Everyone is talking about AI in health care, but I bring a somewhat unusual perspective to the discussion. In addition to serving as chief medical officer for NextGen Healthcare, I am a practicing family physician in a practice owned by a local hospital. I also own and operate a small practice with my wife, which gives me firsthand experience with the realities of independent medicine. On top of that, I am board-certified in clinical informatics, so I spend much of my time thinking about how technology can support clinical decision-making. Because of that combination of experiences, I have developed a few opinions about AI that may not be universally popular.
Hot take #1: Nearly every clinician will be using an AI scribe within a few years
I believe AI-driven ambient listening documentation will soon become part of everyday workflows for almost every clinician. That statement often generates skepticism. Many physicians have experimented with early-generation AI scribes and concluded they were inaccurate, disruptive, or simply not worth the effort. I understand that reaction. Early versions of any technology often create as much friction as they remove.
But focusing on what AI scribes were two years ago misses the bigger story: the trajectory of the technology. The quality of speech recognition, context awareness, summarization, and workflow integration has improved dramatically, and the industry is rapidly solving many of the problems that once made ambient documentation feel cumbersome or unreliable.
Real physician voices, twice a week
Free, and one click to unsubscribe.
I frequently hear physicians say, “It’s still faster if I just do it myself.” Maybe today, but not for long. I often spend time talking with clinicians who have successfully adopted ambient AI solutions. The most effective implementations are not replacing physician judgment. They are reducing cognitive burden, minimizing keyboard time, and allowing clinicians to focus more attention on patients.
The real question is no longer whether AI documentation works. The question is how quickly it becomes the standard expectation for clinical practice.
Hot take #2: Agentic AI will accelerate the shift to value-based care
For more than a decade, health care has talked about value-based care. Progress has been real, but adoption has often fallen short of expectations.
One of the biggest barriers has not been clinical philosophy. It has been economics. Value-based care depends on activities that happen outside the exam room: care coordination, patient outreach, chronic disease management, preventive interventions, follow-up reminders, and more. Delivering those services consistently requires people, and people are expensive.
Many organizations simply have not been able to scale the staffing necessary to achieve the full promise of value-based care, or they have to hire the staff far in advance of expected value-based payments. This is where agentic AI becomes interesting. When used responsibly, AI agents can help capture structured clinical data during visits, identify care gaps, prioritize outreach, support care managers, and maintain engagement with patients who might otherwise disappear between appointments. In other words, AI can make health care teams more scalable.
If health care leaders are serious about making value-based care sustainable, they should be paying close attention to the role that agentic AI can play in creating operational models that are both clinically effective and financially viable.
Hot take #3: Physicians need to stop fighting patients’ use of AI
Whether physicians like it or not, patients are asking ChatGPT and other generative AI tools questions about symptoms, diagnoses, medications, and treatment options. Too many discussions about patient-facing AI still focus on whether patients should be using these tools. I think that is the wrong question.
Patients are going to use AI. The more productive conversation is how physicians can help patients use it safely. The physician’s role should not be to police AI usage. It should be to evaluate, contextualize, and validate (or correct) the information patients bring into the exam room. In many ways, this is simply an evolution of a challenge clinicians have faced for decades. Before AI, patients arrived with information from search engines, social media, well-meaning friends, and countless other sources. Now, the source just happens to be generative AI.
The future of health care is not AI-free medicine. The future is better medicine because clinicians understand how to assess AI-generated information and partner with patients to separate useful guidance from misinformation.
Hot take #4: AI could make independent practice viable again
Many physicians did not join large health systems because they wanted to. They did so because the administrative burden of operating an independent practice became overwhelming. Could AI reverse some of the consolidation that has transformed medicine over the last two decades?
If AI can automate documentation, improve coding accuracy, streamline follow-up workflows, simplify administrative tasks, and reduce operational overhead, it may create new opportunities for smaller practices to compete. I am not predicting a return to the health care landscape of thirty years ago. But I do believe AI has the potential to change the economic equation that has pushed so many physicians toward consolidation, and maybe make them happier, or at least more professionally satisfied.
Hot take #5: Governance matters more than capability
Health care spends far too much time discussing what AI can do and not nearly enough time discussing how it should be governed. Health care organizations are understandably eager for solutions that reduce burnout and administrative burden. After all, clinicians have been asking for relief for years. But unchecked urgency can create risk.
The next stage of AI adoption cannot simply be about faster outputs. It must be about reliable outputs. Across the industry, we need to be having conversations around governance frameworks, validation standards, transparency, quality assurance, and ongoing performance monitoring. The goal is not to replace clinical judgment. The goal is to create tools that clinicians can trust while preserving the expertise and accountability that medicine demands.
Robert Murry is chief medical officer, NextGen Healthcare. He brings to this position more than 20 years of extensive clinical experience and background in health IT. Previously, Dr. Murry served as the company’s chief medical information officer (CMIO) since May 2017. During his time as CMIO, he was the “voice of the physician” across specialties, product safety, and government/regulatory affairs. Before becoming CMIO, he was the company’s vice president of clinical product management, responsible for clinical oversight and workflow design.
Previously, Dr. Murry served as medical director for ambulatory informatics and CMIO for Hunterdon Medical Center, where he continues to practice family medicine at Hunterdon Family Medicine at Delaware Valley.
He is board-certified in clinical informatics by the American Board of Preventive Medicine and board-certified in family medicine by the American Board of Family Medicine. He is also a fellow of the American Academy of Family Physicians.
Dr. Murry holds an MD from The University of Texas Southwestern Medical Center, Dallas; a PhD in physical chemistry from Boston College; and an MA in physical chemistry from the Massachusetts Institute of Technology.
