Let's talk about AI

I'm going to develop a game with an AI that bans ppl for using spoilers on an imageboard.

sorryfortriggeringyourautismofellowanoniwilltrytoavoidthatnexttime:^}

Because programmers degenerated, all the good ones moved to more lucrative and less stressful jobs so you have smoothbrained millenial soygoblins and pajeets who can't program for shit.
Technology is stagnant while subhuman niggers don't learn to optimize their spaghetti code because

Audio engineering was better 20 years ago
here's 1998 sound engine.
Also bear in mind that the more complex AI you make, the easier it'll be to abuse it, hitman is a good example of it.

Well yes, but you gotta understand suits in video games are retarded in general.

A good example of this is the pre-release footage of The Last of Us vs the final product. In the pre-release game the AI was super intelligent and could arguably rival a human. However according to rumors AI was dumbed down due to suits and devs at sony worried about the games being too hard for casual gamers and the uncharted crowd. So what we got instead was a movie game.

...

The prerelease footage was so obviously staged for that piece of shit. They never made good AI and never had any plans to.

True, I don't think modern naughty dog has the talent to make such a system. Especially since all their games play themselves, regardless my point is that the reason we haven't seen any real advancements in vidya AI is because the people with the money don't want to invest in single player, favoring into investing in microtransaction systems and online modes. 2K even admitted as much in their messages to investors.

Agree on that, but tricking suits on thinking memehard=good would help push for a better AI? The means justify the ends
Think about he souls games, how many casual go around saying they beat the hardest game ever? How many cucktaku article state X us like the dark soul of Y

I think sound design is pretty underrated. Typically most good games have something memorable sound-wise whereas mediocre games do not.

I remember before the release of Red Orchestra 2 the developers were touting a situational awareness mechanic that would visually alert you to any threat-worthy noises that happen in proximity. They never did implement this and they never allowed their sound engine to pick up the slack.

The biggest disappointment about that game is the sound design.I think it's what makes it a mediocre game.

Psh, why go to all that effort when throwing players into a fancy rat maze and having them butt heads will sell better?

Well aren't they stuck with sport games? What the fuck do you sell for microtransi
Right

Then single player games like butterlord or hitmen?

My man. Some of my favorite moments involved an excellent aural element.

god i missed using realbot for cs 1.6. by far one the best i've used.
now bots are just rush machines if any even get put in. i miss my customizable bots and don't have the brains to make one myself

Nah I don't think I'd work. The reason for that is that youtubers already shills games with weird quirks and mechanics in them. A good example of this is Cuphead, it's a niche game that's basically a bunch of bosses. a few levels and plays like Metal Slug. I don't think Cuphead would do nearly as good if it wasn't for the Xboners dire need of exclusives and youtubers like Pewdiepie shilling the fuck out of it. Now while Cuphead was a huge success sales wise, most publishers and AAA developers see it as a failure because it didn't implement an online mode or micro transactions. It sold 2 million, but it didn't make 500 million dollars.

Right now the video game industry suffers from having bad business people in charge of these companies and relying on market research instead of just looking on the internet and taking the gut punches. Additionally SJWs also control the market research teams which is why they have so much power in the industry. If anything we should be attacking the market research and providing actual factual evidence to these publishers rather than the lies they get from the market research departments.

Either way the reason why we don't have proper innovation in games is a number of factors. Those factors being publishers being obsessed with market research, casual gamers buying the latest sports game, CoD and GTA while also buying the micro transactions, SJWs in HR departments ruining the careers of developers by blacklisting them, the SJW press teaming up with market researchers to try and profit off the industry, business men who don't care about long term growth and only what's in their pockets, "community managers" who have no idea what the fuck they're doing, etc.

Like I said, the developer teams who want to improve the AI don't have the budget to do so. most of the money is spent on making the best graphics for sports games rather than making the game with the best gameplay and innovative systems.

That's because anyone capable of coding good AI is no longer working in the vidya industry. In many cases AI have degraded in quality not increased.

Nah, I don't think that's necessarily the case. There's still good people in the industry, especially coming from Holla Forums. The issue is one of finances and social standings. These young developers have talent, but no one to guide them. Like when you see publishers having creative input rather than just managing the money, time and budgets then you start to see issues with the projects.

Because all the talented people has left the business.

Even the fucking sound design has degraded since Windows XP times, sure, Microsoft was partly to blame but Microsoft ain't in charge of coding your videogames.

Red Orchestra Ostfront in comparison to Red Orchestra 2 is a fairly good example. In Ostfront they didn't have fancy recording equipment but they still did a extremely good job. Sounds may not have been realistic but they felt right, submachine-guns sounded coarse, everything sounded coarse. Artillery especially stood out in Ostfront if compared to artillery in Red Orchestra 2. In RO2 when artillery is falling on you it just sounds like a minor inconvenience, while in Ostfront it is completely on purpose extra loud and bit coarse, just so that it's borderline painful to your ears.

While there are some talented and properly learned individuals making vidya, the number of those is dismally small in comparison to developers that produce substandard code.

Fucking kill yourself.

People don't actually want smarter AI. They want patterns they can learn and adapt to. It's part of the reason why DOOM is so good, you know exactly what every enemy will do/is capable of.

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

okay

You post a Warhammer 40k image. You and I surely can agree that the sound design of the first Dawn of War takes that game a tier above it's sequel in terms of overall quality. Everything about it is so much more interesting - from unit dialog responses to the soundtrack. Another good example in my opinion.

Play Decent.

You mean Descent.

Artillery fire is downright terrifying in Rising Storm 2: Vietnam, especially when it's raining down directly on you and you know your only chance of survival is finding cover really fast.

Gunshots are pretty great too.

Did someone call for an Earth' Shaker?
Three
Two
One
F I R E

Because for most genres the AI is good enough.
Realism is not fun. It's not hard to create realistic AI which would see and hear you from realistic distances.

I recall an article in a which an FPS dev talked about how when they implemented better AI schemes, players would feel like enemies were being randomly spawned behind them instead of reacting to actual environmental cues indicating player presence and sneaking up on them

So it's not just that programmers dropped in quality, crippling the ability to produce good AI but also, gamers dropped in quality, crippling the ability to even recognise good AI.

I'm convinced most playtester complaints and just there to justify scrapping something that they think would take too much work. You never see them complain about unskippable cutscenes for example despite playing shit on loop for days.

...

I think with good communication from your friendly AI, that wouldn't be a problem.

The main issue is, no one in reality fights alone, because a single person is easy to flank and surround.

It's because it's all done with behavior trees, which if you want anything not awful requires a lot of tedious coding. Some are coming out with neural network (true AI) AI. Elon Musk has one for the least worst ASSFAGGOTS called OpenAI. Hooktube embed is video of it.

I agree completely with you. Good sound design is such an esoteric thing too. There's no objective measure I know of and you won't find any instructions anywhere to do it right. Who today even thinks about things like how well different sound effects like attack, enemy death and music mix together, yet they can make or break an excellent game.
It's not surprising you see so little care taken for these considerations today when the average consumer doesn't know any better, and actual game enthusiasts are plagued with gameplay only fags that actually feel proud of having a lack standards. Don't get me wrong, you can't have a fun game without good gameplay, but all the other things are what elevates it above a mere pastime.

I call horseshit on OpenAI. Not because of it's merits but how overly praised it was for simply being able to do 1v1s with one of the most simplest heroes to play in a game all about drastically different characters. It's certainly a good starting point but this is simply the chess AI shit and making out that the AI can beat anyone in a contained environment is just fucking hilarious.

It's also a real time game so it can abuse it's virtually instant reaction time just like in fighting games.

It only has to be a starting point- this one was done in 6 months, so imagine one with further cooking.

If this is "revolutionary" AI, AI will never be a serious thing.

you're not really gonna be satisfied unless we use a neural network in place of AI
and we're far from being able to do that well, let alone efficiently

Since you insist on both wasting dubs and being a retarded mongoloid, have a lecture and educate yourself. These convolutional neural networks are essentially the same as the human brain- the revolutionary part right now is that they exist at all, even if right now their complexity is that of a toddler.

Check out AI War, I haven't played it but remember hearing that the dev has poured a lot of time into the AI in the game and it's leagues better than any other RTS AI. IIRC he accepts demos of any victories against it on the hardest settings and will tweak it to be able to counter it in future.

I call bullshit. In his own anecdote he talks about how his AI outsmarted players when it didnt, the players outsmarted the AI, they ran into hallways and the AI didnt react intelligently. What a stupid blogpost. Also saying players hated it because the AI "outsmarted" them is just typical lazydev. Anyone can program a bot to snipe across the map and headshot someone. Its like his team was trying to make good AI and stopped halfway through because…..why exactly?

It's not just that the hero is simple to play, the mechanics make it virtually impossible to lose in a mirror matchup once you get the slightest advantage. He gets a damage boost for every unit kiled but loses the bonus once if he dies, so if it kills the human player even once it gets a huge advantage that can't really be overcome since it can deny it's own units with the increased damage and keep you from ever getting bonus damage.

I guarantee on literally any other hero in the game it wouldn't be able to do nearly as well even with all the other bullshit stipulations they throw on the match.

wherefore doth I even liveth

Why was this the only game that did this.

is this dragon's dogma?

GalCiv2+3 are the best there is, and they're not that great in fact they're pretty bland to play against after a while.

Because it's so hard to tell that block 1,1 and 1,2 are different than block 2,1, and 2,2

Nigger, Dragon's Dogma has 9 prebaked AI behaviors, with 5 of them being shit and the other 4 being just about who to target first (strongest enemies, weakest enemies, ranged enemies, etc). The inclinations go from 0 to 1000 in scores. The game arbitrarily decides which retarded events raise one inclination (hurrrr durrr asking for help always means medicant it's not because I want an enchantment) while slightly lowering all the others. The closer an inclination is to 1000, the more likely the pawn is to select a script from that inclination.
That's it, that's all there is to it, they don't learn shit.

It is for a computer, you don't understand because you have millions of neurons and even more connections, but this might have just thousands. The average male brain has 86 billion by the way.

They definitely learn, but the degree they learn is vastly overstated by some people. They have certain bits of information that they learn if you defeat enemies in certain ways (wolves are weak to fire, Saurians are weakened if you cut off their tails, etc.).

But, yeah, that shit like "I taught my pawn to play catch!" is bullshit.

No, its dark souls.

honestly vidya AI reached its peak 10 years ago and there hasn't been any significant developments in the field since then.
games like crysis, stalker and f.e.a.r are about as good as it gets for shooter AI (and RTS AI bots have always been as dumb as a bag of hammers)

I'd say its probably because AAA games swung heavily towards being multiplayer oriented around 2006/2007 ish so developers felt more justified in spending most of their time working o n the multiplayer and just leaving the singleplayer as a narrative driven shooting gallery where you mow down waves of stupid mooks to make the story keep moving forward

Is Mirai Akari a real AI like Kizuna or is she just a person? What about Kaguya Luna?

That's just bestiary knowledge.
Every monster has three stars in the bestiary, gaining one star unlocks one of these little enemy-specific scripts.
These all all bound to the utilitarian inclination.
You gain starts by defeating X amount of enemies, reading an enemy tactics scroll or pulling off the tactic yourself in a way that the game registers it.

You more or less just described learning behaviors.

Yeah. They gained knowledge. The shorter way of saying that is "learning." It's not glamorous or fancy, but it fits the definition. 99% of all machine learning is going to be the equivalent of setting a flag somewhere.

Kaguya Luna isn't an AI, AI's are incapable of huffing glue.

Good AI is not fun to fight. It just fucking rapes your ass or constantly sits in cover waiting for you to move then aimbots you.

We could have good AI. It's just not fun.

I'd say good AI will behave like a human while Hard AI will behave like a robot (which will result in you being raped)

...

sincerely kill yourself faggot

Absolutely witnessed, I hope he records it.

It was nice knowing you anons

...

Because good AI isn't sick graphics.

Sorta related, by Quake's AI are scary with their chat abilities.

Great AI can destroy players.
Most players want to escape reality and have fun for a while. They don't want to get destroyed.

Multiplayer gives competitive players a great experience without costly programming to create amazing bots.

It's a lose/lose for developers.

...

...

You're assuming what I wrote was an autobiography.

You're also a dumbass if you think the vast majority of people that play video games are hardcore gamers. That's why no one caters to people that want extreme challenges. Because there's a marginal market for it. Video games are a way to make money and the market responds to what will make money.

A marketing term meant to describe people who in total only make up about 200k-500k people across the entire United States. Half the guys on this board don't even fit that term the way most people think it does, and even less the way marketers actually use that term.

Then define the (((marketing team))) definition of "hardcore gamer" for us.

...

Someone who buys and plays every game, in full to full completion, and has a minimum of 2.5+ hours for games they don't finish. When I mean "every game" I mean every single game released throughout the year including steam, including AAA releases, and a mishmash of a common collection of games that is considered great although it varies by the region which collection to consider. That's less then 2% of the entire market, the technically correct term you're looking for is "challenge gamer".

Interesting trivia but the primary reason for linear forced cinematic experiences being pushed is not SJWs but Valve finally being able to gather important data like "how many people actually finish the game" for every game it's a grand whopping quarter or less, two thirds don't even make it to the halfway point before dropping the game they played outright. Executives and developers were not happy when they found that out.

it's not even that, making an ai that's impossible to win against isn't that hard you just let the ai cheat in various ways depending on the game.
making an ai that can actually play the game properly would be a massive accomplishment in modern games. some games have gone backwards in terms of the ai's ability to play the game (namely the civ franchise).
i'd be thrilled to have a single player fps/tps game where the ai wasn't pants on head retarded for pathing, just having the opponent always know your location and score headshots at a high rate isn't good ai.

Escape from Tarkovs AI is so shit about this. The bots are either pathetic or one tap you through walls. And their searching algorithms are nonexistent. They either know exactly where you are or forget you exist.

*
I'd like to add a little bit more before I hit the haystack "hardcore gamer" used to be used to describe basement dwellers before steam achievement results and the trend of datamining really hit off but the aforementioned results completely changed to consider total number of games, time spent playing, and how many consoles and the like is owned that is how completely game changing the datamining is. Assfaggots stick to their favorite game and only branch out to other flavors of the same game, then go right back to their favorite one after few months maybe more. The phone demographics are practically their own market but they're only included to inflate gender demographics; the only demographics that is smaller then "hardcore gamers" are a subset of people who do nothing but buy games and never play them at all. One dev user a long time ago who was a budding developer actually complained about it cause he listened and believed in no advertising being needed to sell his game but those 10-12 people who buy everything were his only customers.

This is probably the best answer. Good AI takes a dedicated, if small, team of programmers. That needs a real studio. But the studios have investors who want maximum return on investment, which means EA style kikery and multiplayer.

Except that isn't true

Let's see buyfags who would buy every game for the sake of it
That's three of the twelve accounted for. Who are the other nine autistic spergs?

This, unfortunetelly is the truth. As you can't give the A.I a specific purpose or completely agreed upon purpose without similar or the same ideas being put into it to use. Whether gameplay concepts or shilling.

Realism can be fun if the game is designed and balanced for it. More to the point AI isn't about realistic detection distances, it's about realistic response. An enemy that forgets you're there ten seconds after you shot his buddy is boring.

NEET oil barons who pay lip service to Islam and then buy shitloads of western goods.
Manchildren with Richfag parents, crossover with autistic jew.
The most normalfaggy normalfag to ever normalfag, like Patrick Bateman with aspergers instead of sociopathy.

Aimbotting isn't AI, and it can be fun to have to kill an NPC whose actually using decent tactics.

If that's happening the game is unbalanced. But I think by "AI" you actually mean magical knowledge, aimbots, and other shit that has nothing to do with intelligence.

LOL you can't be serious? Like, it is possible. But really now, no fucking way. I won't believe till I see it.

If a game is fun enough I can play it for thousands of hours without ever actually finishing. (since I play sandbox RPG's) Some games I've played many times and feel I got way more value than I expected, yet never finished.

So you actually have no idea what you're talking about then? Why would anyone want to discuss this subject with you when you're so ignorant about it?

Good AI can be done , and it's generalizable enough to work from one game to another in the same genre. That just hasn't happened yet.

this, (((they))) pic related just want it all for themselves. I can't remember an example so please forgive me but I know that you guys remember some older games that have much better AI than some newer games. I know I'm right, go ahead and prove it so. I know you remember.

Well I see why you are just shitposting then user

So apparently the fill tool IN basically every paint program (hell every paint program ITSELF for that matter) ever made is an impossible herculean task since it can figure out that at these x,y co-ordinates this RGB# or similar is present.

This right here is what basically everyone wants even if they're a turbo casual, something vaguely competent that doesn't wall hack, aimbot, spawn right behind you, have infinite resources turned on, rubber banding etc.
Good luck getting this ever if here it is "revolutionary" AI needs majority stock in a crutch factory to not get it's shit insta pushed in by even a newborn.

Look up the Royal Family of Qatar for reference. Or the Saudi royal family for that matter, but then again the house of Saud are pretty much autistic Jews.

Why would they care? It's not like they get less money if people drop the game 3 hours in (unless dlc or multiplayer micro transactions faggotry) if anything they'd be happy about that since it means they can sell other games that might not have been bought if the consumer was still playing the first game

I can imagine the day when real AI becomes a thing, MIT or some tech company like that will probably test it as a 3D avatar in their dev software
One day, NPCs will have a conscience
Maybe it's already the case and it's stuck somewhere on a supercomputer in a fake 3D world, knowing he can be deleted by the simple press of a button

You know that classic game room guy on youtube? Half of his shit was gifted to him by a oil baron from Qatar. This isn't even a joke.

Pawns are less of a technical accomplishment than the creatures from Black and White from ten years before. Not that creatures are any good but at least they can learn things like that if the player is looking at them they're in for a beating.

They consider it a wasted investment and think a game 75% shorter would be more cost efficient.

Look at thread more closely next time, machine learning is still just an AI trying to make different pathways to get to a predetermined flag. We have had this for years now and everyone making a fuss about it since Elon musk said its important really goes to show how many people here go on reddit. I said it once and I'l say it again, until the learning process ends up with the AI setting another flagpole to reach its goals we will never see sentient AI and seeing as how the best of the best bots right now can barely do the task they are given unless given a lot of fucking time and effort to find all possible variables.

Yeah its not like over 75% of a games funding goes to marketing or anything, a real wasted investment.

From a developers point of view what's the point if nobody is going to see it?

Because video game players are often ignorant of how intelligent the AI of a game is. They might be buggy, but most of modern AI have a lot of complex functions that didn't exist back in the old days. For example, gamers complain about how Fallout 4's AI is shit even though in reality they have the capability to decide when to take cover, flank, and throw a grenade to drive you away from your cover, and it's done more seamlessly than the event location triggers that existed in HL1 and FEAR.

...

I doesn't even need to have new AI from scratch….like in alien isolation you just need two different for role, one knows what to do and the other do what is told…
They are all the same type of AI but makes for different roles

Majesty had the best AI of any videogame, prove me wrong

Good or realistic AI is especially hard to make.
Even for a game that is as easy to understand as chess it took millions of dollars, supercomputers and many years of development before they had a competitive chess engine. And even then the best chess AI engine in the world ("Stockfish") which will quite easily beat all the top players, does not play like a human.

AI hasn't advanced because the people facing the AI have regressed. In order to evolve, you need environmental pressure to act on a population with selection. Modern game players aren't capable of exerting this pressure, nor do they exercise any form of natural selection, because they'll pre-order whatever garbage their corporate masters throw a hype budget and shiny graphics behind. If anything, AI is either stagnating or getting dumber because the people going up against them are stagnating and getting dumber.

because devs don't make good games anymore, they don't even have the skill to balance a game let alone design complex ai for it

Why do you need to spend resources on AI when you can have human controlled bots and humans even pay you for this? Online kills AI.

Pls. Open most played steam games list. 90% of hours played are played in competitive (aka you need to git gud to win) online games.

As other anons in this thread have probably already said, developing a good AI that is also fun is very difficult.
Developing the AI system with only its efficiency in mind will lead to the players getting the feeling that the system is unfair and that enemies are manipulated (spawned, moved etc.) to be behind them.
This is a result of the fact that it is hard to judge what is happening when the enemy is performing offensive action against you while unseen.
As for non-hostile AI, much of it relies on handcrafted schedules, which of course require man hours to make. Your average casual shitter will not miss an in depth AI character system, because they spend very little time playing video games, and rarely make an effort to really immerse themselves into the game.
I like F.E.A.R's approach. While it's not a very complex system, most shit the enemies do is telegraphed in some way, like them shouting to their buddies what theyr'e about to do. This gives the player the ability to prepare for what is about to come, without it feeling unfair.

Probably not true but.

They rewrite the AI every game so technique and experience is what creates a smarter AI. Too bad neither can be found in the gaming industry since nobody stays long enough to gain experience or think up new techniques.

Easily fixed by killcam. Praise COD for good ideas! So "positioning cheating" argument is just developers blabbering bullshit.

Really comes from the another angle. What is difference between good aim and aimbot? Well there is none. Robot to good at his job is not fun. People do miateks and are not perfect all the time. So they expect their opposition been on teh same level as they are. (level of noobs and scrubs who can't git gud). Only then game feels "equal" and therefore fun. And here is the problem, programming "mistakes" is more difficult then programming good job.

FUCK YOU

Making a good AI requires you to be good at your own game, which is something developers don't do anymore.

You online fags are all kinds of retarded.

This sums it up.

Not a significant selling point.

In online PvP game pressure comes form competition with human players. And level of human players has no ceiling of growth. AI is replaced by humans as better and cheaper (it is hard to beat negative costs) alternative.

all chess ai does is look through every possible move though, after around 8 turns have passed it can do that very quickly even on consumer computers.
it is nice having the ai look through movesets of old masters to choose its moves from in modern chess games though.

at least part of the problem with ai development definitely stems from the high rate of turn over. people with passion go in, get eaten up and tossed away. i'm not sure how much these devs are taught in their studies about ai either, i know ai (or machine learning) courses around here are all focused on data management which probably doesn't translate to teaching a computer vidya well.

It's not totally vidya but it's a begining

I was sort of more impressed by pawn pathfinding, maybe I'm easily impressed but I thought it was neat they could go basically wherever you can go and would eventually catch up if you got tearing off ahead.

For the same reason some people did this.
These are engraved on the silicon chips, 99,9% of the people who had those chips never saw this.

The jews in vidya have yet to achieve maximum jewery.

...

I used to have a Windows XP rig with a Montego II but it's gone now. It still hurts to see this tech get bought out and left to rot. I hope every single kike at Creative burns in hell.
IN HELL

Didn't an AI beat the shit out of everyone in the Dota 2 competitive scene? I think only one player was able to beat it once but afterwards the AI learnt how he played and he couldn't beat it anymore. It learnt the game by itself. Other than that, the guys who made it said they were planning to make an entire team made out of AIs and they would beat everyone with it.

you're retarded, they're infinitely more simple than even a fucking insect, even the most complex have like 12 neurons. The difference is they are highly specialized so they are better at performing a specific task, just like how human hand can do a million different tasks well but an ice pick is pretty much limited to picking ice and being used as a improvised weapon - both things a human hand can do (worse).
It's basically like a jack of all trades vs master of one.

...

Because intelligent AI makes video games sell less.

whats going on in this picture?

Kingdom Come Deliverance AI looked dumb in the combat video. Archers ran out of the fence (and the melee defenders didn't) encampment only to get slashed by swordsman. Very sad, I wish they would stay back and use cover, height, and wide angles.

Also the enemy infantry has no line formation, they break and charge into a blob of melee.

Will Volta's new gpu AI Cores be utilized to learn AI to have human-like team play in combat? Archers maintain distance, infantry attempt 2v1s, flanking, cohesive lines maintained during melee?

>will be utilized?
Oh user.

RTS and TBS games are too complicated to create a good AI. Look at how much effort it took to develop a good one for chess, and chess is much simpler compared to most video games.

In FPS and TPS games, it's pretty easy to make an AI better than the player, but then you would get aimbot headshotted as soon as you stepped out into the open. You would still always lose if they were as intelligent as the player (ignoring the fact that most players are imbeciles) because you're pretty much always outnumbered. And stealth games don't work at all if the AI is intelligent. There's a reason why people in real life don't go around being sneaky ninjas.

Ninjas were highly effective during their time, what's your point?

Strategy games are stuck between a rock and hard place because if the AI is too smart, the game would be unbeatable, but if the AI is too dumb, it will not be a challenge.

So they come up with a compromise, dumb AI that has some sort of cheat to make them challenging. Obviously, it still doesn't work all that well.

This is why game ai is shit, the players don't know anything about what they're talking about.

Your argument is held up by a shitty presupposition.
It's pretty self evident why this is false.
The real question is: what is good AI?
Obviously, it's one that emulates a variety of humans, who're conscious beings; each with their own intrinsic values, playstyles, and motivations.
However, for games at least, it's only necessary to emulate the properties of conscious beings that matter (for the context of a game environment, which has intrinsic limits). In the context of the particular AI purpose/genre, and hardware/algorithm capabilities.
Such as FEAR utilizing the aspect that conscious beings communicate (they had a neat approach, the agent that was triggered to speak, selected the response line of their ally, and was thus lending credence to the perception of the AI cohesively communicating).
There are many different ways in which conscious beings act, and we're made up of a plethora of behaviors, values, and other intrinsic/extrinsic characteristics.
Another, slightly more verbose example: the "need based" AI of the sims (utility of an action, with a "smell" centric perception system); which demonstrates the lower tiers of maslow's hierarchy (or any other "primal to abstract" values hierarchy).
It's just a matter of emulating the correct characteristics for the environment (much like life itself), and not projecting some ultimate idiolized AI that trivializes you as a conscious being (that has limitations, such as making an aimbot that reacts in a single frame unfair).