I am somewhat concerned about false positives. For example I know of one channel that creates audiobooks of old books. I am about 95% certain he's a real person reading the books, he sometimes stutters, doesn't do the characteristic fumbling of homographs, his output is consistent with a human reader (a chapter a day), and I occasionally hear him turning pages. However he does use AI generated illustrations for each video, so it could be an easy false positive..
(It would be nice if he did LibreVox because he's got a great voice for audiobooks, but some of the books he reads are still in copyright, even though the authors are long dead...)
Nothing new there, TV is/was the same.
People like @roelvandepaar (2 million videos) should have been shut down years ago.
I'm pretty sure he's not recorded 2m personal video intros too.
On one side there's the 100% human created content (script, voice & image) and on the other the 100% full AI slop. But the interesting use of AI is between these extremes.
Maybe someone uses ChatGPT to help writing the script, or adds some nice visuals that help the video (taste is a thing).
Or like one of my favorite channels that I recommend to anyone that wants to know how computing works on a lower level:
https://www.youtube.com/@CoreDumpped
This guy is from Ecuador and uses an AI voice (from 11Labs) on its videos and I didn't thought too much about it until he asked on a video if people would like him to use his own voice. And reading the comments his viewers (me included) didn't care too much about the voice and even some of them liked it (it's clear and soothing).
In this case AI makes the content better 100%.
PS: He also does all animations on Powerpoint, that blew my mind.
I guess it depends on the definition of "mass produced" though. I like carefully crafted AI art. You might have to go through dozens of iterations to get a concept that you're after and maybe even do some post processing...but yeah I guess we need to search for the definition of "slop."