI's behavior is scripted off of known maps that the company themselves built, tested, and then packaged. However, with SM, that is not going to be the case. 90% of the maps are going to be -user created-, and therefore, if there -is- AI, it's going to have to function under any circumstance

You just proved my point. Give the users the ability to script AI functions and place AI waypoints and objectives and the problem solves itself. No one is expecting AI and bots that are 100% functional on warped and ill-defined. This way, the AI bots work under all circumstances granted the map designer planned for it.

Trackmaniack wrote:. And, as a final note, Namus, if I was satisfied with two people standing in a room throwing rocks at each other, I'd've never progressed past doom, thank you very much. If that's all you see in a commercial shooter, what are you even doing here?

Because this game has the potential to not be another generic commercial shooter. This appeals to me.

Too many designers now feel that adding bullets and guns to an unworkable, hardly playable system is what makes a AAA game. Nadeo understands that user-created content is the core of many highly successful workable systems. If an FPS comes out that is based on user-created content with a workable editor, I want to be all over that because it sounds awesome.

Trackmaniack wrote:@Unit's last post: The waypoint and shootpoint system are already integrated into the game

This isn't directed at me, but I want to point out that you don't know that. At all.

Trackmaniack wrote: I'm saying that I don't think SM will even go that far, especially if you want a flexible AI that can be modified. It'll be up to the user to code behaviors, and then how and where those behaviors work. And that brings me to my previous point of not everyone's a programmer.

Then not everyone should be making maps that have AI in them.

You're acting as if making AI is some dark and shadowy art that few understand. But it's not. There were systems for making dynamic user-specified AI in an editor back in the Warcraft 2 days. If Nadeo just gives us basic AI actions and maybe some templates, maps with AI integrated in them will come out and be great. End of story.

Trackmaniack wrote:My data, facts, and info? Playing years of multiplayer games, both with mods and without, and seeing everything break at least once.

Your experience with MULTIPLAYER games sure is a testament of how much you know about AI in video games.