Sam, thank you very much for your kind words and additional insights shared here. If we casually use LLM-based GenAI (e.g., ChatGPT, Claude, Gemini), we merely think about behind-the-scenes datasets. However, if you collect and/or generate data for your own model building, discussion points here persist.