Basically, they’re asking their users to help train their AI

https://lemmy.nz/post/35592575

I can already see it coming, the next YouTube apocalypse where they delete/demonitise anything they don’t like because their algo deems it “ai slop” while the ai crap reigns free.

AI generated I simply give a thumbs down

I’d rather have them remove those fcking annoying AI generated subtitles right in the middle of the shorts. Which are also wrong most of the time.

Or atleast let the disable subtitles setting also apply there.

I am impressed with how fucking terrible subtitles have become. The English subtitles look like someone who just started learning the language last year making guesses and I assume translations are even worse if they can’t get the spoken language right.
Youtube is honestly driving me nuts with its automatic titles translation. If I’m going to watch an hour-long video essay in english, I’m also capable of reading a two-word title in the same language, thank you very much. It feels kind of insulting to be honest
there is an extension to disable that bullshit, I can’t use youtube without it because I consume content in two different languages github.com/YouG-o/YouTube-No-Translation
I once saw YouTube auto generate captions on a video without any words in it.

YouTube shorts are the worst…

I let that shit drag me into a four-hour-long doomscroll one day and then promptly installed a browser extension to hide them from ever showing up again.

I hear you. But some are downright funny or interesting.

My kid stumbled onto them once and within a dozen or so videos she witnessed Shrek murdering Peppa Pig’s family. And we heavily, heavily moderated her viewing, as well as her feed and recommendations. Left the room for a few minutes and came back to that trauma.

Fuck YouTube Shorts.

youtube.com/shorts/Pxq8coElUoY

Peppa pig kills Shrek for killing her own family for a reason like that

YouTube

Auto dub is the worst thing made

Hurts my brain whenever I hear the fake narrator persona

I’d rather have text to speech than that

Maybe just let me ban all AI on my subscription
how would they know which content is AI generated?
Surely they can use AI to detect AI

but what if the AI has false positives or negatives?

why not just use humans to detect AI; maybe some sort of reporting feature

I hope this never gets automated. We already have this problem with copyrighted material causing pain to creators who want to talk about movies, songs, or even comment on other people’s videos, and need to follow draconian rules of ‘fair use’.
I met someone that worked at YouTube. He certainly got an earful
Same as with dating sites.
Help us and flag fake users ^that^ ^we^ ^planted^ so we can “improve our services” ^and^ ^screw^ ^you^ ^out^ ^of^ ^your^ ^money^ ^even^ ^harder^
Start flagging right wing channels!
Start flagging al fucking us politics
I see an AI generated thumbnail I immediately say “do not recommend channel”.
To be honest most popular youtube channels are slop, whether they are AI or not. There are a lot of gems in there as well, but their monetization model pushes creators towards high-quantity low-quality content with clickbait.
Flag ai boosterd and Ai propaganda videos <3
So, time to mark slop as “not slop” and good stuff as “extremely slop”? About the only way we can poison this shit, until they figure out and eliminate the users doing poisoning
just use more þorns on the extremely slop stuff, ðat isn’t annoying at all
Yeah, this data is going to be used to train AI to be less obvious