Social media moderation

[AllThingsTech]

As y’all know, social media platforms including Meta have recently relaxed moderation policies.

Youtube have now relaxed their moderation policies too - initially, Youtube would remove a video if a quarter of the video violated its guidelines, whereas the threshold is now 50%.

YouTube relaxes moderation rules to allow more controversial content

The Google-owned site has provided moderators with new guidelines and training on how to deal with inflammatory content that breaks YouTube's code of conduct, writes The New...

www.techspot.com

This avoids scenarios where news podcasts are removed for having a small clip of violence for example. However I don’t like the consideration given to the proportion of the video that is rule-breaking alone, given my view that the severity of the breach should be a more significant factor. In addition context and intent should be more heavily weighted e.g. there’s a difference between inciting or endorsing violence and posting such for educational purposes. It’s unrealistic for AI algorithms to dig into this and it’d be a resource burden on human moderators to have to navigate through such reports.

I’d love to hear other ppl’s thoughts

 

[data66]

IMO, they should use the AI to flag the videos, with reference points to where the infractions are. Then, the human mods will have an easier time going through the vids that may be breaking terms.

I agree, we should have news and educational content. Any junk (bad content) posted for "entertainment" or to incite violence or promote misinformation should be removed.

 

[Retro]

As I've said elsewhere, I Facebook's moderation sucks big time and that's an objective fact, not just my opinion. They sanction users out of the blue with only a generic message of having "violated community guidelines" and don't even let you see the post that allegedly did the violating. I believe that these are done by "AI", if you can really call it that hence the quotes, not human mods. This is beyond retarded and after I'd gotten nailed several times by this abuse, with increasingly draconian penalties, the latest penalty being a month long posting ban, I decided to visit Facebook a lot less and hardly post at all. Previously, I was on it for significant periods most days, now I can go for days without even visiting it.

One of the bans where I did manage to figure out which post it was, was when I made a comment about Japanese whaling being an inhumane and barbaric practice done for profit that should be stopped. I kept the post factual and polite, with no swearing. However, I got nailed for "hate speech". Like, wtf?! I think this is the one that might have been the final straw, it was a few years ago now.

@Tiffany I know you have some robust, well justified, opinions about Facebook.

 

[AllThingsTech]

Hmmm, @Retro did you get banned immediately after that post? If so I’d assume it’s AI.

I imagine the reason I haven’t had such an experience id because I only ever used to post photos of vacation and never anything heavy, controversial or political and I rarely post there these days. Note that my friend frequently posts up chess videos and blogs, and has never had an issue.

 

[AllThingsTech]

IMO, they should use the AI to flag the videos, with reference points to where the infractions are. Then, the human mods will have an easier time going through the vids that may be breaking terms.

I’m not sure how effective AI is at identifying AI generated images, given that plagiarism software that identifies AI generated content often has often found to be biased against neurodivergent students. There was a case where a neurodivergent student was initially falsely accused of AI generated content and received a failing grade.

Do AI Plagiarism Detectors Work? Here's What the Research Says

AI plagiarism checkers are under scrutiny as studies reveal false accusations, racial bias and a growing student backlash.

www.vktr.com

Of course there’s an appeals process and in the interest of fairness and avoiding potential discrimination, an “innocent until proven guilty” approach is taken. This means that a guilty person can deny plagiarism and get away with it!

 

[Retro]

Hmmm, @Retro did you get banned immediately after that post? If so I’d assume it’s AI.

No, I think it took a while, but not very long, perhaps a couple of hours or so. It's not like I got an email alert, either. Just next time I accessed FB, I got this great big in-yer-face popup blocking everything telling me I'd been nailed. So unpleasant.

 

[AllThingsTech]

Have these Meta bans been more recent - when did these occur? If they are more recent and haven't happened in the past, this correlates with AI moderation, and is more likely to be such