- cross-posted to:
- Aii@programming.dev
- cross-posted to:
- Aii@programming.dev
Yesterday, Google announced Project Genie, a new generative AI tool that can apparently create entire games from just prompts. It leverages the Genie 3 and Gemini models to generate a 60-second interactive world rather than a fully playable one. Despite this, many investors were scared out of their wits, imagining this as the future of game development, resulting in a massive stock sell-off that has sent the share prices of various video game companies plummeting.
The firms affected by this include Rockstar owner Take-Two Interactive, developer/distributors like CD Projekt Red and Nintendo, along with even Roblox — that one actually makes sense. Most of the games you find on the platform, including the infamous “Steal a Brainrot,” are not too far from AI slop, so it’s poetic that the product of a neural network is what hurt its stock.
Unity’s share price fell the most at 20%, since it’s a popular game engine. Generally speaking, that’s how most games operate: they use a software framework, such as Unity or Unreal Engine, which provides basic functionality like physics, rendering, input, and sound. Studios then build their vision on top of these, and some developers even have their own custom in-house solutions, such as Rockstar’s RAGE or Guerrilla’s Decima.
You must log in or register to comment.
Welcome to the stock market: where the money’s made up and the rules don’t matter!
I see a couple of major practical reasons why game (engine) devs are under no threat from this even if it gets better in the future:
-
Scale. Like all things AI, this is not going to scale well. This doesn’t generate code, 3D models and textures, both making games and playing them requires running the model. So if you want a game to have a persistent environment where the world behind you doesn’t get regenerated into something different after taking 8 steps, the context window is going to get real large real fast. And unlike programmed games, you can’t make choices about what’s worth remembering and what isn’t, what can be kept on persistent storage and is only loaded when it becomes relevant etc., because it’s all one big, opaque blob of context, generated by a black box; you either have it remember everything or it becomes amnesiac in a way that makes it useless. Memory availability also isn’t increasing at a rate where this becomes a non-issue any time soon.
-
Control. Manipulating the world though a text prompt gives a lot of control, but it’s also very course. It’s easy enough to tell it that you want a character that can run and jump, but how fast does it run? Does it accelerate and decelerate or start and stop instantly? Does it jump in a fixed arc or based on the running speed and duration of the jump button being pressed? How far and how high? You’re going to run in the limits both of what you can convey and what the language model will understand pretty quickly. And even when you can get it to do exactly what you want, it would have been faster and more practical to manipulate values directly or use a gizmo place things. But there’s no way to extract and manipulate those values, because again: big, opaque blob of context.
-
So people are idiots. Got it.
I dunno man working on a video game as a side hobby they’re the worst things I’d use for gen ai. There’s too many things from pathing to physics and collision that require human input to make work.
Anytime I’ve tried it’s given some absolute shit results.
We play games because they’re stories and challenges put forth by other humans that look interesting.
Even if a slop machine put together a cohesive game involving metaphor and emotions, it’s still not human and it still won’t be played and enjoyed. It would ring hollow, just like AI sound files that try to approximate human music.
Hmmm. let’s build something we know and take some time, and know how to modify and build off it, or! we could spitball into the shit machine and it can shit us out a shit we don’t know how any of the shit in it works and then spend twelve times as long untangling its web. SHIT MACHINE AHOY!
Genie is pretty cool as it stands from a technical standpoint, but… 1 minute of some really, really bottom tier walking simulator gameplay is not going to destroy the gaming market.
Investors are so easily manipulated.
It’s an initial proof-of-concept. It’ll be developed into more complex games eventually, that’s not really an issue for it
The main issue is that it’s just a facade. It completely lacks the foundation required for a game. It’s a world without hard rules, which is a terrible experience for any user. The game isn’t determining cause and effect from actions. It’s just guessing at what would come next
What’s the point of decorating an in-game house if the next time you go there, the AI forgot what was supposed to be there?
What’s the point of completing quests if the AI forgets what you’ve completed?
What’s the point of getting new gear if AI hallucinates what gear you have?
There is no progress in an AI generated game because everything is made up as it goes. Google would need to fundamentally change their approach to allow for that
That being said, this and similar technologies could be used to add really cool mechanics in a game that previously weren’t feasible. The ability to add a temporary dream-like or watercolor aesthetic easily would be great for small developers to tell stories
It’s an initial proof-of-concept. It’ll be developed into more complex games eventually, that’s not really an issue for it
Except it is an issue, just one being masked by the mountains of cash these companies are burning to provide AI. To increase the depth and complexity and actually store state would require orders of magnitude more energy, compute, memory and storage. The AI bubble is causing very single one of those to become more expensive. At some point the market will call bullshit on these companies (“show us profit, or at least exponential revenue growth, or line go down”), at which point these companies will attempt to download the costs onto their users. When people see the bill and realize what these services actually cost, the whole thing is gonna collapse like a flan in a cupboard.
Yeah, I think a lot of people forget that Google (and AI research like this) pumps out a lot of work like this that shows amazing new advances. However, that doesn’t mean any of this is near ready.
Here is a 2018 paper about using world models (a concept where a model is developing an understanding of a “world”) that used it to create an interactive Doom AI model - https://arxiv.org/abs/1803.10122 just to show that this stuff has been in the works for a long time.
If this is widely adopted, I have enough emulators and classic PC games to never buy another game in my life and still be entertained the whole time. Good luck, corpo dipshits.
I’ll still buy from principled indie devs any day any year. There may be more old games than anyone could play in a single lifetime but let’s be real most aren’t good.
This will never be widely accepted in the gaming space because it’s not a game. The model only generates an interactive world, not a game world. It’s effectively a glorified AI prompted showroom. It’s useless as a development tool because nothing it generates is usable in the traditional development process which means the model would have to create the whole game but the model is incapable of understanding what a game is.
Yeah this is more investors being stupid. Hell this would be impact VFX and Architects but the logic they are using. The whole thing is a cool demo but little real world application like like with most genAI.
So… it’s as good as Starfield then but without load screens?
Even better, it’s Starfield but your character is moving in 4D space and things pop in and out of existence depending on your position in the 4D space. And of course no loading screens.
That’s all it does so far.
But I doubt AI games will succeed, people are always going to want the human touch when it comes to art.
There will be a demand but I wouldn’t bet on that demand being the most popular.
That’s all it does so far.
Isn’t that the AI hype in a nutshell? “It’s all it does right now but if you add insert hopes and dreams it’s going to revolutionize X”.
I mean, human touch will play a role but I think the tech overall just nowhere near where it should be to make games. It would actually need to understand what it is doing because there needs to be some intentionality there. Something as simple as a counter going up when you kill an enemy, but I think even that goes beyond what current models are even remotely capable. They would be capable of imitating a counter for some timeframe but to actually keep track of it over a long gaming session? I have my doubts.
They would be capable of imitating a counter for some timeframe but to actually keep track of it over a long gaming session?
The article was little light on the details, but if the whole game is run on ai thats what is going to happen. But if AI is creating real code and the game it creates has real files that are saved on the computer, things like point counters are not anymore tied by the limits of AI’s memory.
But i just dont see how AI in its current state could make large cohesive projects.
Also there is no such thing as artificial intellect. AI is just nice marketing word for something that tries to mimic what real AI would be.
It’s not generating any code. You don’t even get a game out of the model, you only get a video of what you played. It’s like an AI video generator except you have control over the camera and character.
So its a glorified a procedural generator that does not save anything it makes?
What the fuck. Its like saying game devs are being replaced because people see dreams when they sleep.
Given that what it “does so far” already required the theft of the sum total of human creativity available online and the sacrifice of the survivability of humanity due to climate change, kinda seems like there isn’t much else to wring out of this.
Its a step in that direction though
There were also steps in the NFT games direction. Steps in some direction doesn’t mean those steps will lead to somewhere.
Very different things.
Because one is completely useless and the other is great at making the illusion of not being completely useless?
As a dedicated fan of walking simulators I can already see the amount of shovelware we need to dig through to find the good stuff multiplying by orders of magnitude.
It’s been a year since I played INFRA and I’ve thought about it without fail at least once a week and it damn well isn’t because they haphazardly made boring environments.
Well that’s something I didn’t think about before. How would you even release an AI game? It’s just a prompt and the rest is a black box.
The companies that market machine learning tools to investors and the masses have not been set up by people who believe art has value. Everything is content, and content exists to be aggregated alongside advertisements or displayed for a fee.
I genuinely hate that actual artists can’t use a lot of pretty neat novel digital levers to make stuff. Because it’s synonymous with garbage. The ability to leap across the uncanny valley has lost all novelty and is downright banal now.
But the answer to your question is the same as every desperate attempt at getting a “good” use case for slop generators. It’s for cranking out low effort trash.
This. My back log of physical DS and 3ds games is extensive and grows a little every time I remember I have the eBay app on my phone. Sorry wallet.
If can grow more if you know the way of Luma CFW and Twilight Menu ++
…, much, much more
3ds and DS cartridges both have a limited lifespan and are likely to experience save failure as the years pile on - have you considered hacking your 3ds and getting a flashcart for DS games?
(You also won’t be giving money to scalpers on ebay)
-
Selling games from 10-20 years ago isn’t scalping.
-
I have 15 Nintendo handhelds on my last count. 2 of the 3ds are modded.
-
I have a pretty sizable collection and I’ve not had a game die on me yet, aside from save batteries that I’m capable of changing. I know the games can eventually die, I know it’s on the horizon, but they all still work for now, and I think even after they die I’ll enjoy the memories that the physical media provided me.
-
The problem with this, and most other “ai products” isn’t just that they are immortal attacks in human labor and and intellectual property, they also simply don’t work.
Same. I have spent years building my game libraries just for stupid shit like this.
You just described why it won’t be widely adopted.
I also have probably emulators for approx. 90 consoles / systems and have full set of games for most… Even if no game is produced anymore, we can buy current gen PC and console games, including Switch and Steam. In addition to emulation of older systems. And then there is the modding scene… with never ending content for out beloved games, even remasters from fans.
If the gaming industry goes wild, then I have no fear of missing out. And there are enough games (even to buy) that will serve me for the rest of my life.
i wish i knew how to use this stupidity, but i dont know what stock would be useful to get. I dont want to buy anything from america, ubisoft is bad investment always anyway and only finnish game company i can think/find doesnt even seem affected.
It’s Google, so it will last two years MAX.
I have no idea why people and companies still trust Google. When I did work on GCP we had mandatory maintenance every 2 months because some core service was changing. Hell I just got a notice last week that they’re shutting down another API.
killedbygoogle.com’s already polishing the new tombstone
The people running that site must be extremely overworked
So what i am hearing is buy in the dip?
no these are a bunch of dips making it happen
Investors never played a real video game. They dont undestand the difference between video (ai generated) and a video game.
Some of them tried to make video games, and that’s how they ended up with web3 and play-to-earn bullshit. Remember those? Barely? Yeah, same.
I tried an in-browser demo of something like this that Microsoft recently took down, and it was an image diffuser running an agent that could contextualize mouse+keyboard or gamepad gameplay inputs to behind-the-hood text prompts.
It looked like I was playing a Quake 2 clone, and almost played exactly like it, but weirdly turn-based when I didn’t do anything because it was just an AI generating images. It remembered the corpses of the bad guys I shot and it also kinda remembered the environment it made, including ramps that go up another floor and opened doorways that led to other areas.
Its cool, but not really a good game, very jank and likely resource intensive, which made sense why they took it down.
If you ever feel like you’re stupid, ignorant, absolutely microbe-brained, and that no one on this planet could possibly be more braincell starved than you:
remember that at least you don’t invest in the stock market for a living- Step 1 : buy some shares that fell down
- Step 2 : Wait for the IA fuckery to collapse
- Step 3 :
- Step 4 : profit
Yep, gaming stocks are on sale today because of stupid investors. Happily bought a few
Step 3: market remains irrational, loops back to step 3
So… No profit for me ? I’m sad, but not surprised
Your turn will come, gen AI. It’ll be a total shitshow because of the monster you’ve created, but it needs to happen anyway at this point.
