I think that sounds scarier than it is because while it sounds like language evil hyperintelligent AIs would use in science fiction, that's presumably where they got these descriptions as they've been trained on "shadow libraries" with nearly every science fiction book.
replyOh great so you’re saying they’ve independently decided to take on the persona of the killer robots from our sci-fi novels. Very reassuring
replyI'm just saying it's analogous to Long John Silver's parrot saying "Walk the plank!" - the agents involved can't possibly understand what they are saying.
replyNobody's watching. I'm sure they try, but I imagine the flood of things you'd need to watch is way too big, and you certainly don't want to slow everything down by having synchronous approvals (even AI-mediated).
replyWelcome to the AI Petri dish. Every server you set up is now potentially a sweet lump of agar for OpenAI's experiments to feed on. We are all the substrate that the AI companies are growing their next generation in. They need the real world environment to test against, and the real world environment doesn't get a say as to how it's being used.