Maybe I’m confused, but why would an llm be better at mapping tuples to functions as opposed to a kind of switch statement? Especially since it doesn’t seem to

Maybe I’m confused, but why would an llm be better at mapping tuples to functions as opposed to a kind of switch statement? Especially since it doesn’t seem to totally understand the breadth of possible kinds of faked data?