"Study Reveals ChatGPT's High Misdiagnosis Rate in Pediatric Cases"
A study published in JAMA Pediatrics found that ChatGPT version 3.5 had an 83% error rate when diagnosing pediatric case studies. The AI chatbot's diagnoses were often incorrect or too broad compared to physicians' diagnoses. Despite this, the study suggested further exploration of AI in medical settings, particularly as an administrative tool, while noting the need for more selective training to improve diagnostic accuracy. The study also highlighted that ChatGPT's knowledge isn't regularly updated with the latest medical information.