How is AI used in healthcare?

Where is AI already used in my care?

The World Health Organization groups health uses of large AI models into 5 areas. Most of what you will run into fits one of them.

You may not see most of it. An AI tool might help analyze one of your scans, draft the note from your visit, or help write the reply you get in your patient portal.

• Diagnosis and clinical care: such as responding to patients' written questions.
• Patient-guided use: looking into symptoms and treatment options yourself.
• Clerical and admin work: documenting and summarizing visits in the health record.
• Medical and nursing education: simulated patient visits for trainees.
• Research and drug development: including finding new compounds.

Which area of medicine uses the most AI?

The FDA keeps a public list of AI-enabled medical devices that are authorized for sale in the US. As of its September 4, 2026 update, the list had 1,614 entries. 1,230 of them, about 3 in 4, had radiology as the lead review panel. Cardiovascular came next with 154, then neurology with 73.

Radiology covers imaging, like X-rays, CT, MRI and ultrasound. To be on the list, a device has met the FDA's premarket requirements, which include a review of its safety and effectiveness for its intended use.

2 caveats from the FDA itself: the list is not a complete record of every AI device, and new authorizations can take time to show up. The FDA also says it will explore ways to tag devices that use large language models (LLMs), the technology behind chatbots, in a future update. The list does not do that yet.

Is AI listening to my doctor's visit?

It might be. Ambient AI scribes listen to the visit between you and your clinician, then draft the visit note. In a 2025 study, clinicians edited the draft before it went into their note.

Use is growing fast. In the AMA's 2026 survey of nearly 1,700 physicians, 81% said they use AI at work, more than double the 2023 rate. 28% used it to document billing codes, charts or visit notes, and 19% used it to draft replies to patient portal messages.

The evidence so far is about doctors more than patients. In a 2025 study in JAMA Network Open of 263 clinicians at 6 US health systems, burnout fell from 51.9% to 38.8% after 30 days with an AI scribe, and clinicians reported focusing more on the patient in the room. The study had no control group, relied on self-reports, and the scribe vendor helped collect the data.

In that study, clinicians asked patients for verbal consent before recording. If you are not sure whether your visit is being recorded, ask, and tell them if you would rather it not be.

What about AI chatbots and AI doctors I can use myself?

This is the newest area: tools you text or chat with to make sense of symptoms, lab results or a new prescription. Pymander is one example, an AI doctor you text for guidance on what to do next.

WHO warns that these models can produce false, inaccurate, biased or incomplete answers, and that people can fall into automation bias, trusting the output without catching mistakes. Models can also be trained on poor-quality or biased data, for example biased by race, ethnicity, sex, gender identity or age.

A few habits help you use them safely:

• Give full context: your age, medicines, conditions and how long symptoms have lasted.
• Treat the answer as guidance, not a final diagnosis. Decisions about prescriptions belong with a licensed clinician.
• Check how the tool handles your data before you share anything sensitive.
• If the answer and your gut disagree, or you are getting worse, get seen.

How does AI help with drugs and drug safety?

Drug makers now use AI at many stages of a medicine's life, from lab work and clinical trials to manufacturing and tracking problems after a drug is on the market. The FDA says it saw more than 500 drug submissions with AI components from 2016 to 2023, and in January 2026 it and the European Medicines Agency issued 10 guiding principles for using AI in drug development.

For drug safety, the FDA says early adopters use AI to automate routine steps like taking in, entering and processing side effect reports, and that it may make safety surveillance more efficient. The FDA runs a program to learn how companies use these tools and what the risks are.

What does all this mean for me as a patient?

Much of the AI in your care works in the background, and doctors want guardrails on it. In the AMA survey, 88% of physicians said strong safety and effectiveness testing is critical for wider AI use, 86% said data privacy is critical, and 85% wanted a say in how AI is adopted.

You have a right to ask questions. Is AI being used in my visit? Who reviews what it writes? Where does my data go? If an AI-drafted note in your record looks wrong, you can ask to have it corrected.