Despite these shortcomings, researchers believe that the chatbot could be improved with specific and selective training on accurate medical literature, as well as real-time access to medical data. The medical field has been an early adopter of AI technologies, and many see the integration of AI chatbots into clinical care as inevitable. However, the study underscores the invaluable role of clinical experience in diagnosing complex medical cases.
Key takeaways:
- ChatGPT-4, an AI bot, has shown poor performance in diagnosing pediatric medical cases, with an accuracy rate of just 17 percent, according to a study by researchers at Cohen Children’s Medical Center in New York.
- The study suggests that the AI bot struggles with recognizing relationships between conditions and requires more consideration of the patient's age, which is critical in pediatric cases.
- Despite the low success rate, researchers believe that with specific and selective training on accurate medical literature, and real-time access to medical data, the diagnostic accuracy of AI-based chatbots could improve.
- The medical field has been an early adopter of AI technologies, with mixed results, but many doctors see the integration of AI chatbots into clinical care as inevitable.