zlacker

>>feross+(OP)
I don't want moderation OR censorship. I want proper labeling so that I can self-filter out things that I don't want to see. Just like the labels on cans of food tells you exactly what ingredients are found within the can so you can skip the ones with peanuts, gluten, or some other substance you don't want; I want labels (and levels) on data.

If I don't want to see profanity, I should be able to set my filter to exclude profane comments. If I don't want to see nudity, I can set that filter too. Just like movies get a certain rating (G, PG, R, etc.), we should be able to properly label data.

>>didget+PG1
I'm with you on this. It also occurs to me, though, that by rigorously labeling all offensive content, you give ai a huge boost in creating it with perfect fidelity.

Which inspires another really weird, super uncomfortable thought. If the CSAM producers had cheap, reliable methods of creating their awful content without the use of real people, would that reduce the harm done?

I can't remember the last time I felt so conflicted just asking a question, but there it is.

>>pksebb+HP1
I hear you. But maybe AI shouldn't even be in the equation. It's expecting a robot to solve a people problem. We try to do this a lot in the technology industry. But no matter how great technology is, it's a mistake to think that it can fully solve social and communication problems.