Was thinking that a lot of people will begin worrying about over sharing details of their life online when one of those chat gpts gets decent at scrapping the social media accounts on the fly and you can ask it to write a little biography of a specific person.
I think people are really underestimating how much the base model for these latest LLMs can infer about people's private lives just from what they publicly share. The models that people interact with are fairly neutered and don't display this capability, but the "base" models (without reinforcement learning with human feedback or RLHF for short) are much scarier in this regard.
