Logan Zoellner comments on “Learning to Summarize with Human Feedback”—OpenAI