Your Patient Brought an AI Diagnosis. Now What?

From "Dr. Google" to AI, help patients turn more information into a better conversation.

Long before AI, there was "Dr. Google." Patients brought printouts and search results to appointments. Now they bring chatbot conversations. The habit is familiar: people want to understand what's happening to them before they walk through your door.

We live in an information-rich world, and that's an opportunity. Patients who have done some reading can ask better questions and have more meaningful conversations about their care. But the same search can send someone down a rabbit hole, worrying about a diagnosis that doesn't fit their symptoms.

The job is to help patients make sense of what they've found. Welcome the effort, find out what prompted the search, and bring the conversation back to the concern that brought them in. You don't have to review every result or agree with every conclusion to take someone seriously.

"What were you hoping this would help you figure out?" is a useful opening. If something alarmed them, ask, "What part worried you most?"

The answer gives the clinician somewhere to start. Perhaps the patient wants to know whether a spot is dangerous. An acne patient might be more worried about the side effects they read about than whether the prescription will work. Those are different conversations, even if both begin with the same app.

In Edelman's 2026 research with Yale, respondents said being understood and having concerns addressed without judgment would help them consider a recommendation they doubted.

"Don't believe everything you read" can sound like a verdict on the person who read it. Thank them for bringing the question, then ask to see the part they want to discuss. They shouldn't have to defend their curiosity before getting help.

If their research raises something useful, say so. If it leads in the wrong direction, explain what doesn't fit and return to their concern. Respecting the effort doesn't mean accepting the diagnosis.

A reassuring answer needs more than "I'm not concerned." Explain which findings matter and how they connect to the recommendation.

A useful structure is: "I understand why that caught your attention. Here's what I'm seeing, here's how it fits with your history, and here's what I recommend we do next." The clinician supplies the specifics, including what the screenshot leaves out. A change over time, prior treatment, or something found on examination may change the assessment.

When a patient asks for a test or treatment, discuss what it would help answer, whether it fits their situation, and what the alternatives involve. Ordering something to settle an argument sets a poor standard. So does dismissing the request without explaining the decision.

If the concern involves medication, ask whether it changed what the patient actually did. As we covered last week, writing a prescription doesn't tell you whether treatment began.

End with a plan the patient can repeat back: what happens next, when to follow up, and what changes should prompt earlier contact. If uncertainty remains, explain how it will be addressed.

An owner or administrator can improve this experience before the clinician enters the room.

Agree on a welcoming front-desk response: "Please bring the part you'd like to discuss. I'll note the question for your visit." Staff don't need to decide whether the screenshot is medically credible. Their job is to capture the concern and route clinical questions through the practice's usual process.

For portal messages, ask patients to identify the question they want answered and the relevant excerpt. A long attachment shouldn't obscure a report of new or worsening symptoms. Tell patients when to expect a response and how to reach the practice for urgent concerns, using your established triage guidance.

Also be clear about where a substantial discussion belongs. A clinician can decide when a message needs an appointment. An automatic reply shouldn't promise that someone will review every page before tomorrow's visit.

Keep screenshots in the practice's approved communication channels. Don't paste identifiable patient information into a public chatbot to check its answer. The AMA's patient guidance makes a useful starting point for education: protect privacy, and don't rely on AI for diagnosis, treatment decisions, or emergencies.

At your next huddle, have one clinician and your administrator agree on the opening question and the route for questions arriving between visits. Test it for 2 weeks.

Ask the team to bring back a few de-identified examples: where the wording helped, where a patient still didn't know what to do, and where the same question reached several people. Adjust the response together. You don't need a new dashboard or a tally of patients who mention AI.

Put the agreed wording where staff already work, such as an existing portal reply template. The administrator can set the review date and make sure the afternoon team hears about it, too.

An informed patient can be a partner in a better conversation. Whether the starting point is Google, a chatbot, or something a friend sent, help them sort out what applies and leave knowing what to do next.

Clarity RCM manages revenue cycle for 200+ dermatology practices across 44 states. It's all we do. See how we work.