Patients are already asking AI their health questions. Patient education is how health systems stay the trusted source for answers.
It’s 9 p.m. the night before a colonoscopy. The prep instructions are on a printout somewhere in the house, and the patient has no other way to look them up. They aren’t sure when to stop clear liquids. The nurse line closed at 5.
So, they open a chatbot.
Twenty years ago, clinicians worried about “Dr. Google.” Today, patients ask AI chatbots. The behavior is familiar. The answers now arrive faster and sound more personal. According to a recent Gallup poll, roughly a quarter of U.S. adults use AI chatbots for health information or advice.1 While concerning, the good news is that 73% still turn to their doctor with questions.2
Boards are asking what the health system’s AI plan is. A more useful question may be where patient trust starts. Health systems want to be the first place patients turn with a health question, including through the branded AI tools the system offers. Call it the AI front door.
The goal is a worthy one. But a front door only works if people already come to it. Patients need to rely on their health system for trusted answers before they will use its AI tools. Patient education is where health systems can build that habit today.
Trust has to exist first
Patients trust their care team more than any app. The open question is whether that trust carries over to a health system’s own digital channels. Evidence on system-led, AI-enabled interactions is still early, but hospitals need to meet patients where they are.
Habit is how trust gets there. Gmail was easy to launch because people already used Google Search. A health system can’t introduce a new AI experience to patients who have never come to it for valuable information. Today’s patients are like consumers. They want easy access to clear answers on their own time between office visits.
Education is infrastructure, not a single moment
The colonoscopy patient who couldn’t find the prep instructions points to a bigger need. Printed materials still work well for many patients, and many prefer them. But when paper is the only option, a misplaced page leaves nowhere else to look. Education also isn’t a single moment. Patients need it from scheduling through recovery at home.
And, every patient learns differently so the information needs to be multi-modal, in their own language and reading level. Meeting that range takes a connected approach.
Clinicians already carry a heavy load: too many clicks, too much documentation, and the effort of deciding what matters most to tell each patient. Ask a nurse how many times this week they’ve explained the same procedure prep or the same discharge warning signs. Each time, they also must make sure the patient understood.
With workforce still the top concern for health system executives in 2026, and more than 90% prioritizing productivity improvements,3 that is skilled time and mental energy spent on repetition. A connected approach takes on the routine explanations, so clinicians can focus on the conversations that need their judgment.
The results from personalized education show up in operations. One Virginia health system automated its colonoscopy prep education and reduced prep-related same-day cancellations by 50%, an estimated $100,000 or more in annual savings. Each canceled colonoscopy costs an estimated $700 to $1,000.4
Cancellations, readmissions and patient experience scores all depend on whether education reached the patient and was consumed and understood.
Where does your health system sit?
Health systems can’t skip from manual education to an AI front door. The middle stages build the habit and data AI depends on. Four stages describe where most health systems are today and where AI readiness begins.
| Stage | What it looks like | AI readiness |
| 1. Print only | Printed materials, shared at a single moment. No digital option. | No digital relationship with the patient. |
| 2. Digitized | Content in the portal, assigned manually, one patient at a time. | Occasional logins. No habit yet. |
| 3. Automated | Education triggered by the EHR from a procedure, diagnosis or order, and timed across the journey. | Repeat portal use. Habit forming. Measured time spent in the portal consuming education. |
| 4. Connected | Education adapts to each patient’s language, literacy and format. Engagement data flows back to care teams. | Trust, habit and engagement data that future AI tools can build on. |
Many health systems land between stages two and three. They’ve digitized, but education still reaches patients one at a time. Moving from two to three is the hardest jump, and the one that matters most. It means education is triggered automatically, from the procedure or the order, and arrives when the patient needs it across the care continuum.
Stage four is about the individual patient. Personalization is more than translating a video. It is the right language, reading level, and format, at the right moment, working together.
Education has to be designed for the full range of patients, not an average one. Hawaii Pacific Health, for example, built culturally tailored NICU education for Chuukese- and Marshallese-speaking families. This kind of personalized approach not only helps patients and their caregivers understand their care, but it can also help enhance their confidence. Education works best when it reflects the people receiving it, and when care teams can see how patients use it.
Once engagement data flows back, knowing what patients watched or read and when becomes an input for better care and, eventually, for AI. Engagement data suggests patients respond when education is relevant to their procedure and easy to reach. For the first six months of 2026, 49% of Medicare patients having bypass surgery (CABG) engaged with their assigned education, along with 44% of hip replacement patients and 37% of knee replacement patients. Knee replacement patients watched 15 minutes or more, with reminders.
A quick self-assessment
- If a patient is scheduled for surgery tomorrow, does the right education reach them automatically?
- Is education timed across the journey, from scheduling through recovery at home?
- Does it adapt to the patient’s language, reading level and preferred format?
- Do you know what patients watched and when?
- Does that data flow back to the care team?
If the answer to the first question is no, start there.
Governing what carries your name
Patients still see their care team as the source of truth. That’s a real asset, and it has to be earned again with every digital interaction. Connecting patients to a health system’s AI agent with human oversight is safer and more trustworthy than relying on standalone public AI chatbots.
Trust also supports safety. When a patient already uses your portal and trusts what they find there, you can route them between self-service information and a clinician the moment something looks urgent. Without that relationship, there’s nothing to route.
The responsibility grows when AI carries your brand. A health system’s tool also faces different regulatory expectations than consumer AI. Any patient-facing tool that handles protected health information adds security risk, so governance needs to cover access controls, data handling and vendor risk, alongside clinical review.
Governance for patient-facing AI can include:
- Test the tool against high-risk scenarios before launch.
- Have clinicians review a sample of conversations every day.
- Use an AI reviewer to check every conversation.
- Report regularly to an AI governance board.
Boards can add two questions: How are adverse events detected? Can patients report a problem?
None of it works without good content underneath. AI is only as trustworthy as the clinical content it draws from, which is why evidence-based, expert-reviewed content is the foundation.
Start with the algorithm present
The health systems patients trust five years from now will be the ones that treated education as infrastructure today, without waiting to see how AI develops. Those that wait aren’t only behind on AI. They are already carrying the cost, in cancellations, repeated staff time and lost trust.
To win the AI future, invest in the algorithm present. Build the foundation AI depends on now.
Before you explore introducing patients to an AI tool, find where your health system sits on the curve. If education still reaches patients one at a time, or in one moment, start there.
The patient with a prep question at 9 p.m. should find the answer in a place their health system built.
Discover how you can transform the patient experience through integrated technology and personalized education. Learn more: elsevier.com/patient-engagement
Sources
- Gallup, Americans Turning to AI to Supplement Healthcare Visits, April 14, 2026, https://news.gallup.com/poll/707789/americans-turning-supplement-healthcare-visits.aspx
- Gallup, Four Ways Americans Seek Medical Advice, February 12, 2026, https://news.gallup.com/poll/702164/americans-rely-mainly-own-doctor-medical-advice.aspx
- Deloitte, 2026 Global Health Care Outlook, December 11, 2025, https://www.deloitte.com/us/en/insights/industry/health-care/life-sciences-and-health-care-industry-outlooks/2026-global-health-care-outlook.html
- Internal data from a Mytonomy by Elsevier customer, January to June 2026.
- Fierce Healthcare, Patients prefer healthcare providers’ AI agents to public chatbots, with human oversight non-negotiable, survey finds, June 24, 2026, https://www.fiercehealthcare.com/ai-and-machine-learning/patients-prefer-healthcare-providers-ai-agents-public-chatbots-and-human
