DanielFilan comments on EIS VI: Critiques of Mechanistic Interpretability Work in AI Safety