Comparing the writing styles is reasonable. One caveat is that when using ChatGPT to check for grammar, spelling, and punctuation mistakes, if you don't put in the prompt not to rewrite sentence structure, ChatGPT will do that as part of English correction and then the original writing gets biased toward AI model. Not 99% or 100%, but it could make original content more than 51% that way.
People that couldn't afford fancy writing assistant software before but can now use ChatGPT for free will have much better original content.
The "personal judgement" is good. No matter what the test is, one should never assume it is infallible.
Playing devil's advocate though, I will point out that judging writing style might be a good approach, but ChatGPT can be trained to emulate that also.
I am being devil's advocate. I agree with what you are doing and I support the cause against derivative AI generated content, but I want to make sure that going after AI generated content on Hive won't be like workplace drug testing and eating a poppy seed bagel will result in a positive test result.