A rating system (1-5) that categorizes the clinical harm level of an AI error, from minor to critical.