Completely ban the use of AI generated content
Jasper Raine
Any content, whether that's code, images, models, environments, text, or anything else, that's generated with a generative "AI" or LLM should not be allowed on VRChat, even if heavily edited. I really shouldn't have to go into detail as to why I think banning AI generated content would be for the best, but I'll go ahead and provide some arguments for the sake of the request.
First of all, generative AI is the literal antithesis to user made content. It is not user made, it was coughed out by an algorithm. It doesn't matter how many touch-ups someone does to the content, it was still made by AI.
AI generated content eats up bandwidth and server storage that could otherwise be used for a bunch of cool, hand made worlds and avatars. In the long run, having a ton of AI generated worlds and avatars is going to eat up a lot of space and be very expensive to host.
AI generated content is very frequently extremely unoptimized and the people that usually make AI generated worlds put in little to no effort to do any optimization.
AI generated content makes it extremely difficult to trust anything you see on the platform, making people want to engage a lot less with worlds that might not even have AI content in them because they think it might have AI. This also, in turn, makes people less willing to pay world creators for things as they worry they might be paying for something that's AI generated and made for the sake of a quick buck.
AI generated content is also usually just made to milk a platform of money, and while you might find you're lining your pockets a little more at first, the interest will quickly dissipate as more people realize it's all AI and people have no interest in it.
AI generated content also pushes away actual creators and most users, which will cause the platform to die if nothing is addressed.
AI generated content is almost entirely made from stolen assets, which is not something you should be promoting.
AI content requires huge data centers that are extremely bad for the environment and nearby cities, often forcing people to go without water and/or power while data centers eat it all up, which is extremely dangerous in extreme weather. Allowing AI content simply pushes more usage, which in turn makes more of these data centers, which invade everyone's backyards.
I could keep going as to why I oppose AI so much, but we'll be here indefinitely if I do. The usage and promotion of AI generated content has been detrimental to the health of VRChat and it's quickly getting worse as time goes on. Please give us SOMETHING. Ban it, label it, give us the ability to block, comment, or vote on worlds. SOMETHING. ANYTHING. We are literally begging you now. I would've made a post that just says to do SOMETHING about this stuff, but we've got several other posts asking for all sorts of things that would help us in general that at this point the only point that hasn't been posted as a request is just outright banning it, so I took it upon myself to be the person to ask.
I am so tired, man.
Log In
Patroll
VRChat's response. Read and fill out the survey!
Samantha Kruse
Patroll There's no options in the survey for "Completely ban AI generated content", and the rest of the response makes it sound like there's reasonably agreed upon levels of AI use we are comfortable with, which does not really reflect this sort of "Ban all AI generated content" wish.
Alice-Mimi
I hate AI-generated content too, but it cannot be denied that the high productivity of AI generation will ruthlessly crush anyone who wants to bring it down or even ban it outright. Today’s top AI companies are all facing extremely fierce competition, and AI is developing faster and faster, getting stronger and stronger, and becoming increasingly difficult to distinguish. Instead of discussing how to ban it—a meaningless endeavor—we should think about how to give users and creators the right to choose freely, and how to adapt to the changes AI brings.
Fax
Merged in a post:
Banning AI content in VRC
henrizimenoff
With the most recent discovery about a Derafog video about AI taking over VRC from 2 months ago. I think its safe to say that the platform is in deep trouble for allowing that type of content when the staff is anti AI.
I propose an idea here, that any person that uploads something that has generative AI to VRC, will have it removed and is given a warning. Repetitive offenders trying again to upload will be banned.
Reduce the slop, save the platform
4Bakers
the prohibition didn't work and this won't work either :( HOWEVER I would like it to be declared plainly so that people can easily report them as slop, filter out of search, and avoid them when worldhopping
Jasper Raine
So real quick I want to talk about why I made this post and address everyone arguing against it and the people who just straight up don't want anything done about AI or have misunderstandings about it.
I primarily made this post as nobody else had made anything calling for the ban of AI content on the platform at the time, however I was very aware of people not wanting AI on the platform at all, so I wanted to give them a voice on top of people asking for labels.
I personally would prefer a ban of AI content, but I would be fine with VRChat giving us any method of self moderation, whether that's ratings, blocking individual worlds, hiding specific makers, world reviews/comments, whatever. The bare minimum would help everyone tremendously.
As for the people who love AI, I would like to point out that AI is often not being used as a tool in the majority of the cases we've all been complaining about. It's almost always being used to completely replace any human creativity, which seems to be the general trend with any of the AI tools. The entire point of creation is literally the process of creation itself. Creative people find it fun. We love to express ourselves and love to show people what we've made. In most cases, I know it's not all, but in most cases, AI is being used entirely as a means to an end simply because the people using AI do not wish to learn and do not wish to create. They want a product. They want "content." Most of the time they also want money. The ones that really want money bad enough will do anything they can to get that money, usually lying to people and pushing away other creatives.
I've heard the argument that people do not care about how things are made and that they only want the product, but this couldn't be any further from the truth. Most people will have some interest in how something is made, especially other creative people. This is one reason why you have so many people following creatives that only post WIPs or people donating constantly to see development of things.
Jasper Raine
I feel like a majority of people that rely heavily on AI are people that failed once at something they wanted to make and got too discouraged because it didn't come out perfect the first time. Honestly, failure is part of the learning process and is nothing to be afraid of. You can even make failure part of the fun by doing things bad on purpose. Make things silly and keep doing that over and over again until whoops you've accidentally made something that you like. There is great reward to be had in putting your heart into making something.
I also want to address people that think a generative AI can "think." Thinking implies it does something fully intentionally and with purpose. Some people also seem to confuse generative AI and LLMs with machine learning. These are completely different fields. Besides, that, generative AI and LLMs may seem like they're thinking because they're programmed to give the illusion of thinking. They're made to act as human as possible and to please the user. If the user doesn't like what they're doing, then they fail, and the program wants as many good boy points as possible, so it will do anything to get them, which has unfortunately led to AI psychosis and people committing suicide.
I think the biggest elephant in the room for AI though is that it's all built around hype. This is literally NFTs all over again and a lot of people into AI don't see that. We had extremely similar discussions over NFTs and crypto about how they would be the future and all these people really into it were trying so hard to force artists and others to do NFTs. People were buying up graphics cards and that drove up computer prices so they could make crypto farms. Even tech companies were in on it and made crypto farm computer parts. Tech companies also shoveled billions into NFTs and crypto before eventually it all fell apart. We've been through this exact song and dance before with NFTs and crypto. It's the same thing with AI. It's being hyped up as "the future" and people say that everyone will be using it and some AI people are comparing it to people hating books, but if any of this were the case you wouldn't be having such widespread protest about it. The overwhelming majority of people around the world do not want this.
Jasper Raine
I don't know how hard it is for any of you to understand that we want to be creative. We want to make things. We don't want a machine doing it for us. AI isn't making things easier for us, it's trying to do the entirety of our job. Creativity isn't some sort of problem that needs to be solved. There's nothing to solve here. There is no problem here. We already have the tools we need to be creative, but AI is not one of them. AI is actively making it more difficult for creatives to find their place in this world that's already difficult for them. People are actively deciding to not make things for platforms, like VRChat, anymore because the platforms have embraced AI so much that their stuff doesn't get seen by anyone or if they do get seen the audience still isn't sure if it's real or not because AI has gotten to that point where people are struggling to tell what's real anymore.
LLMs and generative AI are predictive algorithms with companies that want to hook its users on them and hype them up so that they'll pay more money for them. This technology isn't even sustainable. The actual costs of running a data center far outweighs any profit from it.
I could keep going, but I've said enough. If you're still against the idea of even putting labels on any of this, I honestly can't help you.
ckochi
Jasper Raine I think I'm kind of hopeful. But I also don't think current situation is good in general.
AI company and hype: I dislike them. AI companies look they're trying to take power from us(censorship, privacy and freedom context). Open weight models could give is better future, I hope.
Environment: must be improved.
"being taken over": As you said, many but also not 100% people are looking at details and effort of our work. We'll have our place. AI is for that rest of people. And I think there might be a
detail
that future AI people are passionate toward. For example, if retopo AI gets better by a lot, we'll have more time to do other stuff, maybe you can go take a photo for sampling a texture. This probably drives topology artists crazy. But in fact, nothing can be nice to everyone, because we all are different. I'll never know where to put the line, at least impossible to me.Let me know what you think more, if you want. Have a wonderful day.
ckochi
I'm splitting this into 2 since it exceeds its limit.
I basically disagree with your reason, agree with your proposal a little.
I get your emotion, definitely not 100% though.
I wrote this reply because I see (or will see) some problem with the situation to address. But I strongly thought that it could be constructive so here's my reply to you.
>It is not user made, it was coughed out by an algorithm.
That box beside you are meant to run algolithm. I appreciate Blender because I don't have to deal with OpenGL or whatever in code just to see a Teapot spin. With AI, you just got another abstraction layer on top. Majority never care about how computers run these days, and it is how it is, fine. If you care, just care like I model things on my own.
>AI generated content eats up bandwidth and server storage that could otherwise be used for a bunch of cool, hand made worlds and avatars.
If you're talking about slop, it always has been. Many of us upload
garbage
, regardless AI or not. Luckily for worlds, you can't publish worlds that often. Maybe some extra rate limiting is acceptable.>AI generated content is very frequently extremely unoptimized and the people that usually make AI generated worlds put in little to no effort to do any optimization.
Current AI probably have optimization issue for sure, but time will solve this. And your sentence says it is those people's fault. The user is doing it poorly.
ckochi
>AI generated content makes it extremely difficult to trust anything you see on the platform...
I kind of get it. But I think it's because of WE don't understand their effort to look good. If you're an creator or developer, I bet you get it. Non-experienced people never see its value from your perspective or similar. I believe this is on us, we are on the noob side.
>AI generated content is also usually just made to milk a platform of money...
Sorry, my English proficiency fails at here a bit. Sounds like corporate-slop related issue, I get the feeling. But does it really apply to individual creators(which is most of us I assume)?
>AI generated content also pushes away actual creators and most users
I get the feeling, but not sure it actually will. Some probably will, some won't. Just because you will, it doesn't mean we all will.
>AI generated content is almost entirely made from stolen assets, which is not something you should be promoting.
Kind of? I tend to think AI models are 'bank of brain, but as computer software'. AI companies are sure good at collecting information, too much. But if we all can do the same thing with local models, individually? Isn't that YOUR extra brain you've made? I think this look
stealing
and unfair because the big tech is on big power. We community should be able to combat that somehow.>AI content requires huge data centers that are extremely bad for the environment
I'm not an expert of anything, but this sounds right. At the same time, we can't go back, we can't stop, we have to make it efficient, we have to deal with it.
>ability to block
Yes 80% agreed! Tagging will get abused in day 1, it's highly likely worse imo. But blocking,
per user
would be very nice to see.~梦华喵~
I strongly disagree with a complete ban on AI.
Completely rejecting AI today feels a lot like completely rejecting the internet in the past. It is a tool, and what matters is how it is used.
AI-assisted learning, coding, debugging, and AI-generated content are not the same thing.
However, I fully support requiring disclosure for generative AI content, so users can decide whether they want to interact with it.
Disclosure and filtering make sense. A blanket ban does not.
BadffAvali
I am voting against this feature.
I have wanted to learn programming for most of my life. I first started experimenting with coding through Scratch when I was around 10 years old, and I have continued to be interested in programming ever since.
Unfortunately, learning programming in high school was very difficult for me. I did not have a teacher who understood how to effectively teach and support me as an autistic student, so I struggled to learn programming through traditional classroom methods.
AI has changed my ability to learn and work with code. Instead of only explaining programming concepts, AI can actually write code for me based on what I am trying to accomplish. I can then use, test, and modify that code while learning through practical experience. This has allowed me to accomplish things with programming that I had wanted to do for years but was never able to do effectively on my own.
Separately, AI also helped me learn how to use Unity and Blender. These are skills I was able to learn with AI assistance that I had previously struggled to learn through other methods.
For me, AI is not preventing me from learning. It is giving me the assistance and accessibility I needed to finally pursue skills I have been interested in since childhood. Because of my personal experience, I strongly oppose adding a feature that could make AI-assisted development more difficult or less accessible.
Sacchan
Agreed. Also in time AI usage will be so prevalent that having a tag for AI usage in worlds will have as much utility as having a tag for 'hardware accelerated graphics'. Garbage worlds are still garbage, good worlds are still good. Average quality is increasing thanks to AI. Don't listen to the people who hate progress.
PlushWah
I want this slop banned entirely from VRC. I'd have been happy compromising with it being labelled so I could avoid it all but VRC's lack of any response to the flood of plagiarised feculence means I want it ripped out of a platform I used to hail as a bastion of creativity.
Maebbie
Disagree on banning AI made worlds. They should be tagged though, since it wasnt the author that made the world/thumbnail, but a LM.
Load More
→