What patients really think about AI in their healthcare

A men's health clinic called us in July, a little rattled. They had just turned on an after hours virtual assistant to catch the calls their front desk kept missing, and the owner could not sleep. His exact words: "What if a patient realizes it is not a person and feels tricked? These guys already do not love opening up. I do not want a robot to be the reason they walk." Reasonable fear. It is the same fear we hear from dermatologists, dentists, and med spas every month. Everyone assumes patients are one AI discovery away from storming out.

Then, in late August, Pew Research Center published one of the largest looks yet at how Americans actually feel about this. The answer is more useful than the fear. Patients are not anti AI. They are anti surprise. And most practices are managing the wrong risk entirely.

The data: openness with one big condition

The new report, Americans want transparency when AI is used in their healthcare, comes from a survey of 3,488 U.S. adults run June 22 to 28, 2026. Start with the headline most people miss: Americans are more optimistic than not about where this is going. In Pew's earlier tracking, 44 percent expected AI to have a positive effect on medical care over the next 20 years, against just 19 percent who expected a negative one. That is not a public that hates the technology. That is a public leaning cautiously toward yes.

Here is the condition. A full 72 percent said it is extremely or very important that a doctor or provider tells them when AI is being used in their care. Another 16 percent called it somewhat important. Add those up and the message is almost unanimous: use the tools if they help, but tell me. Notice what patients are not asking for. They are not asking you to ban AI. They are asking you not to hide it.

That 46 percent is the number to sit with. Nearly half the country has no idea whether AI has already touched their care, and among the people who know it has, only 22 percent feel they understand how it was used. This is the actual problem. Not resistance. Fog. Patients are standing in a room they cannot fully see, and human beings do not trust what they cannot see. The practice that turns on the lights wins the room.

Patients draw a very clear line

The survey also settles an argument we have a lot with owners: does the type of AI matter? It matters enormously. Patients sort AI into two buckets, and they feel completely differently about each.

• AI near a medical decision. When AI helps analyze a scan or reach a diagnosis, 81 percent want to be told. For explaining lab results, 80 percent. This is the high stakes bucket, where patients want their doctor's judgment front and center and AI clearly in the passenger seat.
• AI near the admin work. For taking appointment notes, 72 percent want disclosure. For prescription refills, 64 percent. For scheduling an appointment, it drops to 56 percent. Still a majority, but far more relaxed. Nobody thinks a smart scheduler is playing doctor.

Read those two buckets together and the psychology is obvious. The discomfort was never really about AI. It is about control. Pew found 63 percent of adults want more of a say in how AI is used in their care, and 53 percent feel they have little or none today. Patients are fine with AI helping their doctor. They get nervous when it feels like AI is replacing their doctor. For most of what a practice actually automates, the phones, the texts, the reminders, the booking, you are squarely in the relaxed bucket. You are not asking a patient to trust an algorithm with a biopsy. You are asking them to trust it to book a Tuesday at 3.

Where AI already touches your patients first

Owners picture the scary version of this, an AI reading an MRI, and skip the version that is actually live in their office. For most practices, the first and often only place AI meets a patient is the front desk and the phone. That is the bucket patients are most relaxed about, and the one with the biggest payoff, because missed calls are lost patients. When a call goes to voicemail, most people do not leave one. They call the next name on Google.

So the real question is not whether to use AI there. It is whether you use it honestly and well. A virtual receptionist that picks up on the first ring at 9pm, answers the three questions every caller asks, and books the appointment is a service upgrade, full stop, as long as the patient is told what it is and can reach a person. If you are weighing this, we walked through the trust side of it in whether patients hang up on an AI receptionist and the disclosure question in whether to tell patients when AI answers the phone. You can hear the honest version in action with our AI receptionist, which introduces itself plainly and hands off to your team when a caller needs a human.

The same rule covers everything else you might automate. Appointment reminders, review replies, first response to a web lead. Each one is fine, even welcome, when the patient can tell what is happening and never feels cornered by a machine. The tools are not the risk. Silence is.

Uncertainty is an opening, not a threat

Step back and look at that 46 percent again, the near half of patients who do not know where AI already sits in their care. Most practices read that as a warning. We read it as an opening. When a market is confused, the clear communicator wins by default. You do not have to be the biggest or the most advanced. You have to be the one who explains, in plain words, what your tools do and where a human takes over.

This is also why we tell owners to sell the outcome, not the acronym. Patients do not want AI. They want to book at 9pm, get a text back in ten minutes, and reach a real person when something feels off. Lead with those promises and mention AI plainly as the how. The moment you turn AI into a buzzword on your homepage, you sound like every vendor. The moment you explain what it does for the patient, you sound like the practice that respects them. We made the fuller case in whether you should market AI to patients and in how to use AI without losing patient trust.

Our honest take

The men's health owner from July did one thing right and one thing wrong. Right: he cared enough to worry about how it would land, which is more than a lot of practices bothering to bolt on AI can say. Wrong: he almost hid it out of fear, which is the one move the data says actually backfires. We changed his assistant's greeting to name itself in the first breath and promise a human, added a line on the booking page, and told his front desk exactly how to describe it if a patient asked. The revolt never came. What came was more booked calls after hours and, weirdly, a few compliments about how easy it was to reach the office.

That is the whole lesson in one clinic. Patients are not the obstacle you think they are. They handed you the instructions in this survey: use the tools that help, tell us plainly, and keep a human within reach. Do that and AI stops being a liability you hope nobody notices and becomes a reason patients pick you. Try to sneak it past them and you will forfeit the exact trust you were trying to protect.

How EtherealMinds handles this for practices

When we build a practice's patient acquisition system, the AI is never a secret and never a gimmick. The virtual receptionist introduces itself honestly and always offers a live handoff. Booking pages and reminders say in plain English what is automated and where your team steps in. And we write the marketing around the benefit the patient feels, answered fast, booked easily, a person when it counts, not around a technology they never asked to think about. That same honesty runs through the websites we build and the social media we manage, because trust is not a feature you add at the end. It is the whole product in healthcare.

So here is the thing to do this week. Make a short list of every place AI already touches a patient in your practice, the phones, the texts, the forms, the reminders. Then ask one question of each: would a patient be surprised to learn this? Anywhere the answer is yes, add a sentence. That is the entire fix, and most of your competitors will never get around to it.