This is it. I posted a comment that mentioned "dumb ass" and Youtube blocked it and told me it could be against the community guidelines while showing it to me. I accepted anyway and posted it.
AI makes the panopticon possible. They're going hard.
Comments (15)
sorted by:
I was able to pick out words or phrases every time I encountered it, iirc. It was usually pretty obvious.
I only tried a couple of times. It might be that when I'm left with a phrase it's so innocuous I dismiss it. I do suspect it can only be so simple because it's so fast. It is however possible that they use more extensive content processing to derive their simple high performance filters.
A lot of these are basically the same as spam filters, literally one and the same in some cases. People do train them to filter out political content. They can be trained for example on the number of reports something gets so if you report something maliciously and a bunch of others get the same idea...
If you're in the UK you can do a SAR when they remove a content and by law they must tell you how they processed your data and made and automatic decision. If we can convince someone from the UK to go on the rampage with the DPA they can really fuck things up sideways for big tech operating in the UK. The SAR is based on the 1984 DPA not the GDPR.
I don't have enough money for chasing up Google on the SAR and suing them when they refuse but someone moderately well off and with some experience with using legal services will find it a breeze and quite cheap. Just ask people how to trigger filters and get things removed so they can then send the SAR to find out why.
Yeah, it filtered right-wing ideas, but I can't remember having trouble creatively rewording things to get the same point across.