avturchin answers Are LLMs being trained using LessWrong text?