No, I mean managing an AI to achieve a random task. Prompting, iterating, filtering - they all require high level input from the user. A LLM is a complex beast, not easy to use (yet).

Students that can't write well can't tell if the generated output is good enough either.