Washington, D.C., is home to one of the country’s more ambitious efforts to bring artificial intelligence into medicine responsibly. The MedStar-Georgetown Collaborative Center for Artificial Intelligence in Healthcare Research and Education, known as AI CoLab, pairs a major regional health system with a leading academic medical center to research, test, and teach AI applications in clinical care.
Its work spans applied research, training the next generation of clinicians and data scientists, industry partnerships, and convening policymakers, researchers, and innovators around a shared question: how to make the technology serve patients rather than the other way around.
Dr. Nawar M. Shara is chief of research data science for the MedStar Health Research Institute, co-director of the Center of Biostatistics, Informatics, and Data Science (CBIDS), and chief of AI applications in health data science for AI CoLab.
Dr. William J. “Billy” Mea is senior director of research at PsycHealth, LLC, a firm focused on AI-supported healthcare and research, and a member of the AI CoLab research team. At Georgetown, he teaches policy courses grounded in innovation, change management, program evaluation, and leadership, and mentors both graduate and undergraduate students.
Together they bring data science and clinical psychology to the question now confronting every hospital and medical school in the country: how to use a powerful new tool without hollowing out the profession it is meant to strengthen.
The part no algorithm captures
Shara does not treat AI as infallible. It makes mistakes, she warned. It generates artifacts and can occasionally produce output “that might kill your patient.” The safeguard is a physician who still knows how to question the answer.
Some knowledge never makes it into the data at all. “There is a dimension that we can never capture, that you only get when you’re sitting right there with your patient,” she said. That quality lives outside textbooks and numbers. She described it as something close to magical, understood only human to human.
Preserving it is the work she and her colleagues have taken on. They are equipping the faculty charged with training future physicians, a calling she described as sacred, to bring AI into the curriculum in ways that protect critical thinking and the drive to become a good doctor. Her forthcoming book, AI in Medical Education, will be published by Springer next year.
Signs of de-skilling
Mea worries the erosion has already begun. Data he gathered in Bangladesh show medical students consulting AI before their textbooks, and some skip buying the books entirely because a chatbot delivers an acceptable answer. Overextended attendings aren’t always watching closely. Closer to home, a physician assistant told him that an attending had grown concerned that fellows were losing their capacity for critical thinking and leaning too heavily on AI.
The challenge reaches well beyond medicine. Because AI can write a paper, Mea noted, high schools and colleges have had to rethink grading. Many are returning to in-class quizzes and handwritten exams, and the blue book is making a comeback.
Lynn Schmidt7 min readOne ask of Congress
Asked what single bill, regulation, or funding decision he would request from a member of Congress, Mea proposed an accountability framework: “a national healthcare AI accountability program that ties Federal support to independent safety testing, auditable records of consequential AI actions, and meaningful clinician authority to question and override recommendations—without allowing vendors or hospitals to shift the entire responsibility to the clinician.”
That final clause carries the weight. Without it, the clinician becomes the default bearer of blame for systems they did not design and cannot fully inspect. The independent judgment Mea fears is eroding would then become the only protection patients have left.
Shara’s request pointed in a different direction. She would fund legislation that lets AI deliberately overhaul the nation’s food system as an industry, in service of health and healthy living.
Winning the future
Every mtf.news conversation closes with the same question: What’s one idea that would help America win the future?
Mea came prepared. America wins, he said, by building greater human capacity rather than human replacement. Technology naturally tends to replace people. The better use of AI is to absorb low-value tasks so that clinicians can strengthen their judgment, widen access to care, and spend more time with patients. He recalled his years on the inpatient traumatic brain injury unit at Walter Reed, where roughly two-thirds of his time went to the computer, toggling between one system for labs, another for notes, and still another for orders. “That kind of irritation should be a thing of the past,” he said. “We need more time with patients.”
Shara offered what she called a non-traditional answer, drawn from more than two decades in the field. AI, she argued, can shorten the learning curve enough to shift medicine from curing disease toward preventing it. She pointed to a paradox. A nation with extraordinary brainpower carries one of the sickest populations, while people in remote, deprived regions often post better metabolic and blood-work numbers. “There is something fundamentally wrong in our society,” she said, faulting a culture that chases balance sheets over human lives. She named food, lifestyle, and isolation as the targets: “Loneliness is killing most of us.” She acknowledged that many people won’t like hearing it, but said she was speaking her conscience. In her words, AI is “a huge blessing from our own doing,” the culmination of everything humanity knows. The task now is to use it to back-engineer the life Americans want to live.
Two answers, one conviction
Shara and Mea approach the question from different directions, but they arrive at the same place: the technology is only as good as the human purpose behind it. For Mea, that means giving clinicians back their time and their judgment. For Shara, it means aiming AI upstream, at the conditions that make people sick long before they ever become patients.
