Daniel Kokotajlo comments on Six Thoughts on AI Safety