Jen on Nostr: it's difficult to build an algorithm based on ethics. Can it distinguish an innocent ...

it's difficult to build an algorithm based on ethics.

Can it distinguish an innocent photo of near-naked child enjoying a day at the beach, from a far more sinister image?

Is that picture of a cut encouraging self harm, or teaching first aid?

Is that nipple porn or breastfeeding advice?

Can it recognise sarcasm and satire?

Of course, there is also nothing encouraging Meta to try, as it thrives on views. It just needs to maintain enough trust to keep viewers.