I mean, how else are you supposed to treat an LLM when the interface is prompting? You seem to get better results from them when you anthropomorphize them no?

I mean, how else are you supposed to treat an LLM when the interface is prompting? You seem to get better results from them when you anthropomorphize them no? So it's less a choice and more just using the tools as they are designed to be used and best used.