RussellThor comments on Recommendations for Technical AI Safety Research Directions