Loading…
18
Ran a small bot to find duplicate comments across 12 forums, the result split my whole view on AI moderation
It caught 3,000 repeats in under a day, but half were subtle typos or tone shifts that looked human to me, so is that proof AI can police intent or just proof it's training on noise, has anyone else seen this kind of split result?
1 comments
Log in to join the discussion
Log In1 Comment
hayes.tara1mo ago
Right, this EXACT same thing happened to me with a comment scraper I ran on a few gaming boards. It flagged stuff that was clearly the same person rephrasing their hot take, but then it also caught two posts from different users that were word-for-word identical except for one missing comma. That split result really makes you wonder if the machine is just matching patterns we've trained it to see, not actually understanding the MEANING behind the words.
7