The study conducted by researchers at the National Institutes of Health found that an AI model, GPT-4V, was accurate in diagnosing patients based on clinical images and text summaries, but often made mistakes in describing images and explaining reasoning. The researchers highlight the potential of AI in medicine but emphasize the importance of human expertise in accurate diagnosis. The study compared the AI model’s performance with that of physicians, showing that while the AI model excelled in selecting correct diagnoses, it struggled with image description and reasoning. Further research is needed to evaluate the effectiveness of multi-modal AI models in clinical settings.
Source link
New findings shed light on risks and benefits of integrating AI into medical decision-making
