|
Top-level
 Also can I just point out how fucking hilarious it is that they've only been able to do this with DOOM and Minecraft, the two games that have more hours of gameplay footage just lying around the internet than any other thing which has footage ever created And it still sucks And somehow, this is going to "revolutionize" game development, when it can only do the two most successful (in gameplay footage) games in history and it STILL CAN'T DO THEM RIGHT 26 comments
@eniko There was a scifi writing prompt on tumblr where they invent a transporter with lossy compression and people become more and more like each other over the decades due to average regression. I keep telling people hyping genAI to go read it. Of course the ingest-millions-of-examples-and-extract-the-most-predominant-features-machine can make DOOM-looking sorta-ok-looking screenshots! That's literally the easyest ask! Now try get it to make something niche like Microsoft Flight Simulator  @lunarloony @elrohir @eniko Well, I mean, this is kind of like the argument that AI can't make pixel art because when each "pixel" needs to be a 16×16 square, it fails to line them up properly, but has anyone tried to have it make actual 16×16 sprites and scaled them up themselves? Artificial Intelligence is just a force multiplier for stupidity, as Mark Stanley says  @FrankHghTwr @elrohir @eniko I'd be tempted to try it for science if I wasn't vehemently opposed to doing so! @FrankHghTwr @lunarloony @elrohir @eniko Many decades ago I taught remedial summer school math for middle schoolers. My students were not happy to be there. One of the first questions they asked was if they could use calculators. I don't know if I knew the phrase "force multiplier" back then, but I told them that if they didn't know what they were doing, a calculator would just give them "bigger wrong answers." AI is the same on super steroids. @elrohir @eniko be careful what you ask for. It could probably make A flight simulator. It could probably replicate the leaked code for some version of Microsoft's. But if what you're asking is "make a model of planet earth to fly over" it'll probably pull a Douglas Adams by consuming the planet, and outputting 42 @eniko This so much. Every time they're pushing a new iteration of a product, when they're showcasing its output with HAND CURATED best case picks, it's blatantly, obviously dogshit if you pay any attention. It will always come short. With exponentially larger and larger indiscriminately pilfered datasets they define the exact contour of falling short with ever increasing precision.  @gkrnours @eniko they can just do what capitalism has been doing for the last 15 years (or more, that's just around the point it's noticeable to me!) and rehash the same thing over and over and over and over again forever! "Western" culture has been doing a 1980s revival for longer than the real 1980s have lasted  @hazelnot @eniko I think full AI game will remain a niche because the quality will be too low for people to keep buying. Game are a feedback loop of progress and reward. If you traverse a corridor, look behind yourself and it's a whole new thing, it's not a place where you can progress, it's a fever dream. Which sounds interesting but not battlefield interesting. @gkrnours@mastodon.gamedev.plac Everything is enshittifying without losing out on profits, I'm not convinced the same thing won't happen with all kinds of art, from music to movies to video games 🥲 @eniko stable diffusion has limitless potential for games! you just have to:  1. Release the first game as early access @fingerless @eniko just make a vertical slice demo, record millions of hours of those 15 minutes, and tell the AI to expand it to a whole game, duh! /s @eniko absolutely. Also like most ML things I don't even find the tech interesting. It's just a big optimizer and if you add some billion parameters to a model, that might just cover like any surface level data variation by sheer size alone. And it immediately breaks down with something like games or videos because the additional temporal dimension just would add impossible memory requirements. So tired of this wave of blackbox brute-force "programming" approaches these last years @shaperOfDefiance sure, if you ignore the environmental and ethical aspects it'd be great for Fever Dream Simulator 2024 but I feel like that's a pretty niche use case @eniko Yeah, that's the thing, I think this _looks_ very interesting and it would be incredibly cool if achieved by carefully (surgically, even) applying procedural generation to terrain, with properly herded generation algorithms, in a very limited fashion, for a particular level, with the intention of capturing a specific experience during a concrete moment of the game. This is not it and this is never going to be it. @eniko This is in general my posture with genAI, tbh. IF we were to look past all the layers of issues (and I want to be clear: we should NOT look past all the layers of issues) at its core it's an interesting tech demo and that's all there is. @eniko My favorite part about the Minecraft one is that they were crowing so hard about it being "playable". But "I can technically input controls and get a response" has never been what "playable" means in gaming. If a game from a normal publisher accidentally forgot the whole environment and erased all your work when you turned around, literally every review for it would have the word "unplayable" in it. @eniko I'm surprised that AI doesnt hallucinate twitch chat with random nonsense, considering if was probably trained on twitch VODs. @eniko what makes the whole "AI will always be bad" much worse is the environmental effect it is having, just to provide something that approximates an approximation poorly based on stolen data |
Gregory's Server
@eniko Once people are presented with 100% LLM-generated games, they'll like them just as much as they like the books that produces.