How is this an "anti AI hot take"? It's discussing using one type of LLM output versus another...