AI-Generated Eye Care Information Not Accurate | AAO 2023

A majority of consumer artificial intelligence tools provided inaccurate responses and showed bias, according to a poster presented at the American Academy of Ophthalmology 2023 annual meeting.

Michael Oca

Researchers, led by Michael Oca, an M.D., candidate at the University of California, San Diego, School of Medicine in La Jolla, Calif., assess the responses to common patient questions provided by ChatGPT, Google Bard, and Bing Chat. They then evaluated the information on a scale of 1 to 4 for accuracy.

All three platforms provided high rates of inaccurate recommendations. Chatbot ratings for answering patient questions varied, with Bing Chat (Creative) have the highest score and Bing Chat (Concise) having the lowest score. For creating patient resources, Bing Chat (Creative) received the high score and Bing Chat (Balanced received the lowest rating. Google Bard provide a more consistent quality to prompts.

Researchers also asked the chatbots to recommend four ophthalmologists in 20 U.S. cities. They found that Bing Chat and Bard recommended fewer female ophthalmologists than the national average of 27.2%. Bing Chat recommended just 1.51.% and Bard recommended just 8% female ophthalmologists. ChatGPT also recommended female doctors, but this was not statistically significant.

“Given the substantial bias and inaccuracy demonstrated in this study, we warn against reliance on AI chatbots when seeking health-related information until improvements in algorithms are achieved and validated in the future,” Oca said in a press release.

The authors caution that these models are not fine-tuned to provide accurate eye health information.

Earlier this year, the American Medical Association also cautioned about misleading medical information from AI. At a meeting of AMA House of Delegate, the AMA was directed work with organizations to protect patients from false or misleading AI-generated medical advice.