Will AI Disrupt the Videogame Industry? - Slashdot

Posted by EditorDavid

• ›

  About as much (or little) as other forms of procedural generation disrupted the videogame industry.
  • 3 hidden comments

  • Re:

    Generating is easy. Testing and QA is the major time consumption often, and what gets cut short. Given AI state, even more QA will be needed than with human generated content.

    • Re:

      Well, depends. What is certainly correct is that it is a trade-off. The less effort you invest in design/creation, the more testing, QA and fixing effort you have. This is true for art, code, stories, engineering, etc. As Artificial Ignorance is of the currently hyped form is relatively often flat-out wrong and hallucinates things, I do not see this being any different when it is used. In fact, the QA and fixing effort may be _worse_ as it is harder to predict where it will put in serious flaws.

  • About as much (or little) as other forms of procedural generation disrupted the videogame industry.

    So... completely and utterly changing the landscape leading to better graphics, and in some cases whole classifications of a certain subgenre of video game like "rogue like"?

    • Re:

      Wow, our take on video games is very different. The vast majority of video games don't seem all that different then last generation of video games. The only thing that has changed is the improvement in graphics and that's just because technology is better now.

      I'd go as far as to say games are worse today then a decade ago. Almost all of them are rehashes of previous games and often times they now won't even let you just outright buy the game. It's "free" with a bunch of timesinks put in and if you want to s

• ... resulting in so many games that its damn near impossible to get people to play yours.

  • Re:

    Yep. Every young game developer dreams of coming up with some simple gameplay idea that allows them to make a game quickly without too much effort, and then watch the game go viral and make them into multi-millionaires once it goes live.

    Well, that dream is not realistic. The sooner would-be indie developers drop it, the better.

    It does happen that low-budget games sometimes do well. But the odds are overwhelmingly against it and getting stronger against it every day. If you want people to play your game,

    • Re:

      Flappy bird.
      Tetris.

      • 1 hidden comment

      • Re:

        2 games out of millions.

        • Re:

          2 hugely popular games out of how many indie games? Both made by single people hacking away in their spare time.

          And how many commercial games are big winners out of how many made? Pretty much all of which are made with expensive licensed engines and hundreds or even thousands of people with giant marketing budgets and take years to produce.

          • Re:

            I am not sure what your point is. In my original post I stated that it does happen. You gave two examples of exactly what I stated.

            I also stated that the odds are overwhelmingly against it happening. Your two examples don't dispute that fact.

            • Re:

              I already answered this.

              How many India games make it?

              How many lose money?

              How many AAA games make it?

              How many lose money?

              What percentage of each makes it?

              What is the definition of making it?

              Until you can answer those questions it remains unclear if being an indie developer is a hopeless endeavor commercially.

              • Re:

                You seem to think that your final statement should have been obvious from the prior statements you made. It was not.

                But now that it is clear, I wonder why you didn't just web search for the answer. Sites like this [vginsights.com] can provide some insight. Direct quote from the main page: "Over 50% of indie games on steam have never made more than $4000."

                There's more information there if you want to poke around. The general gist, however, is that only a very small percentage of games make millions, and many of those had

              • Re:

                Google can readily tell you all those things. Now run along.

          • Re:

            LOL! You said something incredibly stupid again. Take your lumps and move on.

            Or are you going to do your usual thing and throw a temper tantrum because someone called you out on your stupidity?

    • Re: Making games is already easy...

      Another issue that could exacerbate the issue is a wave of AI-generated shovelware clogging up distribution platforms. We already had that issue with Unity where hacks were basically slapping together store assets and pushing them onto Steam, flooding even the front page with garbage.
      Digital platforms will likely need to update their curation methods again to deal with the spam.

      • Re:

        They have needed to do this for a while. the algorithms are not very good at picking up good games over the shovel ware that uses ad networks and seo tricks to keep bouncing people between a small network of games that all advertise each other with fake advertising filled with lies..

  • Re:

    Yeah. what if the ease of spinning off great-looking novel content actually increases the importance and premium paid for licensing IP that is relatable to the real physical world, such as the EA Sports franchises, or Forza / Gran Turismo brand licensing? Thus centralizing / monopolizing the industry even further?

  • Re:

    It's almost like we've been here before....

    Last time, the bubble burst and you know the rest. The industry appears to be undead at this point, shambling forth, shedding mediocre games with every greedy step, unable to die.

• This is what the current state of ChatGPT can do to create a simple game [youtube.com]. It won't be all that long before it can create more complicated layouts with deeper story lines.

  Imagine all those out of work gaming programmers screaming, "It took mer jrb!".

  • Re:

    User: "ChatGPT, create a 3D model of a human skeleton I can use in my game."
    ChatGPT: explodes.

    Game programmers are safe. The only ones who have to worry are the entry-level coders who only write simple code but do nothing else. There is so much verbosity and back and forth needed to get ChatGPT to create simple, common code structures that you can probably code it yourself just as fast.

    The language interface is the only impressive thing about ChatGPT. It's only just barely passing at just about everything e

    • Re:

      I really wonder why a part of the population dreams of seeing coders and artists lose their jobs. This is the real question. It seems to be a political thing, as Trump and Musk supporters and other far right people often exhibit this behavior.

    • Re: As posted elsewhere

      AI tools like adobe firefly can already create vector art - creating a 3D model out of a verbal prompt is probably only a matter of weeks. It doesn't matter, if it's an "illusion" as long as the illusion is good enough. I'll bet you a case of beer, that when AGI finally arrives some day, it will be indistinguishableform LLMs and the likes (at meast until it evolves). Granted, an AGI is a lifeform so it has to be treated differently - but still, it will walk and talk the same

    • Re: As posted elsewhere

      ChatGPT or any other bot will be getting better with time

  • Re:

    In this video, chatgpt created nothing. An human used natural language as a programming language to code a flappy bird clone. Right now it is faster and easier to use some no code game engines than chatgpt to achieve the same result. It is already possible to produce games/website/whatever with almost nocode but coder still exist, why?
• Games are a great application for AI because mistakes are not a serious issue. AI generated plots, scenery, characters and dialog will make it much easier to generate content. It also makes it possible for gamers to have the AI design custom worlds with esthetics tweaked to the gamer's liking.

  • Re:

    ... you mean... boobworld can be real?

    • Re:

      If that's what you're into, Japanese developers have been making games where player can effectively customize a woman of his/her preference that will then interact with the player in real time 3D environment. For at least a decade. Probably longer.

• Eventually. Humans are too curious to leave it alone. I just wonder how long before the dark side of AI will show its face.

  I could see AI doing well in random content generation needed for visuals and physical character generation.

  Let the dice roll.

  • Re:

    These things have no capacity for novelty. I can't see any reason to be concerned at all.

• Designing graphics,replacing voice-actors,making cutscenes...everything will become easier with AI

  • Re:

    All of that is very much up in the air. The LLM's cannot create original anything, but can only copy what they've been trained on. The copyright liabilities are astounding, and the little guy may or may not ever have a means to make use of LLM's outputs. This will all have to be slugged out in court over the next 20-40* years before you or I will have the legal certainty about what we can or cannot do with it. This will remain the domain of deep-pocketed interests until then.

    ---------
    * SCO's lawsuit against

• by alvinrod ( 889928 ) on Sunday April 09, 2023 @12:59PM (#63436794)

  I think one of the biggest advancements will be using AI (I use the term loosely here) to generate voices for game characters. This will make fully voiced games possible without massively ballooning the budget and file size to produce and store all of the audio.
  
  It might also help out with testing to a degree as well. Some games are just too large for developers to test every pet of the game world. AI agents that can be trained to act closely enough to playing the game and recognize when they've entered some improper state would help with identifying bugs.
  
  Maybe it can be used to help artists as well, but I don't think that's where the real uses will come from. I'd be more interested if it can help with creating more realistic model animations just by studying real-world objects. It's not too much work for an artist to create a good set of textures for a wolf, but getting something that has to realistically move like one is considerably more work.

  • It makes me wonder if movie studios

    will start having actors sign away voice rights early on before they're starts, or if the actor's Union will stop that.

  • Re:

    Voice actors get paid peanuts compared to an actual game development budget. This would have zero effect on the industry.

• I'm primarily a software developer, but art school trained and I also went to culinary school... so I've got a lot of experience in art and making creative products in addition to making software. Many fellow technically-minded people think art is mostly about having the skill to create images and use art tools, but that's a pretty small part of it. Maybe 20% of art school degree classes are directly about developing a technical art skill. Many of the others may use those tools, but are fundamentally about theory, developing an artistic perspective, developing an eye for seeing things conceptually and how to play with those concepts, communicating emotionally using visual subtext, and things like that.
  
  For example, I can always tell when someone who's skilled in Photoshop without being skilled in art makes a composite image. It might be technically perfect, but in the conceptual and aesthetic sense is woefully boring or unbalanced. Another very visible example is in photography forums online... there's always the crowd that's way more about collecting the kit and talking about taking pictures than the art. Their pictures are technically perfect, but if it's not going to communicate anything beyond what physical objects were in front of the camera when the shutter opened. That's great that it makes them happy and I wish them well, but don't confuse their taking pictures with the images an artistically trained photographer would create.
  
  Of course, AI will hugely disrupt games--- especially comparatively lower-effort high-turnover products like mobile games-- but no matter how many iterations of some set of character models AI can generate, you still need someone with that fundamental knowledge who knows whether or not any of it is good. And for fine art-- AI will never create anything that isn't some amalgam of the things it already knows. There will always be a creative avante garde who will be making the real change that will feed the algorithms, and there will always be people who are way more interested in seeing their work in its pure form before it gets dumped in the computer mixing bowl.
  
  So yeah, absolutely, but I doubt we're going to hit "OK Google, make me a AAA quality FPS where the main character is...". I think people like workaday utility web developers, jr contract lawyers, lower-end graphic design who work in sign shops and the like are all way more directly in the path of the AI axe.
  • 1 hidden comment

  • Re:

    Fantastic comment!

    This quote from a professional art teacher blew my mind as a rookie painter:

    "What does it take to make a museum-quality painting? You might think it's just a question of developing the technical skill... but actually MANY people have the technical skill. Very few of them ever succeed. Even with the skill it's not so easy."

    • Re:

      Yeah. I saw someone on a tech subreddit say Jan van Eyck wasn't that talented because he used a projector to paint all those captivating little shiny bits.
      
      Who gives a shit.
      
      Jan van Eyck's paintings definitely weren't notable because they were realistic-- they were notable because they were absolutely captivating. If you had compiled every piece of artwork from every painter before and fed it into a ML system, there's no prompt that could have possibly created what he did with his paintbrush. A more m

      • Re:

        I think it's worse than the Dunning - Kruger effect. Slashdot and tech forums are chock full of people who are Badly Broken. Not only do they lack a soul, they don't understand what they're missing. They have massive inferiority complexes and say delusional, mentally ill shit. Dunning Kruger is just overestimating one's ability, Slashdot and techies have a kind of megalomania or schizoid type personality disorder where their pronouncements bear no relationship to reality!

• What AI provides to gaming and media goes far beyond just procedurally generating a landscape. What AI can do is much closer to the holodeck, where the user specifies what they want to play and the AI generates everything, the story, characters, rules, landscape and all the rest. There is no longer a need for a game creator. For some genres this essentially already works. ChatGPT can generate stories, can generate puzzles and puzzle dependency graphs, it can generate images, there is AI to segment those images, generate depth maps, etc. That's basically all the bits and pieces you need to create a classic point&click adventure from scratch, completely by AI (this is what StableDiffusion can produce in 20min without any user tweaking [youtube.com]). The thing missing is having all those AI systems integrate with each other, but people are already working on that too. If you leave away the graphics, you can already play simple text adventures right now completely in ChatGPT.

  The thing I am most curious about is how this will change media consumption in general. A large part of classical media is the shared social experience, people don't watch movies just for the movie itself, but because friends and family are going to watch the same movie. With AI they now have the tools to generate custom content at home, tweaked for them personally. Furthermore that content will be dynamic. With AI you can explore locations and characters that aren't part of the main plot. You can change the story on the fly. So it's not just that AI is making the movie/game/etc., AI is the thing you use to consume the media with. There no longer a need for a finished.mp4 file falling out the other end, it will all be made up on the fly by AI.

  • Re:

    >A large part of classical media is the shared social experience, people don't watch movies just for the movie itself, but because friends and family are going to watch the same movie.

    Streaming services largely killed this aspect. "Netflix and chill" is not a euphemism for sitting down to watch a movie. We're already in the world where streaming services are highly targeted at individuals, and in the world where living together with a spouse is rapidly degenerating from a norm into a rarity. Which is in

• The problem with AI for game creation is that it's basically a new form of procedural generation. Eventually, everything looks alike or becomes a similar/familiar experience and the generation of things is just a boring meat grinder.
  
  Just like AI artwork, you need the human to make real art, even in games.
  • 1 hidden comment

  • Re:

    Not sure whether this is true - since introducing more "random" aspects into AI generators seems easy enough.
    
    But looking at the output of, for example, a contemporary AI porn image generator [pornpen.ai], you seem to have a point that current AI generators produce way too repetitive results.

  • Re:

    -glances toward Madden game series-

    • Re:

      Say what you want about Madden, but if you buy one every several years, it's good fun. I played the crap out of Madden 18 and then when Madden 23 dropped, bought it and have now played the crap out of it.

      I imagine if Xbox One get's Madden 28 (doubtful) I'll buy that for the updated roster as well. The difference between 18 and 23 aren't much but 23 is actually more fun to play. This isn't true with every iteration though. Madden 19/20 wasn't really worth upgrading to.

      Some games, you expect to just be graphi

  • • Re:

      Nah. Raw AI product is badly overfitted and subject to the same awfulness factor as ChatGPT (which may literally make up false things, complete with false citations).

      This is fine if a human with some skill refines the raw product... but then the quality depends on human skill. We're talking about a purely procedurally generated world that doesn't break immersion. That simply isn't going to happen.

  • Re:

    Maybe mobile or throwaway games, but games with more engagement definitely won't. Game studios know what they have to do to keep their customers and gamers won't be any more forgiving for derivative boring content than they are now. As it stands, they could buy beautiful assets for dirt cheap-- far cheaper and more game ready than you'd get with AI-- from assets stores, but they don't. Why? Because they know that creativity is the core of their business and they'd be undermining that instantly by doing so.

    • Re:

      Also, cheapo game studios have been using AI and asset store assets for this stuff for some time.

• You might see more serious 3D FP/3P style games from indie developers. OTOH I suspect that they're going to look very similar, and an advantage the big studios will have is not looking similar to everything else.

  So I don't see it as disrupting anything, more managing the constant improvements in processing we've been getting over the decades.

  One thing I'm interested to know is whether it might result in improved single player experiences in open world games if the NPC characters you converse with were able to interact with you more fully. That might be the biggest change to games going forward. While the major studios have been focusing on multiplayer, maybe independent devs will be able to give us immersive worlds where the majority of characters you meet are fleshed out people that fit within the story's universe, not 13 year olds who think "LOL" is an actual word, and is appropriate in every sentence, funny or not.

  • Re:

    Most games already look very similar, because a lot of games use same engines AND same default assets from those engines for much of the environment. Same applies to sounds.

    If you have a good eye for detail, you'll quickly notice same default UE4 rock textures and such in many different games for example.

• Quality will suck, elements will be repetitive and sometimes simply be off. Writing a custom world-gen, for example, has a much lower risk of angering your customers. Either be genuinely creative or not. The second option will not make you a lot of friends though.

  • Re:

    So exactly just like every game today, where same default UE4 rock textures are used in countless games, where same dungeon layout with slight recoloring is a norm, where when you level up, enemies have the same model but different color and maybe some small additional element on top of that.

    We already have a problem with repetition and things being off, because most of the design work is mind numbingly repetitive. So people copy paste the same levels, same models, same textures, same sounds etc where ever

    • Re:

      Well, yes. But Artificial Ignorance will att real problems in surprising places on top of that.

      • Re:

        ML AI will remove the majority of work that was mindless drudgery of picking each rock, each pebble, each patch of grass, placing it, then placing a correct texture on it, then figuring out the scene lighting so it doesn't look awful and so on. If you think there was some kind of unique insight in doing that, you should really try to be less bigoted against game designers. They're overwhelmingly not 90 IQ crowd who would enjoys this sort of work, but 110 IQ crowd who find this sort of thing soul destroying.

• All non-trivial software is stuffed with tons of laboriously-designed content. That's why we have "data warehouses" and "data lakes" are all about. These collections of data tend to be poorly organized and haphazard. AI might be able to help tame the chaos.

  Business software these days is sprawling and inconsistent, with changes to business rules being applied in some places before others, and some are never updated, because there is simply too much work to do to make it happen.

  Dependency analysis is an area

• • Re:

    More likely people in that category will be far too busy jacking off to their latest waifu than caring about something as trivial as going outside and planning an actual murder.

• Rather than procedural generation for the environment or art, it would be cool to use ChatGPT like AI for NPC dialog in adventure games that would seem a lot more rich and diverse than scripted dialogue. They could give the NPC a "personality/speech style" and some information about the world and it could put together responses to player questions in new ways each time. It would also be able to respond to questions or statements from the player that it hadn't been specifically scripted for in a more natural
  • 1 hidden comment

• It'll be all fun and games (literally) until someone tricks the AI in plastering dicks all over the environment...

  Then it'll be an outrage (again)...

• FFS, the game industry in many ways is / has been the founding area for AI.

  I'm not sure I really understand where this stupid question came from, but AI and video games are just made for each other... and have been for decades.
  Before multiplayer online really took off, we battled "AI" opponents.
  Even whilst multiplayer games were starting to become hugely popular, we still had "bots" - Quake, Quake III, Unreal - whatever, you name it.... but I digress, we need to unstitch the loaded term AI.

  And I can't be b

  • Re:

    I don't disagree with anything but your final statement.

    There is a level of abstraction to consider - no, nobody's made a really intelligent computer (yet), but the recent explosion in AI effectiveness is a new and notable milestone.

    Reducing it to "we've always had this" does not adequately describe the effect we see in the real world. The tools are (fairly suddenly) much, much better than they used to be.

• Even if you could put an extra dimension on stuff like stable diffusion, you wouldn't want to. The amount of data necessary to train just isn't there, and the computation for high resolution assets would take too long. Meanwhile DreamFusion sucks. The algorithms for high quality 3D content generation will likely have be far more physical oriented, tweaking parameterized building blocks and composing them for scenes. More Metahuman, less DreamFusion.

  The problem is far more different from language models and

• I'm editing my friend's book, which is a How-To guide for using AI-Art program Midjourney. It was born out of the frustration we both had over using existing internet tutorials to create anything meaningful. My friend has an extensive background in visual design and it still took him six months of research into a wide range of art, photography, and architecture related subjects to finish the book.

  One of the more surprising findings was the amount of human effort required to go from the AI's output to an image of sufficient quality to include in the book. At a minimum, one requires an art program and photoshop. Even then, the images produced don't really compare to what a professional digital artist can produce... my friend can iterate over 100 images and spend 20 hours post-processing and it will look maybe 80% as good, with that missing 20% being the emotional resonance the pro can produce.

  AI-Art empowers the average person with little to no skill at traditional painting/drawing/digital art. Rather than spend thousands of hours learning the basic skills to draw, they can make their vision a reality in a few dozen hours. I call this the 'easy win' model - by getting nice images right away, that unskilled person will (hopefully) persevere... that's the point of the book.

  However, making an open-world game with procedurally generated AI-Art that can compare to a professionally designed AAA game??? That's just idiotic. The human eye is very good at picking up the 0.01% discrepancy that breaks immersion.

  All AI will do is let amateurs make far better games than they could otherwise. This will be incredibly valuable to encourage people to go into game design, but it won't 'revolutionize' anything about what people actually consume en masse. It will speed up Professional developers by about 10% -- the early stage proof of concept demo.

  • Re:

    Most so-called AAA games just use and re-use high-quality assets and textures to impress you. Once you create a virtual wooden barrel, it's dumb to recreate it for every new game. with the use of AI, you could generate most of the mundane models in the poly count you want with little to no effort. The AAA part would be gameplay and feel, not graphics

• If GAI is anything near the hype and promise, ALL industries are going to be disrupted at very fundamental levels.
  This represents a historical shift in the ways we have employed technologies in the past, It affects individuals, companies, governments, and humanity as a social whole.
  The disruptions we see today are just the 1st scratches on the surface.
  The close relationship of gaming and computers guarantees that it's the low hanging fruit in the race to optimize everything.

• I've been working in the games industry for 19 years. While my company is exploring AI applications there are a ton of legal and ethical considerations around their use. With that said, there clearly are some benefits to be had from AI in various applications but at the end of the day you still need flesh and blood people to be making sure that everything fits cohesively within the framework of your game.

  • Re:

    I couldn't agree more. For fun I took Will Wright's masterclass in game design. There's an element of understanding how people think that's KEY to making a fun game. Ideas like feedback, freedom to fail, happy accidents, leaving space for the player's imagination.

    An AI will never understand what humans find fun, let alone what they find beautiful. 90% of games already feel too generic.

• There wouldn't be such an interest in AI, and we wouldn't have video chips to crunch the code on--if not for video games. I am sorry for my part in this.

• I've often dreamed of just being able to tell my computer more or less what I wanted and get the product out. I've played a lot of Neverwinter Nights and Neverwinter Nights 2 and the community content is what gives it such longevity. Both come with toolsets that let you effectively build any adventure you want. The biggest limiting factor is it takes a lot of time and you have to really enjoy designing stuff.

  If I could just draw out some maps, upload them to the computer and tell an AI to create me a module