Aaron Gertler comments on Our forthcoming AI Safety book