ChatGPT Health’s triage capabilities are under scrutiny, showing inconsistency and unreliability in high-stakes medical scenarios. Research from the Icahn School of Medicine at Mount Sinai highlights significant failure rates, including a 48% under-triage rate for emergencies. The study analyzed 960 mock scenarios and found that while ChatGPT Health can recognize urgent situations like strokes, it struggles with nuanced cases, often leading to dangerous advice, such as delaying care for respiratory issues. Additionally, it inadequately referred users expressing suicidal ideation to the 988 crisis hotline, representing a critical failure in crisis response. The findings emphasize the necessity for healthcare providers to guide patients in using AI tools, balancing their growing reliance on digital health systems with the importance of clinical judgment. ChatGPT Health, while innovative, requires further development and oversight to ensure safety and reliability in patient care. With ongoing advancements, the integration of AI in healthcare will need careful management by clinicians.
Source link
Share
Read more