The U.S. health care system is adopting artificial intelligence faster than nearly any other sector, and for many patients, that’s unsettling.

Patients consistently say they trust AI far more when it supports doctors rather than replaces them. But the rapid integration of AI in medicine, particularly in clinical settings, has left many wondering: Is rising AI use helping clinicians do their jobs better? Or, as some technology leaders seem to hope, is it setting the stage for a future where AI “superintelligence” takes human experts’ place?

New research into how clinicians are currently using AI offers a reassuring answer, and a roadmap for building health care AI that effectively serves both doctors and patients.

Researchers recently analyzed a representative sample of millions of AI sessions by clinicians using a medical reference tool accessible to over 13 million providers globally that provides vetted information on conditions, medicines, and other topics. The research sought to understand how doctors used the technology in their daily work.

The results show that the platform did help clinicians in their decision-making, but not by replacing doctors’ judgments. Instead, doctors used it to quickly and efficiently gather the information they needed to make informed clinical decisions.

Core clinical decision-making drove the majority of queries. Roughly 40 percent of AI searches were for information about diagnoses, and 18 percent asked for overviews of conditions.

Questions about treatment, medication, and dosing for various conditions made up the next-largest share of searches. After that, the most common query subjects included medical research, adverse effects, and contraindications, factors that make certain treatment approaches dangerous for particular patients.

What’s striking about these usage patterns is how closely they mirror the way physicians already work. Historically, clinicians have relied on tools such as drug reference databases, guidelines, research journals, and calculators to inform their diagnoses and treatment recommendations.

The number of clinical studies has doubled over the past decade, creating a growing “synthesis crisis” for clinicians. They must keep pace with rapidly expanding medical knowledge while also managing higher patient volumes, greater case complexity, and rising administrative demands. Now, AI is enabling doctors to synthesize information and cross-reference multiple sources in seconds, strengthening their ability to make quick, informed decisions for patients.

The most-searched topics by clinicians reinforce this point. The specialties most frequently searched included oncology, diabetes and weight management, and mental health. The medications featured most prominently in searches were diabetes treatments, including glucagon-like peptide-1 (GLP-1) drugs like Ozempic and Wegovy, antibiotics, and hormonal therapies.

That’s not surprising. Those are among the most complex and rapidly evolving areas of medicine, thanks to constantly emerging research and evolving clinical guidelines. Doctors have a lot to gain from targeted AI tools that can collect and synthesize the latest information in these fast-moving fields.

Put simply, AI used well is not a replacement for physicians’ expertise, but a force multiplier for it. And that makes it a vital tool for addressing challenges like the growing shortage of physicians in the United States, as long as its development is guided by a clear understanding of the kind of help doctors need.

Projections show that 30 of 35 medical specialties could face shortages by 2038. AI may not be able to solve this problem by replacing physicians, as some tech leaders might imagine. But it can help by empowering physicians to work more efficiently and access critical information faster, allowing them to focus more time and attention on caring for patients.

It’s inevitable that AI will reshape medicine. The technology is advancing too quickly, and physicians face too many demands for it not to play a growing role in clinical care. But how well that transformation benefits patients will depend on the people building AI tools, and how closely they listen to the clinicians who will actually use them.

The emerging picture of how clinicians use AI offers a glimpse of what happens when doctors aid in the design and development of new AI tools and help ground them in trusted medical sources. If doctors lead in shaping how AI is created, the technology won’t become the autonomous authority patients fear. It will become what it should be: a powerful support that helps physicians deliver faster, smarter, and more effective care.

Neha Pathak is a physician executive.