Human Authorship Ranking System
Kazy
So 95% of folks you surveyed want at least
some
AI label.But you don't want to have an "all-or-nothing", in case say, someone uses an AI-built tool in their world.
Well I have the system for you!
Kazy's Really Cool Authorship Ranking System
So it's based on the Avatar Performance Ranking System, which goes from Very Poor to Excellent. It's imperfect, not everyone likes it, but it's
something
. I would say it works because
it is imperfect.These all have metrics that let users set their own optimization standards. But what if we also applied it to authorship? Alright, I don't mind demonizing AI users with a "Very Poor" label on their world, but you probably do. So how about instead of the above, we have this:
Authorship | Performance
|Prefab/Unknown Sources |N/A
Generated | Very Poor
AI-Driven | Poor
Tool-Assisted |Medium
Human-driven | Good
100% Human | Excellent
Names pending. (this is
your
job, you should have thought of this idea anyway. I can't do everything.)The rankings.
Or "BUT WHAT IF THEY JUST USED A TOOL THAT WAS MADE BY AI?? HUH???".
Prefab & Unknown Sources
Since this is an Authorship scale, a non-ranked label should be "Prefab", where any upload of a prefab world where little has been changed gets this tag in addition to whatever else is here.
Unknown Sources
is use of a tool or shader that you don't really know if it's authorship is 100% human or not.Generated (Very Poor)
Much like any avatar over 70,000 polygons, any generated artwork or geometry goes here. People should hire artists and graphic designers, and there's plenty of non-AI stock stuff out there. This is also what most of the people have an issue with, so being able to filter just this will probably make a good 70-80% of the people who want a label happy. Also any untested AI code goes here.
AI-Written/AI-Developed (Poor)
This is where purely AI written code or LLM story content lives, but you've (hopefully) thoroughly tested it and at least have a rudimentary understanding how it works.
Tool-Assisted (Medium)
The middle ground. Someone who knows Udon, but uses AI to help write code, or heavily uses tools or shaders developed by AI. Probably the kind most people are okay with.
Human-driven (Good)
Most of the code is human-written, with maybe a vibe-coded app to help create tools. Most of this is where "well this person uses AI, why not be mad at them?? actually goes when it's not the sort of generative AI people usually get mad at.
100% Human (Excellent)
A lot rarer, like an Excellent avatar. 100% pure hand-crafted.
Filtering
Like Performance ranking, the upper ones should be an always-strived for goal, but not mandatory. If a filter is implemented,
it would only block AI-Written/Developed & Generated
worlds, as not to overly punish people who use these tools to actually be creative (Similar to how you can only use filtering to block Poor & Below ranked avatars). This is a compromise by me, because I genuinely would rather not have to deal with anything
AI, but I feel this is also a compromise most critics would be okay with. Along with Good/Human-driven and up getting its own category in Jams and Spookality
It's a very imperfect system, and would mostly be an honor system thing. I feel like knowing you can still be seen would encourage honest labeling, and only extreme dishonesty would require intervention (eg, something labeled "Human-driven" was filled with obviously generated artwork). Punishment would stop at labeling with a "Contested" label at most, with
maybe
a force change of the tag.##
There are probably tons of opinions on what usage would qualify for which label, but in the end, it doesn't really matter beyond slop, which is what most people care about.
Thanks for coming to my KAZYtalk, and please paypal me my consultation fee for developing this idea.
^this very cool idea is 100% human made and this post was written by hand^
edit: also, like steam, have an "explanation of label" section of world info, also a reply section a world owner with "contested" can fill out.
Photo Viewer
View photos in a modal
Log In
Jay Renketsu
Yes, this sounds like a really good compromise which covers what the 95% were asking for.
mousse․
Absolutely for this. Also i will consume a whole bottle of hot sauce in one sitting if vrchat ever does this.
Jasper Raine
I am 100% on board with something like this.
mBlade
This is a great idea! Lets hope it gets implemented very soon before the slop gets too bad.
Kazy
I'm sure there's a lot I'm leaving out but the Avatar Performance Ranking system doesn't cover everything either so �