Shmi comments on Advice for Entering AI Safety Research