By Bonnie Darves, a freelance health care writer

While some people in medicine may be debating whether it’s artificial intelligence, augmented intelligence, or assisted intelligence, at this point, it hardly matters what it’s called: AI has fully arrived, is being applied across the health care spectrum, and is rapidly altering the landscape of physician practice.

In the past few years, AI has made its way into physicians’ workflow in numerous ways that promise to automate time-consuming tasks and relieve some of the administrative hassles that physicians find particularly burdensome, both manually and cognitively. Examples include use of medical scribe technology and ambient listening during patient visits, chart-summary generation, billing-code documentation, and creation of discharge instructions, care plans, and progress notes. Some physicians are turning to AI algorithms to generate portal notes to patients and provide evidence-based disease and self-care recommendations.

Physicians are also increasingly turning to AI tools to obtain medical research summaries and to tap into assistive diagnosis by quickly accessing up-to-date research. At the more sophisticated end of the spectrum, physicians are using AI to predict patients’ health risks, identify potential quality gaps, and predict their treatment outcomes.

At the practice or health care organization level, physicians who struggle with the perennially burdensome business of obtaining insurers’ prior authorization for patients’ treatment are excited about emerging AI applications that may soon relieve them of that task.

AI is poised to alleviate the burden on physicians, particularly in light of the ongoing physician shortage, the vast amounts of information being generated, and the increasing patient panels. The basic capabilities of AI — synthesizing research, analyzing enormous amounts of data quickly, flagging potential risks, and automating workflows — are already transforming clinical practice.

In all these cases, the promise of AI is to reduce administrative burdens so that physicians can spend more face time with their patients and, possibly, reduce burnout, according to Sarah Nosal, MD, who is president of the American Academy of Family Physicians and vice president for innovation & optimization at The Institute for Family Health, a network of federally qualified health centers in New York City.

“What we’ve seen, with clinicians using these AI tools, is that their [professional] satisfaction is improving. They feel better when they go home — that their work feels more sustainable and that they’re less likely to quit their jobs,” said Dr. Nosal, who is board-certified in family medicine and clinical informatics and serves as the Institute’s chief medical information officer.

Dr. Nosal cites a specific example of the burden-reduction benefits of AI. When physicians have background listening technology in place during a patient visit via AI scribes, with the patient’s approval, AI is “writing the chart note with them and documenting what’s happening. And that changes the cognitive burden that the physician takes home,” she said. Physicians report that they’re less concerned about whether they captured a particular patient-reported health issue, for example, or put in the correct orders, she said.

Even though the time-saving benefits of AI tools in daily practice are modest at present, Dr. Nosal reports, they are accruing quickly. Physicians who have used the tools are optimistic about their potential for streamlining and enhancing care and reducing their cognitive load.

So far, AI gets mostly high marks from physicians

Daniel Yang, MD, Kaiser Permanente’s vice president of AI and emerging technologies, recently helped lead the massive rollout of Kaiser’s AI-enabled clinical documentation system. He frequently cites a humanizing benefit of the tool: freeing physicians from their keyboards and computer screens so that they can look at patients in the exam room. “Not only does the tool drive provider satisfaction, it’s also driving patient satisfaction. Patients are finding our physicians more attentive and transparent,” Dr. Yang wrote in a recent LinkedIn post.

For Paulius Mui, MD, a Massachusetts primary care physician and informaticist, AI tools are delivering big benefits daily. “I am able to do more for patients when it comes to navigating the health care system because I can be more efficient getting them to the right places. I’m also able to automate tasks that would have required hiring people,” said Dr. Mui, who operates a direct primary care practice and frequently delivers presentations to physicians on AI applications.

“The quality work that I do now would not have been accessible to me as a solo practitioner before I started using AI. I can take advantage of having resources that large institutions have,” said Dr. Mui, who specializes in health care quality and safety with a focus on technology governance and is chief executive officer of the quality assurance company Distillemr.

Even though research on the extent to which AI is changing medical practice is nascent, it’s expanding rapidly. More important, perhaps, research is demonstrating that physicians are excited about implementing AI in their practice lives, and many are using the tools when they’re available.

An annual survey conducted by the American Medical Association (AMA) found that 80 percent of physicians currently use AI in a professional context, double the share who reported using it in 2023. Following are other findings from the AMA’s 2026 Physician Survey on Augmented Intelligence:

• More than 75 percent of physician respondents report that AI provides an advantage in their ability to care for patients, up from 65 percent in 2023. Topping the list of benefits are improved diagnostic ability and greater work efficiency.
• Seven in ten respondents indicated that they see opportunities for AI to automate the clinical and administrative tasks that contribute to work stress and burnout.
• Physician training in AI is occurring, but only 11 percent of respondents reported receiving “a lot of” training, and 92 percent expressed interest in receiving more training.
• Physicians, not surprisingly, want to be consulted on AI decisions in their organizations (55 percent), and 30 percent want a role in its implementation.

Despite physicians’ positive views on AI benefits, 41 percent of respondents in the AMA survey cited some degree of concern about the technology’s potential impact on patient privacy, and 88 percent cited the possibility that AI usage might result in some degree of clinical skills loss.

Physicians’ fears about potential skills loss as AI tools proliferate — what some are calling “de-skilling” — are valid concerns, observes endocrinologist Graham McMahon, MD, MMSc, who is president and CEO at the Accreditation Council for Continuing Medical Education. In his June 23, 2026, JAMA article, Dr. McMahon referred to de-skilling as an “insidious concern.” If physicians increasingly “rely on AI systems to generate answers,” he wrote, “diagnostic reasoning may be practiced less and independent synthesis of evidence may weaken.” The key, in his view, is to ensure that AI tools augment rather than replace physician reasoning.

Jennifer Phan, MD, an Irvine, California, gastroenterologist who is active in the American Gastroenterological Association Center for GI Innovation and Technology, acknowledges that AI usage has prompted worries about physician skill-set effects. But she thinks that the tools’ positives, for the future, are more compelling. “With AI, there are concerns being cited about the potential ‘de-skilling’ of gastroenterologists. I think that the focus should instead be on using AI to help with mental fatigue and to improve efficacy and patient safety.” AI applications are already transforming colon polyp detection, Dr. Phan said, enabling gastroenterologists to move beyond identification to include risk assessment and resection recommendations.

Neurosurgeon Peter Nakaji, MD, FACS, founder of Scottsdale Neurosurgery Specialists in Arizona, notes that AI-driven developments in brain surgery are already essentially giving surgeons the equivalent of “second pair of eyes,” enabling real-time analysis of the operative field. “You just can’t watch all of the inputs all the time, so we’re always flipping our attention back and forth in order to do that,” Dr. Nakaji reported in a recent issue of the ACS Bulletin. “I think AI is going to actually provide you much better information than you can get from your own mind because you can’t possibly focus on everything at once.”

Of the multitudinous AI use cases emerging in health care, Dr. Mui cites the nascent developments in AI-enabled safety reporting and continuous improvement as one of the most promising. “This is a very exciting use case because it literally benefits everyone,” said Dr. Mui. “Authors of recent research have identified opportunities in surfacing learnings across different safety events, improved characterization of diagnostic errors, and better dissemination of these insights,” he notes, across local and regional systems.

Preparing to practice in an AI-enabled world

Health care, and physician practice by extension, have undergone significant shifts over the last two decades, but there is agreement that the increasing incorporation of AI may represent the most significant and far-reaching change the health care sector has witnessed.

Understandably, while AI-fluent and technology-adept physicians may welcome the new frontier and are anxious to explore its capabilities in patient care improvement, others are reluctant to dive in, for various reasons. Physicians who are currently behind the curve — either because they’ve been reticent to embrace the technology or because they’ve been too busy focusing on clinical training — may feel so overwhelmed that they’re not sure how to gain AI proficiency.

“What I see are two cohorts — busy residents who are focused on clinical learning and are less engaged with technology, and curious self-directed residents who might be ahead of faculty in AI technology adoption,” Dr. Mui said. Among mid-career and older physicians, AI proficiency runs the gamut, he acknowledged. Some are early adopters, while others have remained on the sidelines.

For physicians in training who haven’t had a lot of AI exposure or engagement because of either their environment or time constraints, Dr. Mui recommends starting with hands-on experience rather than theoretical knowledge. “Residents should start by trying out the different AI tools, learning from their colleagues, and participating in events like hackathons at medical conferences to build practical experience,” said Dr. Mui, who frequently gives presentations on AI applications in practice and its benefits in enhancing electronic health record (EHR) capabilities.

Sources suggested that physicians turn to their professional organizations for AI training options such as webinars and formal courses. For example, the AMA offers the ChangeMedEd^® Artificial Intelligence in Health Care Series and the American College of Preventive Medicine offers comprehensive online learning courses. In addition, major academic centers such as Harvard, MIT, Stanford, and the University of Michigan offer workshops and CME courses. The recently founded American Board of Artificial Intelligence in Medicine offers certification pathways at both the introductory and advanced levels.

Anesthesiologist and clinical informaticist Samir Kendale, MD, who directs the Harvard Medical School continuing medical education course “AI in Clinical Medicine,” recommends that physicians who want to boost their AI fluency and capabilities seek guidance from the individual or team that oversees informatics in their organization. In a recent edition of Harvard’s Insights publication, Dr. Kendale stressed that while it’s important for physicians to have a basic understanding of AI literature and products to determine which tools are suitable for their needs, they also need to understand the organization’s systems architecture before applying AI tools.

Regardless of where physicians are positioned on the AI-fluency spectrum, Dr. Nosal urges them to focus on five important requirements before they begin using AI tools in direct patient care:

— Ensure a high degree of electronic health record (EHR) competency and fluency. This may sound like a given, but physician competency with EHRs varies considerably, Dr. Nosal observes, and as AI becomes increasingly integrated with the EHR, competency is a must.

— Obtain basic AI literacy, to gain an understanding of the terminology, tools, and functions.

— Ensure the ability to identify key possible failures, fallacies, or inadequacies of AI-generated output. Physicians should be able to detect inaccuracies, a commonly reported issue and concern with AI tools, and to spot the so-called “hallucinations” that occur in the output.

— Be able to explain to patients, at a third-grade level, how AI is being used (or might be used) in the clinical setting. Physicians should also ensure patient consent to AI use in the patient-encounter setting.

— Know the AI policies of the organization where you practice, and be prepared to adhere to them.

“These five things are what any physician who is new to having AI in their life or is a new residency graduate who is going to work really wants to make sure they’re aware of before they hit the ground running,” Dr. Nosal said.