AI Oversight and Human Mistakes: Evidence from Centre Court