Study Finds AI Chatbots Like ChatGPT Perpetuate Racial Bias in Medical Advice

AI chatbots like ChatGPT perpetuate false, racist medical ideas about Black patients when asked medical questions, Stanford researchers find.

Incorrect answers included fabricated, race-based differences in skin thickness, lung capacity, kidney function.

Such AI biases could lead to real-world medical harms if chatbots are used for tasks like emailing patients.

Patients already asking chatbots to help diagnose skin lesions, other symptoms, concerning doctors.

Researchers urge more testing of commercial AI systems to ensure they are fair, equitable before use in medical settings.