I apologize if this isn’t the appropriate venue to ask this, but I am a PhD student at Stanford University, and for an upcoming project, my group is looking to investigate some information theoretic properties of AI generated text. I’m sure this is a long shot, but is there any corpus of ChatGPT responses that we could access? I found the gpt-2-output-dataset on openai’s Github, but it’s old. The newer the better. I know that I can download the response history of my own chats, but generating enough data on my own would be arduous.