AI Slop is Detectable

Greg Morris from his post "Instagram's AI Crisis Is Not an Accident":

Meta's pattern is clear: refuse to be "the arbiter of truth" while arbitrating everything in favour of profit. They won't verify content, but they'll verify your credit card. They can't stop AI spam, but they can detect copyrighted audio in milliseconds. They're backing away from even basic responsibility as a distributor of information, the most fundamental duty a platform has to its users.

Instagram / Meta really don't care about stopping AI slop. They just care about numbers go up, regardless of the impact on society. If teachers and people hiring can run essays and resumes through AI detection software to at least get an idea of something being AI generated, Meta can afford to do the same with everything being posted on their platforms. Even a simple reporting that users can do where if you post more than ~10 things a month that are clearly AI generated, your account is labelled as an AI slop account for the next 6 months until you prove you're human again.

Adam Mosseri, the head of Instagram, had this post on Threads that inspired Greg's original post:

People largely stopped sharing personal moments to feed years ago. Stories are alive and well as they provide a less pressurized way to share with your followers, but the primary way people share, even photos and videos, is in DMs.

Stories are alive and well because Instagram shoved it down people's throats as the only way to actually connect with their followers or friends on Instagram. They stopped letting people just see a stream of photos and try to algorithmize the feed so they could mix in more ads. Then Instagram let AI run over their entire platform, and built their own Meta AI they continue to trick people into using by claiming friends of yours are already using it when they're not at all.

Mosseri's entire post feels like a reenactment of the Hot Dog Car Sketch from I Think You Should Leave: