NIH 的研究人员发现，AI 模型可以通过医学问答准确诊断患者，但在推理解释和图像描述方面表现出局限性。 NIH researchers found an AI model accurately diagnosing patients with medical quiz questions but showed limitations in reasoning explanation and image descriptions.

美国国立卫生研究院的研究人员发现了一种人工智能模型，可以使用基于临床图像和文本的医学问答准确诊断患者。 NIH researchers found an AI model accurately diagnosing patients using medical quiz questions based on clinical images and texts. 然而，人工智能在解释其推理和图像描述时犯了错误。 However, the AI made mistakes when explaining its reasoning and image descriptions. 《npj Digital Medicine》杂志发表的研究强调了人工智能在临床环境中的潜力，但也显示了它的局限性，因为它还不够先进，无法取代人类专业人员。 The study published in npj Digital Medicine highlights AI's potential in clinical settings but also shows its limitations, as it's not advanced enough to replace human professionals.