Sherrinford comments on What is AI Safety’s line of retreat?