Yep. An LLM is just always going to say what you’re wanting it to say. Especially with stored context now, all the shit and ideas you dump into one will be referenced and worked with.

It’s not the fault of the LLM, it’s just a machine. It’s the fault of granting access of one of these machines to people that lack certain fundamental understandings. Lest, they seek to have their concepts affirmed by something they don’t understand is any more sentient than their shoe.