RobinHa comments on An analysis of AI-generated content at the Mechanistic Interpretability Workshop