“Dwarkesh Patel finds that for much of the writing he consumes, it is better to have a Large Language Model (LLM) explain the content, as the explanation is often superior to the original human-produced text.”