Skip to main content

Indie game storeFree gamesFun gamesHorror games
Game developmentAssetsComics
SalesBundles
Jobs
TagsGame Engines

I'd like to upgrade my video card to be able to run AI models (Like SD XL, music models or big LLM models) to experiment and find how i could include it into my workflow. (I have an 8Gb VRAM AMD). But yeah, prices of nvidia cards skyrocketed before ram. Luckily i don't have to change RAM yet.

Hopefully when the bubble burst the manufactures will return to provide hardware for the average user. (More offers lowers the price).

Also what mid said 'The majority of software doesn’t justify at all the insane amount of RAMs it consumes, or features it demands.' Is like an 'Unreal 5' situation. If things keep like this, people will start to (hopefully) search for better optimized software and games. Perhaps is a good opportunity for AA and indie devs. Since many of us use have and work in average consumer grade computers, we  can optimize (and pretty much we have to due to the limitations) our games, making them more playable in pretty much every computer.


And it's also like you said: 'Every shortage can be resupplied very fast, because companies fight for customers and the consumer market is still significant.' But if i'm not wrong, many of the RAM stock was already bough even before production (idk if true, but if it's). And the rest of the RAM manufacturing companies  aren't increasing production because they are also waiting for the bubble to burst. (If they expand now to increase the production and the bubble burst in a couple of years the losses may send them bankrupt).

(+1)

So, according to your understanding, the main thing is how fast the AI bubble bursts? Not IF the bubble bursts? I personally think that this bubble won't fully burst and the market won't collapse, AI models will integrate with society as a tool to speed up processes, but they will never replace human mind and decision making. The real question for now would be safety of the customers, transparency of the AI products, user privacy and cultural/social impact.

I am agree. When i said the bubble will burst i don't mean like AI will disappear. But it will be something more like the early internet. (dot-com bubble) Or like with videogames in the 80's (videogame crisis of 1983). And like you said, AI will integrate in society. The same way internet, videogames, music, TV,  etc, did.

And yes, AI will never replace human mind and decision making. It's a tool. Perhaps it will increase the bar of quality standards but not more.

(What i mean is, for example, now indie devs can use pre-made game engines like S2, Unreal or Wicked to create realistic graphics without the need of a huge investment).

'The real question for now would be safety of the customers, transparency of the AI products, user privacy and cultural/social impact.' 


I think that depends from us, both as game devs and human beings. Since years we could barely trust in internet for reliable info (Unless you know where to search.) And sadly, with AI i think the phenomenon will not stop but increase.Perhaps that's good? the more people trust less in internet, the more people will spend their time out there, or become harder to scam, or prone to double check info? That could be a problem for social networks but video streaming services, or games stores may be the less affected.  IDK.

And about AI use. I think the best is try to use local AI instead of online. Using things like GPT4ALL, Koboldcpp or LMStudio for local LLM models (i mean ChatGPT like models) , or using things like SD 1.5 / SDXL (image generator models). More privacy, no ToS.


I think the only thing  we can do is to adapt to the new world and help our friends/family/loved ones to adapt as well. 

(+1)

Okay now I understand your point, thanks for explanation! This is reasonable approach to the topic and I agree with it, especially with local AI usage, I think PewdiePie is doing something like this but I didn't dive into it... I just hope that local models will not become something like Tamagotchis in early 20s (that's kinda a big flex comparing AI to japanese toy collection lol). It was personalized, technologically appealing and interactive, but it remained one-time hit and didn't stick into our culture that much

Tamagotchis? thats a name i didn't hear in a long time. hahaha. You know you can still buy them right? XD.

I don't think local AI will become something similar because even companies need them. For example, if you are a big company and want or need to use AI (any kind) for something. You can use a third party AI But that means sharing private info with the third party ai's owner. So if you want to have your own, local AI running in consumer grade hardware is more cheaper and easy to maintain, customizable and upgradeable than a complete datacenter. 

Also, i think Local AI (like anything local) will still be alive as much as a market for that exist. This may be the only thing that gave us, the end user, a small and sometimes huge amount of power. The less people use cloud-ai's -> the less profitable is -> less companies trying to push it. (This applies with anything.)  Also, alternatives exists since always: In a world dominated by Adobe you can find Affinity, Inkshape,Krita, Gimp, Paint.Net. In a world dominated by Cinema4D or 3DS there is Blender. In a world desktop is windows there is Linux and Mac. None better or worse but different. 

But in a world where most of people is used to use Netflix or Spotify premium (easy of use over privacy), well, it makes me don't put my hopes too high.

And even that, we already have plenty of open source and local AIs (SD, Wan, LLama,Quen,Deekseek,Gemma,LTX,DiffRymth). I also tried one LLM that ran locally in my android phone. (It was slow and heated the phone a bit, but it was something. XD).

Sorry for all the text. I need to learn when to shut up. 

XD