Has AI programming stagnated in video games?

What do you think?

Other urls found in this thread:

askagamedev.tumblr.com/post/76972636953/game-development-myths-players-want-smart
alumni.media.mit.edu/~jorkin/gdc2006_orkin_jeff_fear.pdf
twitter.com/NSFWRedditGif

Realistic AI would snipe you from half way across any map with pinpoint accuracy and basically not let you do anything ever

It is in fucking place since damn F.E.A.R. At least for shooters. I don't know any recent title that would be praised for AI.

The only game I've heard praised for A.I in forever is "Alien: Isolation."

One title, in forever.

Just one title.

it has.
and it is mostly because even though computers have gotten much much more powerful over the last 12 or so years ALL of that extra power has been pumped into graphics

I read on gamasutra I think that the game industry doesn't really try to communicate with AI experts.

I guess they feel the current level (that's been almost the same for a decade) is enough for players these days and is not worth any investment.

Yes developers, continue to ignore the thing that actually makes gameplay compelling: the enemy's A.I.

That and F.E.A.R which have really impressive AI of enemies. Those fuckers really try to flank you and try to trap you into corner.

You know how some people complain about pros in online parts of a game face fucking them constantly ? Yeah, I don't think developers want that as the single player part of a game, scare customers off. Then they'd complain about having to use a lower a difficulty.

Just my 1+1 cent.

Just give the players AI teammates who have "Alien:Isolation"-tier AI.

>we will never get STALKER with the prerelease AI that would literally finish the game

Well metro certainly didn't have Alien Isolation tier, I did find it funny that at some parts it felt like YOU were the escort NPC. It was nice seeing an AI fire and actually kill instead of doing it for show.

No, the devs are too lazy to make good AI.

I honestly don't see why further AI improvements are necessary, what the fuck could that possibly add

You can't market AI, game companies blow their cash on artist and 3d modellers to make their game stand out.

Why waste time on something that won't generate any money.

Shame

The last game I played with good AI was Halo Reach.

AI is only as good as you know how to use it.

Doom is a good example, its AI is simplistic as fuck but they still proved a challenge given ho they're used in the map design

Uh, actual challenge.

I want enemies who fight so well I mistake them for human players.

That's the devlelopers' fault.

You can make money off A.I, as it dramatically improves the gameplay.

You blow all your money on graphics, you make a shallow game with no staying power.

The only place AI matters are symmetrical games, where you and your enemy play by exactly the same rules, ie. any multiplayer game, in which case why not just play against a human opponent.
In a typical single player game enemies act on vastly different rules to the player. They're not designed to be smart they're designed to be convincing npc characters who are easy to kill

Name a game where better AI would dramatically imrpove gameplay

They specifically don't want staying power, unless it's subscription based.
They want people to be dazzled just long enough until they buy the next one.

CIV 5.

>I read on gamasutra I think that the game industry doesn't really try to communicate with AI experts.
AI isn't really a big field and there isn't much AI researchers know that game developers don't that you can apply to video games

It depends on what you're talking about. What is "good AI"? It's really easy to write enemies that just calculate the most efficient actions and pull those off, leading to unfun games. Good game AI would be AI of enemies that do things that result in interesting scenarios for the player to play against, which is more of a game design question than an AI related one.

I completely agree with you, but single-player strategy games like that that make good use of AI are a minority, most games are action games or multiplayer games

>enemies jumping and shooting
>enemies camping spots you can't normally get to

No. Ai tech has greatly improved, but games don't need complex intelligence, they need scipts and variables that can easily be changed and debugged.

Realistic does not mean they don't abide by the rules of the game.

Killzone Mercs has the best AI in a console shooter and should have been the console shooter of its year, however that went for Bio Infinity, and its 90s brain dead enemy AI

>Ai tech has greatly improved
It really hasn't, just because a google bot beat someone at go, it's still using the same AI techniques that have been known about for decades

No it hasn't, there is no "linear progression" of AI in video games. AI is custom made to present challenge to the player that is customized to the game. It is not something that can truly stagnate, no more than art can stagnate, though less care and attention can of course be payed to making a great artstyle in favor of making one that's just passable.

Furthermore specific subsets of what could be considered AI have even improved significantly, such as pathfinding due to more standardization of algorithms and coding practices. Many current engines also streamline making more detailed AI tasks and flags available, as compared to the old days when much more had to be made and accounted for by hand.

The entire "Civilization" franchise.

I can't coordinate at all with nations I team up with, especially in wartime.

Barbarians should be cunning ambushers, a true thorn in my side (as opposed to morons who attack when I have overwhelmingly superiority on defense).

Realistic ai is exactly the opposite. It's extremely easy to write an AI with great accuracy, it's hard to make an AI that has a realistic aim, realistic pathing, realistic indecision.

Yeah this, basically.

In my eyes, good AI are opponents who fight like humans. Our strengths, and our tendency to make mistakes.

I'm not interested in perfect adversaries, I want AIs that simulate the experience of fighting a human opponent.

>Psht, computers aren't really better now compared to the IBM 608, it's all still based on the same techniques that have been known for decades.

why dont you just fight a human opponent

Dumb weeb, but good answer.

No, people just don't want intelligent AIs in their videogames

What about Starcraft 2? I don't ever play against AI opponents but last I saw there were tons of customization options for how the AI players play.

>AI wins a game that was literally thought impossible for computers to beat humans at
>i-it's not a big deal

stalker did a good job of it despite a few issues

>It's the AI accuracy anecdote

AI don't have to learn and that's physics you're talking about in regards to sway etc. which is outside the realm of AI, simulating tiredness and weapon recoil etc.

AI aiming is perfetly fine with about 6kb of memory assigned to it
What's not okay is that sole 6kb being assigned to AI in games.

All games need for good AI is to establish a playbook for AI - that's fucking it. They're not academic, they're not learning the meaning of life, they just need to actually know the fucking rules and play them.

Here's an example: If an AI has what could be considered surplus armaments, e.g. they got your standard gun and side-arm, but oh shit sweet they're full on grenades AND a rocket launcher, and it doesn't seem like there's any immediate threat that needs it, AI can overrule the "never shoot explosives at infantry" rule because of hype.
This translates to AI getting an edge and going fucking mental with it, firing off a rocket into a room to try and blindly catch players out. That comes across as a fun AI.

What doesn't come across as a fun AI is one with a cone stuck around its neck while it follows an invisible line straight through the map on a train track snapping to full auto fire whatever gun it has at the time.

>we will never get a game with FEAR's AI again

Because I want to fight dozens of smart enemies at a time in missions.

I want to be able to say I'm a good video game player, and feel I've legitimately earned that claim.

still not seeing why you don't play a multiplayer game here

>Payday 1 & 2
>Just throw more at them fuck iunno

Could also just have specials use regular SWAT as meatshields while they try to get into position to rush you all at once, but nah fuck that bro. Consoles got no power for that.

The people who thought that where being remarkably short sighted then, Go is still pieces on a grid which is easy to compute, just not as easy as Chess is where you can predict entire end games, it's still built on the same neural network technology that's been around for decades, just applied in a smarter fashion

Most people in multiplayer games are utterly retarded, AI being somehow even more stupid is pathetic.

With decent AI you could just do drop-in co-op vs. AI and not give a shit about the enemy team ragequitting or hacking.

Honestly just study how human players aim in multiplayer matches.

That's the best way to replicate realistic accuracy. Assign skill levels by percentage of occurrence, then distribute those skill levels randomly.

10 percent of the AI guys will be excellent shots, but which specific AI guys are the good marksmen will vary each time the mission is loaded.

Nobody else thought of this? I have a math disability and I thought of this.

Because teams working as a team are quite rare.

You hit the nail on the head.

AI players have the inherent coordination to make it seem like you're fighting a team of human beings.

Fighting a group of actual human players is akin to fighting a disorganized mob.

Well that user is partially wrong, it's pretty easy to simulate human inaccurate aiming, games do it quite frequently, human-like decision making is the difficult part

I have't played FEAR yet, what was so special about its AI?

this is true, imagine having to actually go crouch or prone instead of sprinting and hip firing, half the audience would quit

askagamedev.tumblr.com/post/76972636953/game-development-myths-players-want-smart

Absolutely.

>AI aiming is perfetly fine with about 6kb of memory assigned to it
I like hos you said that dumb shit with a straight face, like, you totally believe it makes sense.

Typically humans in matches tend to make decisions based on either various objective goals or different emotions.

You get the irrational, impulsive decisions when feelings seriously override the objectivity. You get the wrong decisions when a bad call is made trying to fulfill an objective goal.

Often you'll see both. A mistake followed by an emotion-driven reaction.

Skyrim followers did this.

Give them a magic stick and they spam fireballs like crazy

>AI that has a realistic aim, realistic pathing, realistic indecision.
A realistic military trained person will kill you in a second. But you don't want it. You want 2/3 of the bullet to graze your feet to feel like you are a super hero.

It's not my fault most players want things easy.

I don't understand any player who doesn't want to be continuously challenged.

all game makers use rules based programming for ia since forever because it has the best fun/perfomance impact ratio
if you find something better feel free to contribute

realistic ai does not mean fun game
realistic ai takes a lot of power for nothing
realistic ia takes a lot of time to do for something only autistic faggots on 4 chan will notice


You want to play a killzone level with 15 realistic ia? play the multiplayer 15 people agaist you alone
come back here and tell me how fun it was

I said I wanted to fight a human player, not a human soldier; lol.

RTS games AI is terrible in general. they tend to cheat with resources and micromanage several units at the same time when that is impossible for humans

the hard part about making ai make decisions that seem human is usually the context. the best example is when they talked about the oblivion AI, where somebody would kill their spouse to steal an apple because they were starving to death. Considering every single factor and weighing them appropriately is extremely hard

AI doesnt need computation power, it just needs a lot of effort and detail.
Doing actual high level AI stuff like google or deepblue is beyond developers competence.

If you only fight weak AIs, you can't call yourself a good player.

>They're not designed to be smart they're designed to be convincing npc characters who are easy to kill
You forgot to explain why this is how it should be without exception. I for one like challenge in my video games, that's the point of games when you cut the frills, really.

People want better AI because they want the game to be more challenging without just inflating enemy stats.

This, incidentally, is why people are playing multiplayer more.

It gives them the challenge AI fights are currently lacking.

Nah!

I work as a developer in the field of AI.

The vast vast vast majority of the field really has nothing at all to do with what the layman would consider AI. It's a whole bunch of individual niche disciplines, languages, and products that have been lumped together by idiot businessmen and marketers who wouldn't know the difference between Johnny 5 and a toaster.

Further, is absolutely correct. "Progression" or "Stagnation" in game AI (which would more accurately be called enemy behavior) isn't really possible because each game calls for different behavior routines.

FEAR doesn't have that impressive AI, it's actually pretty mediocre. What's so good about fear is the level design and placement of the enemies. alumni.media.mit.edu/~jorkin/gdc2006_orkin_jeff_fear.pdf

Yeah, because apparently shooting cardboard targets is really the more entertaining option when compared to ones that try to shoot back and be competent.

You guys have schools right? Organize your field. Put method to the madness.

It's a mix of the soldiers being smart and using stuff like grenades and corner tactics and the maps being good for them to use. You cannot stay in a single place in FEAR unless ou want to die because the fucks will go to attack you from the back.

Hearts of Iron II and III allows the AI to take control over certain aspects. Really, Grand Strategy is *the* genre that benefits from good AI.

Inflating enemy stats is a sorry way to boost a challenge. I hate losing to a stupid enemy who I only lost to because its health doubled.

>play against maxed ai bots in killzone 2 before i do multiplayer
>dominate since players are not as good as computers

>come back here and tell me how fun it was
Quite? Forces you to think rapidly and develop skills. Far more rewarding than a duck shoot.

Correct.

Through challenge, you are forced to improve as a player.

any tips for a finance major with stats/data analysis background to get into the field? Is it nice? Is it worth it?

AI shouldn't necessarily be realistic but it should be interesting

And it should absolutely never feel dumb or easy to outsmart

It's less that the field is disorganized and more that a million totally different disciplines are considered AI

In much the same way that enemy behavior in games is (erroneously) considered AI

Both developing neural networks and building semantic analysis engines/scripts is considered AI work despite the two being about as far apart as is possible.

There will never be good AI in games, you think its a good decision to have AI being more intelligent than the retards playing?

>AI schools
I bet you're one of the retards who thinks AIs are going to gain sentience and take over the world aswell

Why would developers make enemies more difficult when they're trying to make all games easier and more approachable? QTEs, aimbot abilities, magic wallhack vision, "Fear takedowns"...

Don't blame devs, it's the casual audience that's cancerous.

If you want challenge play a multiplayer game. It's really that simple.

>In much the same way that enemy behavior in games is (erroneously) considered AI
It is AI, even if it's a primitive AI

Screw 'em.

Casuals have plenty of games.

I want games for the people who want to be challenged as players, especially experienced players who mop the floor with weak AIs.

Is that a case study in some sort of education course? Where can I find it?
Why isn't this level of quality and in depth analysis in high schools, rather than make hello world, display a string, make an array, make a loop...

>I work as a developer in the field of AI.
Huh neat, into that newfangled recursive "neural" net stuff, or more traditional type approach? Either way I imagine it must be fun, or at least busy, times with all the ways AI is becoming more and more pervasive.

sophisticated enough AI could make a game infinitely replayable, even long after the servers are down and the multiplayer is dead.

No they wouldn't. Most shootouts are just trading inaccurate potshots from behind cover that usually miss.

I wish I had some tips for you but I really don't. I got into the field because I knew someone who knew someone who had developed a powerful semantic analysis engine and I was smart enough to learn the proprietary language(s) and computational linguistics involved. I didn't even finish college.

It's very nice for me because I happen to be one of maybe three people in the world currently capable of building anything with the engine.

If you can get an internship with one of the fortune 500 tech companies that'd definitely be a start, I suppose. Seems to me with your degree and background you'd be a good fit for marketing and strategy; there's a dire need for sales and marketing personnel that can actually understand what is and isn't possible and what is and isn't difficult. It's a complicated field.

>Huh neat, into that newfangled recursive "neural" net stuff, or more traditional type approach?
you're a fucking retard

Is that Lisp?

Terrible combination of colours/background, almost unreadable.

Wow, rude.