im for of this. I don't say AIs are totaly bad, but at least we must do something AI Enhanced Lier Thumnails. those, "lookin' good at a glance" will break how community labs work.

From the standpoint that we should address the "practical" aspects of the problems surrounding AI, I think the important thing is to eliminate slop.

When it comes to worlds, I believe slop is already being screened out during the process of moving from Labs to Public. I understand that the criteria here are not based solely on visit count (I suspect they use a minimum number of visits as well as the ratio of favorites to visits). Therefore, slop should be filtered out among worlds that become Public.

This does not prevent Labs itself from being flooded with slop, but I think the restrictions of one upload per week and allowing only User-ranked users to submit to Labs keep the problem relatively contained.

The practical problem with AI is the mass production of low-quality content. Mass production is mitigated by the restrictions on Labs, while low quality is mitigated by the process for becoming Public. I think this is what underpins my sense that VRChat is not rapidly deteriorating, unless one takes the position that content should be banned simply for being AI-generated in and of itself.

As a fairly typical Japanese user, I do not see avatars as a problem (because when I use Public avatars, they are generally what you might call joke or novelty avatars; and as for those on Booth, that is not an issue for VRChat to control), so I will not comment on them.

ja:

AIの問題について"実用的"な側面について対処するべきであるという立場から

重要なのはslopを排除することであると思っています。

ワールドにおいてslopは既にlabsからpublic化という過程に於いて判定されていると思います。

ここにvisit数のみではない基準があろうことは認識しています（最低visit数と、お気に入りとvisitの比が使われているのでは無いかと思っています）

なのでpublic化されるワールドにおいてslopはフィルタされるはずです。

labs自体にslopが溢れることはこれでは防げませんが、週間1アップロード、そしてUserランクのみがlabsに提出できるという制約がこれを緩やかなものにして居ると思います。

AIの実用的な問題は低品質の量産であり、量産についてはlabsの制限、低品質はpublic化によって緩和されています。

これが「AIであるということそれ自体によってBANすべきである」立場以外において、VRChatが悪い状況に急速に陥ってはいない体感を担保していると思います。

なお典型的な日本人としてはアバターに問題は感じていない（なぜならPublicアバターを使うときはいわゆるネタアバターであるからだ。かつBoothにおけるそれはVRChatが制御する問題ではない）ためコメントはしない

I get your feeling. But, that is kind of wrong.

I rather want to "follow" or "hide" creators if something need to change.

Questions:

My answers are:

Here's my thoughts of AI in general.

This is the new era, kind of. I think tooling always evolved for efficiency. Like for somebody who want speed, can't do it for various reasons, or being lazy, "i don't care about this process too much"..... Probably majority of us already don't care about how game engine, polygons, shaders are made. It runs, all good. I see AI is the next step of it.

There are creators in this space that majority would never ever understand nor care about what they do. Who cares how my works is aimed to have better corner roundings? Someone does, and it's cool! Others don't so much, and it's fine too.

I'm not gonna AI-generate 3D models or pictures anytime soon, but some snippet of codes. It's a cool technology.

Typed by my hand using my fantastic PC setup.

明確に反対する。いずれAIによる創作と人間による創作はまだら模様のように分かちがたくなってゆく。現時点でラベリングをしたとして、数年後には9割がラベルを張るという結果になるだろう。そのような仕組みに意味はない。

それよりは、手作り側を認証したほうがまだ実用的であるように思う。

--

I strongly disagree.

Over time, AI-assisted creation and human creation will become increasingly intertwined, to the point where drawing a clear line between the two will be nearly impossible. Even if we introduce labeling now, in a few years we may simply end up in a situation where 90% of content carries some kind of AI label. At that point, such a system would have very little practical value.

If anything, I think it would be more useful to certify or identify content that is genuinely handmade instead.

I second the idea of a declarations like thing, "Nutrition Facts". Blanket targeting everything touched by AI in any way will result in people evading self identification, especially in areas where you can't actually see.

upvoted, while Ai has its use like code assistance and dirty work like long research and ui backbone connections, I heavily agree that Ai shouldnt be allowed on the platform period without being required to be tagged because its becoming a real problem to not spot a real world in a literal sea of ai generated worlds

I don't understand how VRChat is supposed to enforce an AI content filter when people can just lie, and outside of using our best guess we won't always be able to confidently discern AI content from hand-created content, which can result in genuine creators being flagged and punished over improper judgement. As much as I hate the surge of AI content flooding worlds, a tag isn't a solution, and will only cause headaches for their support team. It would be nice for some transparency from VRChat with regards to how they are assessing this on their side, even if the answer ends up being that we can't reasonably do anything except support good and authentic creators.

We can't have slop in this platform too

An AI "Nutrition Label" Where world creators can indicate what kind of assets involved AI would be a welcome edition.

It could be something simple as a set number of categories where a developer can check whether or not they used AI in the creation of them. For example, artwork thumbnails, models, code, textures, sounds, layout, and a master checkbox labeled entire world which would assume all of the above.