Dave Lindbergh comments on AI Safety Seems Hard to Measure