Medicine beyond machines: viewpoint on the art of thinking in the age of al

Abstract

The widespread adoption of large language models is increasingly shaping clinical decision-making by altering how physicians engage with data and reasoning. While these tools enhance diagnostic capacity, streamline workflows, and support learning, their misuse may diminish critical, contextual, and humanized thinking, reducing physicians to passive validators of algorithmic outputs. This paper explores the evolution of medical cognition and proposes strategies for integrating artificial intelligence in ways that preserve cognitive autonomy, such as structuring information, reducing bias, and strengthening metacognition. We argue that artificial intelligence should serve as a “cognitive stethoscope,” a tool that refines reasoning without compromising its essence.