Why can't we ever have decent AI?


  • Please log in to reply
57 replies to this topic

#41 Gimli

Gimli

    Dances with Mutons

  • Retired Staff
  • PipPipPipPipPip
  • 1,036 posts
  • Gender:Male
  • Location:Croatia

Posted 22 May 2006 - 03:59 PM

Haven't noticed this posted before, so here goes:

While skimming through a PC magazine today, I noticed a little section of an article devoted to this. Could it be that finally something might change? Does anyone know anything about this? It doesn't surprise me that I haven't heard of this before, since all you here in the media is about the new shiny graphics and physics. Not that I mind those, but they get too much attention, compared to things that are actually useful for gameplay.

I apologize for the late post, but as you can see, it was unavoidable, and I think it was worth it. :P

#42 Crazy Gringo

Crazy Gringo

    War is hell and war never changes therefore hell never changes

  • Site Staff
  • PipPipPipPip
  • 512 posts
  • Gender:Male
  • Location:Albertslund, Denmark

Posted 22 May 2006 - 07:03 PM

I heard of it as an idea when I surfed around the web for info on the Ageia PhysX processor....Not as a specific product....
One more piece of hardware to stuff into a otherwise cramped computer I guess....
Why don't they just made a customizable PCI-card that the user can program to handle one or more of the following tasks: Sound, graphics, AI, Physics and what ever else you might think of.
They are talking about making the graphics processors exchangeable like CPU's today so why not that customizable PCI-card?
Words to live by from Sergent Gunnery Highway: ' You adapt, improvise and overcome!'

'To kill a dream it is not enough to kill the dreamer', -Quote by Crazy Gringo until proven otherwise.

Mental Father of Air Morelman and Air Wargot!

Official StrategyCore UFO: Afterlight previewer. Preview here!

Has joined the league of 'The Crazy 8800s'

#43 Gimli

Gimli

    Dances with Mutons

  • Retired Staff
  • PipPipPipPipPip
  • 1,036 posts
  • Gender:Male
  • Location:Croatia

Posted 22 May 2006 - 10:16 PM

Knowing a bit about the complexity of building these in general, I would say that for the moment this won't be anyone's focus, as it would require quite a bit of research, and I would assume they have other priorities right now.

As for changing different types of processors, that would seem even more difficult, if not too complicated for now. As I understand it, each type of processors (graphics, sound, general etc.) works in a different way, to a degree, so it may not be all that possible right now, physically. Of course, I may be wrong, you should ask someone who knows more about electronics. Although I do have electronics subjects I have to take in my faculty, they're not my interest, so I only force myself to learn enough to pass the exams.

#44 Bomb Bloke

Bomb Bloke

    The Smily Admin

  • Admin
  • PipPipPipPipPip
  • 4,625 posts
  • Gender:Male
  • Location:Tasmania (AU)

Posted 23 May 2006 - 03:01 AM

The problem with a dedicated chip is that you're bound by what it can do, whereas when you're working in software mode you can do what you like.

Video cards are a near identical example. Sure they make things go faster, but if you want to include some special effect, you need to be sure the video card knows how to do it. If it doesn't, then you either don't use that effect, or you go back to software mode.

Most people have video cards these days, but I doubt many people have even heard of the AI chip. So even if you did write a game that supported it you'd need to include your own software AI code for those people who didn't have it. That means more work for developers, so I can see the technology holders paying software designers to use this chip until it becomes popular...
BB's X-Com Projects Page - X-Com Games At GamersGate
You're just jealous 'cause the voices only talk to me :P
We love Tammy! :)

#45 Zeno

Zeno

    เอาเหอะ...

  • Members
  • PipPipPip
  • 58 posts

Posted 23 May 2006 - 11:16 AM

Interesting thread, this.  Advanced AI is a complex field in computing, and not really conceivable in computer games.  Think about it-- Chess is a simple game in that there are a limited number of possible moves permitted in any turn.  Yet it took, what, 2 super computers and decades of work to beat the top human player?  When you have a computer game with hundreds of possible moves, any sort of advanced computer behavior becomes impossible due to the exponential growth in cause/effect calcuations.

Humans have innately reasonable "guesses" and react intelligently based on experience.  Computers don't have these advantages; every decision is based on calculations.  Advanced AI uses case modeling and pattern-learning logic, but it is still a far cry from human capability.  

The best AI we might be able to hope for is something based on expert systems.  Expert systems rely on the accumulated knowledge of humans in a particular subject, put into a data warehouse with optimized search/retrieval algorithms.  This is still outside the scope of computer games; however, an automated database with cause/effect pattern analysis and a number of scripted responses based on human players could conceivably appear to be "better" AI.

I would love to see some advances in this field, but I think business deadlines and market competition make it unreasonable to expect from computer games.  This is one reason why multiplayer-capability is demanded by most game players.  Once a human figures out the game AI, the game becomes unchallenging.  Multiplayer capability allows a game to have much longer playability.

------

Something to consider:  Instead of advanced AI, games are often shipped with scripted behaviors.  Rather than demanding developers to spend months making good scripts, games could be shipped (as some are) with good-quality script editors.  This allows humans to make the scripts better, allows communities to grow around script-writing, and improves the game over time as the scripts are improved.

As game developers learn the advantages of open source code, or providing professional modding tools, I think we'll find improvements in not just AI, but many aspects of gaming.  

--Zeno
The highest-probability X-COM hazards are here:
Murphy's Laws of X-COM

#46 Alienated

Alienated

    Sergeant

  • Members
  • PipPipPip
  • 96 posts
  • Gender:Male
  • Location:Europe

Posted 13 April 2011 - 04:41 PM

The A.I. programming is incredibly hard. Nonetheless, there are certain problems that have no existing algorythmic solutions by principle.

Generally I am satisfied with the A.I. of games. My problem usually is that the game rules and the A.I. are often dissonant and this is the error of both the A.I. programmers and the game designers. I believe that the main job of a game designer is to find out game rules (game mechanics) that require less A.I. programming to provide a still challenging and entertaining game.

For example the combats in RPGs have almost zero A.I. programming. The player has a variety of equipment, skills, and power-ups against a variety of monsters. The game designers just find out a plot, attach a combat system and voila, we get a cool game. The RPG is a classic and evergreen genre, RPG fans almost never curse the A.I. because there is actually none. The best A.I. is the non-existent A.I. This is the golden rule No. 1.

Tactical games have situations where A.I. programming is needed, and here comes the dissonance. The A.I. programmer can't salvage a bad game design if it is better to walk up to the target and release an autoshot in touch than aiming and shooting from ten metres. It is a common error in turn based tactical games that walks take relatively less action points than shots. The average game designer's aspect is this: the player should not move the figures a tile per turn because the gaming time might increase by "tiresome" field movements. The problem could be solved by good game design but the average game designers lack the talent to do that because the average game designers don't like to play video games, they just do it for money. (They should play their crappy games for the rest of their lives as due punishment.) In response, the A.I. programmer creates enemies who do very weird things on the battlefield to beat the player at all cost. I often can't decide whether I should laugh or curse when I am fated to watch my A.I-stricken enemies act.

And this is another common error in A.I. programming: the enemy always wants to win. By this aspect, a potential winner enemy is equivalent to the challenge. The enemy never gives up, never makes peace, never flees. And the A.I. programmer does everything so the enemy could not just try to win but win. As I said A.I. programming is hard, the A.I. compensates this by impertinent cheating including the knowledge and calculation of the players' forces and location as if there was no secret in war. And this is a great let-down in elements of tactical features. There is no need to bar enemy scouting and to risk leaving weak points here and there. The A.I. always knows where and what to attack, as soon as you leave a weak point, the enemy will attack it. This is an easy test to see how much the A.I. cheats. In real war, the commander made weak points look like strong points and strong points look like weak points, or just made risks, every commander had weak points here and there and prayed the enemy not to discover them.

I believe the best working A.I. lets the player win if the player uses the proper tactics, I mean the A.I. programmer has to make the A.I. intentionally weak against certain tactics. But it seems the opposite happens. The A.I. programmers make the computer enemy invulnerable against banal tactics to increase challenge. And the result will be that the player wins by tactics that belong in the world of madness. Actually the player does not beat the enemy in the game but the A.I. programmer, he wins by a way the A.I. programmer did not expect.

Complex A.I. thinking-calculation creates holes in the system. Let me take an example from Warcraft II. The enemy usually has a gold mine nearby its castle. When it is mined empty, the peasants search for new gold mine because more cash means more troops. Then you have to do one thing: build a guard tower nearby a gold mine or deploy some archers in the path to the gold mine and kill the peasants. The enemy will spend all its money on peasants, and when no more peasants are coming, you can decimate the enemy because it is flat broke. You can do that to oil tankers as well. If the A.I. were dumb enough to do nothing but execute the peasants and create combat units as many as the cash allows and wait for the final battle, then the player might have a harder time to win. So actually the cleverer A.I. makes the game easier (after the player has found the hole in the system, which wasn't too hard for me for one). Actually towers are the most dangerous units in Warcraft II in spite of that they are just rooted in place and fire at occasional targets.

And multiplayer mode was recommended as a solution for bad A.I. programming. Come on! People play games because they are lonely. And an unexperienced human player is often worse than the A.I. enemy.

All in all, the A.I. issues should be solved but not by mere A.I. programming but better game design too.

For example the designers could encourage the players to actually help the enemy do trivial things that make the enemy more an opponent. Some things that seem trivial to a human can also be very difficult to program properly. For example the player should be able to tell the enemy not to do this and that stupid thing. Just an example: what if you could tell the alien "do pick up your rifle you dropped in panic and fight like a soldier" in X-COM games? It is very difficult to program the aliens to search for their dropped weapons or even scavenge freely on the battlefield. Safe scavenging is a piece of cake for a human player whereas the mere idea makes me tired in the aspect of the computer-alien side. Aliens unable to scavenge seem very dumb.

Quote

Still, you have to admit AI's evolved from the days of Wolfenstein 3D and Doom, where they just point them at the player and shuffle forward, oblivious to walls and obstacles.
This A.I. tactics can be quite effective regardless of its simplicity (or rather because of its simplicity). If you are out of ammo or low HP, the bravely rushing enemy defeats you. If the enemy would camp here and there, then you could always retreat safely for power-ups. The "Can I play daddy?" level is the easiest mode. If you raise difficulty, then you suffer increased damage and meet more opponents, resulting often low HP and poor ammunition and the dumb rushing enemy becomes frightening. If they rush when you are out of ammo and camp when you have ammo and aim at the door, then the A.I. surely cheats by knowing your power and position.

Quote

One thing I've seen in lots of FPS games through the years is their idea of "good AI". On easy level the accuracy of the enemy is very low. They hit nothing, and you can run circles around them. Then, as you increase difficulty, they get much better. In quite a few cases they go inhuman, and make headshots from distances that no human player could ever manage, NO MATTER HOW GOOD HE WAS.
Even Duke Nukem 3d simulated the imperfection of bullet trajectory. This means you have to be extremely lucky to make perfect headshots from fair distances. No human can do that of course. The almighty God can do that. Lucky Luke can do that. But only a cheating A.I. can do that in reality.

#47 Thorondor

Thorondor

    Thorondoropedia - Your source to everything Aftermath

  • Admin
  • PipPipPipPipPipPip
  • 48,909 posts
  • Gender:Male

Posted 09 April 2013 - 05:34 PM

*dusts off cobwebs*

Whatever Happened to Video Game AI? It's the title of the article at IGN.

::

All hail the rise of in-game Terminators... not. ;)

#48 Thorondor

Thorondor

    Thorondoropedia - Your source to everything Aftermath

  • Admin
  • PipPipPipPipPipPip
  • 48,909 posts
  • Gender:Male

Posted 13 February 2015 - 09:25 PM

*pops in again some two years later...*

Posted Image

RPS' Michael Cook writes about The Lost Future Of AI - part 1.

#49 NikiAlex

NikiAlex

    Squaddie

  • Members
  • Pip
  • 15 posts

Posted 01 March 2015 - 05:57 PM

My humble experience in RTS and TBS tells me this:
There are a quite a big number of conditions for AI to look for in order to be “good” (like comparison of battle stats, defense bonuses, current strength, detection and engagement ranges, weapon type, allies & enemies nearby, etc)
It is more of a challenge from design perspective to properly list all important conditions, behaviors’ and well balanced reaction thresholds. Once this is set up as schemas and functions then programming tackles begin to kick-in.
There more conditions and situation you push to implement the higher the change some condition will go wrong, will be missed at all and this of course will lead to annoying in-game situation. For sake of the discussion let’s assume conditions and responses were set up top notch and you have AI character/unit that appears “intelligent„ and deliver to you nice amount of challenge. Congratulations you have excellent AI for you Character class/Unit. But wait a minute… I have so many different character/unit classes, special traits and so many weapon types. Yeah, I will need to customize AI for most of them.
Now we have some awesome AI let’s do some combat in RTS. Ops AI need to calculate most of his condition every frame in order this to be really efficient. We have a CPU performance issues. OK then let’s reduce condition checking to every second, but not every frame. Well it appears this is not frequent enough. AI is not as good as we wanted.
At some point you are asking, how much time do I need to invest to got my good AI opponents
Just my two cents.
Will be happy to read more hands-on on this topic
Cheerz

#50 ñΩxicity

ñΩxicity

    Resident Åsshole

  • Members
  • PipPipPipPip
  • 616 posts
  • Gender:Male
  • Location:St. Louis, Missouri (US of A)

Posted 02 March 2015 - 06:33 AM

AI is never as good as we want it to be. It's either too moronic, or too abrasive. There's never that fine line in which we say..."yeah that's the kind of AI I prefer". I honestly believe it's trial and error, any AI can be dealt with once you figure out a way to manipulate it's faults and then take advantage of its said faults. If you created the most complex AI (built for use of gaming) and despite all of the brain maneuvering, calculated risks and manipulations you can't progress? It's no longer it's a game, it's better than you are, and you had best not introduce it to your wife, because it'll fuck her better than you ever would. So...there ya go.
"If you win it's just a game, but if you lose it's a complete waste of time"

-Al Bundy (Married With Children)

#51 Thorondor

Thorondor

    Thorondoropedia - Your source to everything Aftermath

  • Admin
  • PipPipPipPipPipPip
  • 48,909 posts
  • Gender:Male

Posted 02 March 2015 - 10:36 AM

Posted Image

::

Over at RPS, Mr.Michael Cook has now aired part 2 of his article.

#52 NikiAlex

NikiAlex

    Squaddie

  • Members
  • Pip
  • 15 posts

Posted 02 March 2015 - 01:09 PM

Fine article. These guys together are serious research capacity in the field.
Unfortunately such “AI mind clusters” are far beyond reach for the majority of game studios.

#53 Fraktal

Fraktal

    Squaddie

  • Members
  • Pip
  • 24 posts

Posted 01 June 2015 - 09:51 PM

While I'm no AI programmer, I am studying software engineering and want to make a turn-based game sometime in the future, just for fun. Of course, that means I'll eventually have to develop an AI for it as well (right now, even the core engine is less than 10% complete). I also almost never play multiplayer myself, so I have experience with the AI of lots of games (mostly strategic ones).

There were some AIs that show promise, despite not actually being all that good. One such example I've seen is the AI of Dawn of War:
  • It adapts its army composition to counter the player's. If its anti-infantry soldiers are getting decimated by my tanks, it starts to roll out anti-vehicle infantry and melee walkers.
  • It doesn't waste troops on minor probing attacks but keeps building up its army, attacking only once it outguns me or once I aggro it. If either of those conditions are met, its entire army attacks all at once in a huge wave, so no luring individual units away for piecemeal annihilation.
  • In team games versus a single human player (like most campaign missions), attacking one AI instantly aggroes its teammates as well. So if you build up your army and attack one AI, the other will counterattack your base while you're away. AI teammates also tend to gang up on the same player to ensure a kill overwhelming numerical superiority; I've played several human-vs-AI 3v3 matches before and saw multiple instances of all three AI players attacking the same human player simultaneously.
  • It concentrates fire on low-health targets but splits firepower to take advantage of hard counters. If a mixed army of anti-infantry and anti-tank guys are fighting an infantry blob who suddenly get armored support, the anti-tank guys immediately switch targets.
  • If it gets into a fight that has it outnumbered, the AI actively tries to kite the attackers back to its nearest armed structure to add that structure's firepower into the equation.
And the AI already pulls these behaviors off at the second difficulty level out of four (on Easy, it is practically brain-dead; Standard difficulty is where it gets unshackled and the fun starts). Now this is what I would call a pretty damn good AI... would. Unfortunately, in-depth observation of the AI's behavior reveals that it is actually rather rigidly scripted. To use the kiting one as example, the AI somehow instantly knows if you give an attack order onto one of its units and withdraws that unit. The withdrawing unit immediately stops fighting and runs to the structure above all else before turning around and coming back, so you can abuse this by giving an attack order onto an enemy unit then cancelling it; the unit will still disengage and make the round trip, putting it out of the fight for several seconds.

Another bad example of scripting in Dawn of War is units capable of jumping or teleportation. Unarmed units and units with a ranged attack more powerful than their melee attack AND set to ranged stance are designated by the AI as priority targets. If the AI has a jump- or teleport-capable unit and a priority target comes into range, the AI instantly jumps/teleports next to that unit. Considering that most jump/teleport units are melee and attacking infantry in melee forces them to stop shooting and melee back, this would be an excellent way to tie up shooter-type units, neutralizing their firepower. Unfortunately, the AI is hardcoded to use jump/teleport whenever the above condition is met, resulting in nonsensical behavior. Behavior like fleeing units unexpectedly teleporting right into the middle of the army they're trying to flee from, then resuming fleeing (and getting gunned down from behind). Or units with two jump charges using one charge to jump next to the priority target, then promptly using their second charge to take off and land ONE METER AWAY. It is painfully evident that whoever programmed the AI didn't take a lot of variables into account, like how far away the target is or what the jump/teleport unit itself is doing right now.



So then, I had an idea. I think that the best way to force the development of a good AI is to write one that isn't embedded into the game engine. I always felt that basing the AI's behavior around data the human player has no access to is lazy and cheap. But if the AI only has access to the same information the player is also legitimately able to obtain, it becomes much more challenging for the programmer to write its behavior, plus it would curb occurrences of superhuman AIs. Ideally, the AI programmer team should hire a tournament-level online player of the game's genre as consultant on how the AI should utilize the information it does have access to for maximum effect: when to build resource gatherers and how many of them, what build order it should use, etc..

Related to this, another one I was thinking of is to dispense with the "if X happens, do Y" kind of AI programming. Instead of events directly triggering scripted responses, the responses all get weighted probabilities, with certain events increasing the priority of some of them. When the AI needs to act, it uses this probability table to decide what's more important to do now and bases its actions around that; in other words, base the AI around emergent behavior instead of pre-programmed responses (Black and White already used this in 2001, I think; I read one example of the AI being taught not to do something it liked doing resulting in the AI only doing it when the player wasn't watching). This would make the AI's actions seem more fluid and natural, instead of being predictable. For example, if an AI is programmed to prioritize damaged enemies but somehow doesn't have access to the HP level of enemy units, it could actively note how much damage it saw the target take and use that to "guess" the target's current health (and if it didn't see the target being healed, the AI would obviously guess wrong - exactly like a human player).

#54 NikiAlex

NikiAlex

    Squaddie

  • Members
  • Pip
  • 15 posts

Posted 03 June 2015 - 06:41 PM

Amitakartok, your thoughts in the last paragraph about response based on weights and probabilities are essentially also pre-scripted behaviors and it’s already in use. Again we run into the issue that we could not define endless set of possible scenarios and responses without messing up things or hit some serious CPU bottlenecks.

What we call AI in games are just scripts for stats and range checks, related conditions and responses. No matter how many scenarios you manage to define it remains a predefined behavior that will look stupid in many cases and human will always find a way to exploit and crush it.

Little bit off-topic, but there is not true AI in games. Even it’s strange why we call it this way. AI would mean full blown algorithmic adaptive behavior, logical decision responses and ultimately the ability to learn and self-improve. We are far away from this in games (maybe it’s for good).
IMHO people with enough knowledge and skills to work on this fascinating topic are already assigned into robotic and military projects. When there is breakthrough there we will start seeing this in games.  Hope we don’t mess up that one… who knows what could happen.

#55 Sunflash

Sunflash

    SC's Resident Pony

  • Members
  • PipPipPipPip
  • 584 posts
  • Gender:Male
  • Location:Oklahoma, USA

Posted 04 June 2015 - 12:41 AM

Part of the problem with any AI in a "Pro" game is that any humans have access to the "metagame"; for example in SC2 you'll see various tactics develop and become popular by everyone to counter popular strat A, then someone comes up with a counter to THAT one and snowball.

I've seen several games won soley because the winner KNEW from watching previous gamed that thier foe liked to switch tech at a precise point or etc. This is sometging that you can practically never expect an AI to get close to replicating.

With that in mind, RTS AI will never ever be able to even try to compete with the big boys without cheating. FPs are possibly a different story, but that's hard to say.

#56 odin

odin

    Squaddie

  • Members
  • Pip
  • 1 posts

Posted 21 December 2015 - 08:21 AM

So if you break down AI behaviour in terms of layers of intelligence.

You'd usually start with some utility functions; generate a path from A->B, follow a path, track other entities in the game, LOS etc. Secondly you'd start tagging meta-data into the game; tag areas as under cover, or full of lava, tagging, tagging some entities as good (healthpacks) or bad (hostile turrets). Here you find your first trade-off; how dynamic is the world? It's easy for a designer to tag a rock as 'good cover', but if you can blow up the rock you now need to be able to re-calculate things like path-finding graphs and cover spots on the run. Personally I'd rather see highly dynamic worlds, but that's me. Mostly games are going the other way, baking in a LOT of information into largely static worlds; a lot of the middleware and engines are also becoming increasingly focused on really pretty, static worlds.

At that point you've got a pretty dumb AI. It can move from A->B, it can find cover, and fire at the player. To move it to the next level you need better domain knowledge, and you need to let the player KNOW that the AI is being 'smart'. Farcry's a good example; the enemies will chatter with each other 'he went over there', 'I see something', 'something's moved', ' cover me', etc. Even if the other AIs were all dumb as bricks the player would READ intelligence into the chatter; if you hear on AI shout 'cover me' while rushing you, and another AI is firing at the same time, you'll assume the AI is cooperating. Granny Weatherwax would call it Headology. In addition to that they had a bunch of little AI scripts that were situation dependent; see a grenade, what do you do (run away, dive on top of it, throw it back, etc). It's the primary way to make AIs appear smart, because it's something the player can see and identify as smart. One a similar vein is explicit set-pieces; the most famous one is probably the one in Halflife where the AI opens the tunnel door, toss in a satchel charge, and close the door on you. It's not actual AI (just clever cutscene work), but it makes you think the AI is planning these kinds of things.

At this point you probably have a pretty decent AI for an FPS, but you're going to be woefully short for a real strategy game, especially a turn-based game where the various cheats don't really work. There's a couple of ways to deal with things; the 'Chess' way is essentially brute-force. It generates a play-space of all possible moves, then all possible responses to that, projected forward several turns (that possibility space gets huge very quickly). For a strategy game, where the possible action set is even greater, it quickly gets out of hand. Usually you end up with a goal-based system. I want to have a huge army -> I need dragons -> I need upgrade my citadel -> I need more gold -> I should take that city to get gold. The goals are thus your 'domain knowledge'; the better you script them the better the game can plan. You need to balance KEEPING a goal (an AI that changes it's ultimate goal every turn is no smarter than one that has no ultimate goal) with reacting to player actions (still saving up to buy dragons while the player is burning your citadel is also dumb).

One poster mentioned various self-learning forms of AI; usually split along Genetic Algorithms, Neural Networks and matrix weights. The consistent issue with them is that computers are both capable of incredible acts of calculation, while also being incredibly dumb. You're going to need a new AI to keep track of your self-learning AI and stop it from wandering into Rainman land.

Genetic algorithms basically starts with some possible action configurations, run them through a game, score each configuration based on some 'fitness' criteria (how well did it do), then take a few of the best scoring configurations, modify them a little, and run them again (iterate this survival of the fittest until you get a great AI). Perhaps the best way to see Genetic Algorithms in action is this little Flash app using Box2Dto 'grow' cars. It's one of those ideas that programmers love, but which usually ends up not going anywhere useful. A couple of racing games have used genetic algorithms to tune AI car behaviour, for example... but usually you'll lock the evolution before shipping, to make sure your price-winning heifer doesn't get any two-headed offspring.

Neural networks are another one of the ideas we borrow from nature; essentially we try to model how neurons function by giving it a big set of matching valid inputs and outputs, and slowly accumulating input in the neural net. It's pretty good at some things traditional AI is bad at (like image recognition), but the more abstract the inputs and outputs get the more likely it will draw wrong conclusions from it. If you have a net that *kinda* works, but is just plain broken in 10% of the cases, trying to work out WHY those 10% of the cases are wrong is a nightmare.

Finally, matrix weighting basically takes our goals and steps, and weight them based on past successes. If every time I try to build up my dragons I get rushed before I'm done, let's change the probability of taking that goal, or change WHEN I try that goal. You want to have some level of control for how that's applied, and how things decay back into neutral, otherwise the AI's likely to draw the wrong conclusions, but at least this strategy works, and is open to normal debugging procedures. If the AI is acting stupid, you can crack it open and see the reason why it's always building dragons.. that's incredibly hard to do with something like a neural network, where the answer is 'the seemingly random sum of 100,000 floating point values made it seem like a good idea'.

TL;DR. AI is hard. The more complex the game, the harder it is. Most of the 'cool new technologies' that is supposed to revolutionise AI ends up creating a bastard love-child of Rainman and Frankenstein, but without the charm.

#57 Thorondor

Thorondor

    Thorondoropedia - Your source to everything Aftermath

  • Admin
  • PipPipPipPipPipPip
  • 48,909 posts
  • Gender:Male

Posted 13 October 2016 - 03:03 PM

Posted Image

::

The Guardian's Keith Stuart writes about the strange future of emotional AI.

#58 KateArcher

KateArcher

    Squaddie

  • Members
  • Pip
  • 1 posts
  • Gender:Female
  • Location:Kenner, LA

Posted 16 November 2016 - 09:53 AM

You mention FPS games and you are right of course. But let me please throw my two cents in under the name of Civilization VI. Have you read the reviews? Oh the glorifying articles with thrilled reviewers giving the game 8, 9 and 10 even. Maxed out scores - it is all peachy. My question is: did they even play this game? I agree that the game has a potential but the AI is simply devastating and absolutely ridiculous. Starting with again the cheat system the higher the difficulty goes (more everything for AI instead of making them smarter), through a ridiculous apostle, missionary spamming and ending with their erratic behavior. One turn AI declares friendship and two turns later yells at the player for having e.g. weak military. No logic whatsoever. Civ VI suffers from few more bugs but the AI issue is the biggest pleasure killer. I could go on but I think I made my point. Apologies for the rant but I thought this thread was good for my complaints.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users