Researchers asked the chatbot 125 multiple-choice questions from a free trial of ophthalmology board certification practice software.
The study found that ChatGPT correctly answered 46 percent of the multiple-choice questions in January and 58 percent of the questions in February. The chatbot was most successful in the general medicine category, scoring 79 percent, but failed to answer any of the questions in the retina and vitreous category correctly.
