One thing LLMs can help with: making simple judgment calls at the level that any reasonable human could do.

"This cake recipe says you should put in five tablespoons of tabasco sauce. Is that reasonable?"

"This flight search says I need to give my Social Security Number to do a search. Is that reasonable?"

These are the kinds of things that are "well, duh" for any reasonable human… but were very hard to do in a formal, computer-friendly way at scale in the past.

But LLMs, as a kind of magical duct tape, allow you to take this baseline for granted.