If you can't critically read the output of an LLM, you shouldn't be using it for the given task. Many people have made the (good) analogy to an intern.