Dr. Amar Rewari on Spectrum News+: Where AI Belongs in Your Health Care

Dr. Amar Rewari joined host Jenna Enoch on Spectrum News+ to discuss one of the biggest questions in health care today: where artificial intelligence fits in the doctor’s office, and where human judgment needs to take over.

How doctors are using AI today

Dr. Rewari explained that AI is already part of many physician workflows. It helps reduce administrative burden by assisting with patient documentation, and it helps doctors keep up with medical research that is growing at a rapid pace.

As a radiation oncologist, Dr. Rewari described how AI has changed treatment planning in his own field. Physicians traditionally had to outline normal tissue structures by hand on each individual CT slice so radiation could avoid them. AI now automates much of that contouring, which saves time in building radiation plans and helps patients start treatment sooner.

From “Dr. Google” to “Dr. Chat”

Many patients now type their symptoms into ChatGPT and other chatbots before seeing a doctor. Dr. Rewari said this curiosity can be a good thing when it helps patients understand their medications or the side effects of therapy. He cautioned against using chatbots for self-diagnosis, because an algorithm may not account for all of an individual patient’s risk factors. Diagnosis, he said, should stay between patients and their physicians.

The problem of bias

Dr. Rewari also raised concerns about bias in AI systems. Underrepresented and marginalized communities are often missing from the data these algorithms are trained on, which means results for patients in those groups may be less accurate, especially when it comes to diagnosis and assessing risk.

A tool, not a replacement

Looking ahead, Dr. Rewari acknowledged that physicians disagree about AI’s future role. His view is that AI is a valuable tool for augmenting what doctors do, but it isn’t ready to replace clinical judgment. There should always be a doctor in the loop to review AI-generated output and confirm that a diagnosis is accurate.