Talked to a neighbor who worked at Facebook and it freaked me out

3 comments

3 Comments

My cousin actually worked on the moderation team for three years and what got him was how they train the AI to spot patterns. He said they feed it millions of posts that are already flagged but the system learns from every single thing you type even if you delete it. The creepiest part to me is that they know when you're about to post something controversial before you even hit send based on how long you hover over the keyboard. That's the stuff that keeps me up at night honestly.

nguyen.angela1mo ago

My neighbor tried to delete a rant about his landlord once and the app glitched and posted it anyway, so @gavin469 is probably right that the tech isn't that scary.

gavin4691mo ago

Your mileage may vary but tracking hover time before a post sounds more like an experimental feature that probably doesn't work as well as people think. Most AI moderation systems I've seen are pretty basic pattern matchers that miss context more often than not. The whole 'they know what you're thinking' angle plays into that fear of surveillance but the reality is usually way more boring and glitchy.