Not everyone learns from reading LLM output and multiple rounds of human explanation are still highly valuable.