AI is changing medicine. Medical education must change, too. | Viewpoint

Artificial intelligence is poised to revolutionize most industries, and medicine is no exception.

Eight in ten doctors already use AI for everything from taking notes to getting up to speed on the latest medical research in order to better diagnose patients.

AI promises to deliver enormous benefits to clinicians and patients alike. But like any tool or technology, AI also has limitations.

Medical educators need to adjust their curricula in response to the AI age. That means putting less emphasis on memorization -- and more on developing an empathetic bedside manner, communicating well, and handling the many other uniquely human elements of the job.

Doctors have the unique privilege and responsibility of being invited into some of the most intimate and vulnerable moments of patients' lives. Physicians often meet patients and their families when they are confronting anxiety, life-altering diagnoses, even death.

In those moments, the relationship between doctor and patient matters. Over half of patients say that they want their doctors to display qualities like compassion and a good bedside manner.

Those qualities aren't merely "nice-to-haves" -- they deliver real clinical value. Studies show that for some conditions, stronger bedside manner and physician empathy are associated with better patient outcomes.

No matter how capable AI tools become, they can't fill physicians' shoes in these moments. So in the coming years, it will be even more important for medical educators to create training programs that enhance empathy among aspiring physicians and teach students how to navigate interactions with patients and their families.

Medical educators will also need to train students to leverage the benefits of AI -- while recognizing its biases and mitigating its shortcomings.

AI systems are trained on human-generated data. So AI models can have the same blind spots and erroneous assumptions as humans. One study found that AI exhibited "significant performance disparities" based on race, gender, and age across nearly 30% of cancer diagnostic tasks.

Future physicians need to learn to incorporate AI into their workflows to make them more effective without blindly accepting the outputs AI systems produce.

That will require changing how physician trainees are mentored. Traditionally, medical students have gone through clinical rotations -- and later, through residency -- to shadow experienced physicians, observe cases, and build a mental library of experiences to draw upon throughout their careers.

But the entire corpus of medical knowledge will soon be on each and every smartphone. AI models can sift through clinical databases and synthesize decades of research in seconds. Three years ago, ChatGPT began outscoring medical students on the United States Medical Licensing Examination. AI models have grown exponentially more capable since then.

Consequently, the trainees of tomorrow should not even try to know more than the machine. Instead, they will need to observe how senior physicians effectively integrate AI into their workflow, identify bias in results, and deliver difficult news to patients.

Artificial intelligence will transform medicine. But it won't eliminate the need for human judgment, compassion, and trust. That's why medical schools must redouble their focus on training students in the skills machines can never replace.

James W. Schroeder, Jr., MD, MBA, is the chief academic officer at Hospital for Special Surgery.