Like it or not, the future of game development is AI: Google Genie 3 video

Darsxx82

Member



Today we are announcing Genie 3, a general purpose world model that can generate an unprecedented diversity of interactive environments.

Given a text prompt, Genie 3 can generate dynamic worlds that you can navigate in real time at 24 frames per second, retaining consistency for a few minutes at a resolution of 720p.

This is advancing fast😲

 
Last edited:
no

Edit: To expand on that, there is so much that this doesn't do that I don't know how it's helpful beyond just ideating. It's a lot of really impressive smoke and mirrors, but the utility of it in development is questionable.
 
Last edited:
Oh, lovely. One step closer to that crap making a whole game. We'll even have a holodeck-style thing where you just describe what game you want and we all end up playing one handed-games with huge titties everywhere.
 
You can either love it or hate it, but in my opinion, we might see more creative games again sometime. I bet there are tons of people out there with amazing ideas who have zero knowledge of programming and neither the time nor the patience to deal with it. Once you can create a game using just your microphone, some really exciting things could emerge. Of course, there will also be a lot of junk, just like there already is from the "professionals."
 
It seems everyone is frustrated with long dev times. So I don't love it but as long as the jobs lost in utilizing AI can be covered by jobs elsewhere, then I just guess it is what it is and it seems this could fix the long development problem of games. Projects that use less AI can market that attractive aspect.
 
Warning T-Rex GIF by Meta Digital
 
We will be not only flooded with endless ai videos but also endless ai games.

The fortunate upside is that all storefronts (save for like... itch.io) seem to be pretty adamantly against AI slop. So far, anyway. If and when this kind of shit tries to flood the market, there will hopefully be no market they're allowed to flood.
 
Last edited:
Well, they write themselves that this is intended to produce virtual training environments to train agents/robots. The context window / "interaction horizon" is only "multiple minutes". Models like these still have problems with simulating physics. And Google have not published a paper on this model, which makes all of this a marketing blurb based on "trust me, bro".
 
AI tools are enabling a lot of things. It seems reasonable that procedural generation of virtual environments would be one of those things. What happens in those environments to turn them into games is probably going to remain the domain of gameplay designers, but they're certainly not going to need a lot of the roles that are relied upon today. Artist, musician, animator and coding in many cases are going to be jobs that probably don't exist in the future of game creation thanks to AI.
 
AI benifits those already intelligent, those with an established sense of critical thinking, patience and an exisisting, serviceable skillset. Retards who use AI, become more retarded.
 
Last edited:
With every AI demo I have seen, it seems like the only thing they seem to showcase is shallow, open world, mmo-like games. I feel like using AI is trying to come up with a solution to a problem that doesn't even need to be created in the first place.

So many talks about long dev time, but so many developers get so bogged down in scope creep and trying to make their games super long or infinitely replayable or some variation due to gamers needing games to be endless because they are clearly addicted and don't have the self-awareness to realize this.

The industry caters to addicts. AI doesn't help game development beyond being used as a tool in a toolbox. It caters to pumping out content for addicts who finish these 50+ hour games in a week, demand games be sold for pennies, and have no appreciation for creatives. I thought something like Sifu was perfectly passable. It has both gameplay depth while also being cinematic without being a blockbuster. It had incredible art direction and environmental design that hooked me in more than straight hyperrealism.

I'm tired of games being entire playgrounds. Of course they take forever. You are trying to recreate a scaled down worlds. Before, big games were comprised of world maps that were heavily simplified. Then you had open zones which were a bit more complex but still manageable. It feels like people are trying to play God with their AI abominations that will look uncanny like something in Parasyte.

AI voiceacting, level creation, and art styles all feel superficial, inhuman, and devoid of emotion and mood. I don't mind AI as a tool, but it is being used as a cheap shortcut to make forever games.
 
This isn't the "future" of anything game-related. The amount of resources required to stitch together these useless scenes is astronomical - a massive waste of electricity.

It's slop, son.
 
This will never ever be able to make games or get used to make one.
Anyone saying otherwise simply doesnt get how this technology works!
Reasons:
- AI is super inefficient compared to game engines. (for example OpenAI is already losing billions - more users, more losses)
- Consistency over a longer period is impossible with LLM based tech. (Why do you think they only show a few mins?)
- Hallucinations are uncontrollable with LLM based tech. (How do you debug an LLM generated AI game?)
- It is impossible to actually design/refine any gameplay with this. (Good luck designing a whole game with just language!)

The only thing this is useful for is for game devs to get ideas for enviroments and test them out.
 
Last edited:
Love the idea of real time experience generation. Some seem so down on this, but I thought the Oasis Minecraft protocol showed amazing promise and would absolutely play a full title in that manner.

It's coming.
 
But can I generate the more risky stuff though? will I be able to generate walking around in strip club with the girls all wearing a cosplay all with a single prompt? If your AI is "censored" and not able to do that then human dev will never be replaced
 
This was one of my favorite features of Oasis. I loved the shifting terrain, the inability to maintain consistency. I think there could be some very good ideas designed around this.
You just said it - designed - which is not what you can do with those AI tools.
You can just run through endlessly generated AI enviroments, which could be fun for a while, but where is the gameplay in that?
 
You just said it - designed - which is not what you can do with those AI tools.
You can just run through endlessly generated AI enviroments, which could be fun for a while, but where is the gameplay in that?
I'm not exactly certain as of yet - but that doesn't mean new ways to use these tools will not come about. I think we can't fathom the future use cases of many of these tools at this present time.

I really do understand the cynicism regarding anything AI right now particularly with quality, but love it or hate it, it's just something that'll be as good as whoever is implementing it.

There is certainly a lot of "head in sand" / denial going on in regard to AI.
 
Last edited:
Oh, lovely. One step closer to that crap making a whole game. We'll even have a holodeck-style thing where you just describe what game you want and we all end up playing one handed-games with huge titties everywhere.
Don't underestimate the tech, should be able to make all shapes and sizes of titties. It's gonna usher in a new era of no handed gaming.
 
I really do understand the cynicism regarding anything AI right now particularly with quality, but love it or hate it, it's just something that'll be as good as whoever is implementing it.

because AI always looks good on the surface, and as soon as you look deeper it falls apart. that's true for basically any use case
 
Top Bottom