Note:
frames one and three should not exist anyway. They are advertising. It was already malign before generative AI.
And Hank Green, despite the morons trying to sound clever by saying he did, did not use an LLM to write the script for the insert segment in that interview. He used a phrase, which was already a common diplomatic phrase, in a completely normal way. The phrase has made its way into ShartGPT outputs because it was already common, and people desperate to seem like they aren’t fooled by LLM output are grasping at straws to find things to point at as ‘tells’ and turning into anti-tech-priests, consulting auguries to check for demons in the words of strangers.
There are no nice, neat tells. There are dozens of models, each with dozens of versions, each with some things that MIGHT serve as a tell, but you can’t keep track of them, and the right prompt will disrupt them enough to ruin the tell, so you can give up. You can’t reliably tell. That’s the whole point. Enjoy living in the derealized, boiling world our annointed leaders have made. Hug your loved ones.