Comment by naasking

> LLMs are not that smart.

They are smart, but they are not aware of the environment they're in, or any implicit context that someone whose doing a job carries with them, that's why all of that context has to be explicitly laid out in a prompt. When the context is provided, they are quite smart.