Seth Herd comments on LLM Psychometrics: A Speculative Approach to AI Safety