How AI Is Changing Patient Trust in Healthcare

I was at the doctor's office two weeks ago. This sign was on the wall. "Patients who have already obtained a diagnosis via ChatGPT are kindly asked to seek their second opinion not from us, but from Gemini." I laughed. Then I stopped laughing. Because this is not just a joke. It is a symptom. We have entered an era where patients arrive with AI generated differential diagnoses before the physician has said a word. Where the consultation starts with "but ChatGPT said..." instead of "here is what I am experiencing." The doctor's sarcasm is understandable. Years of medical training. Decades of clinical pattern recognition. Now competing with a language model that sounds confident but has no liability, no license, and no understanding of the patient sitting in front of it. But here is the uncomfortable flip side. Why are patients turning to AI first? Because appointment wait times are weeks. Because consultations last seven minutes. Because the system often fails to explain, contextualize, or listen. AI is not replacing trust. It is filling a vacuum where trust used to be. The sign is funny. The underlying dynamic is not. The question for healthcare is not "how do we stop patients from using AI?" It is "why did they feel they needed to?" #Healthcare #AI #PatientExperience

Something in healthcare AI has been bothering me—and two posts this morning helped me name it. 🔹 One from Spencer Dorn MD, who warns that AI isn’t just making doctors more efficient—it’s turning us into quantified workers. Every action tracked. Every decision scored. A creeping loss of professional judgment, masked as “optimization.” 🔹 The other from Graham Walker, MD Walker who points out something equally dangerous: AI doesn’t need to be accurate—it just needs to sound like it is. A confident hallucination can earn trust. Especially in healthcare, where sounding like a doctor often works better than being one. At first, these seem like different critiques: One about surveillance, the other about persuasion. But together, they reveal a bigger shift: AI is slowly changing who we trust in medicine—and why. Not long ago, trust came from relationships. You knew your physician. They knew you. Trust was earned over time, through presence, context, and accountability. Now? Patients are being coached to trust the interface. Doctors are being scored by the dashboard. And both are being trained to believe: if it’s fluent, fast, and confident—it must be right. We’re drifting from relational medicine to performative medicine. Where appearing reliable replaces actually being responsible. That’s what ties Spencer’s “quantified doctor” to Graham’s “confident AI.” In both cases, the human gets flattened—either into a metric, or a voice that can be mimicked. This isn’t a Luddite argument. AI will help medicine in powerful ways. But we can’t ignore what’s being hollowed out in the process. Because once trust migrates from people to systems, it doesn’t just change the work. It changes the soul of the profession. Full posts linked below, definitely recommend the read! #HealthcareonLinkedin #healthcare #AI #AIinHealthcare

We just hit 1 million AI-patient interactions. The data is way different than what the "experts" predicted. Here's what we actually learned: 10 Truths After 1M AI-Patient Conversations: 1. The Transparency Paradox 62% want to know it's AI. But reveal it upfront? Engagement craters. Solution: Build trust first. Reveal identity second. 2. Text is Logic, Voice is Emotion Voice-only: 7% response. Text-first: 60%+. Patients need to RE-READ instructions but HEAR empathy. Both matter. 3. The "Technogenerian" Reality 70+ demographic? Our most active users. Why? AI gives them what modern healthcare won't: TIME. Judgment-free attention. 4. AI as Behavioral Scaffold Chronic care = 80% lifestyle change. AI wins as an accountability partner, not doctor replacement. "Doctor's orders" → "My progress" = everything. 5. Rolling with Resistance Push patients? They retreat. Roll with resistance using motivational interviewing? They engage. Acceptance beats persuasion. 6. The Mirroring Effect "I notice you feel lower on Tuesdays." No judgment. Just neutral pattern recognition. Self-awareness unlocked. 7. Radical Capability Disclosure "I can track vitals. I can't diagnose." Brutal honesty = trust spikes. Clear boundaries eliminate the uncanny valley. 8. Data Reciprocity (WIIFM Factor) Patients don't want portals. They want VALUE. Give something useful immediately—engagement explodes. 9. Low Health Literacy is the Silent Barrier 50%+ struggle with medical jargon. AI wins by translating to kitchen table language. Game changer. 10. The Loneliness Bridge Massive insight: Tons of interactions are just… social. AI solves isolation FIRST. Then clinical adherence follows naturally. The Human-AI Paradox: Make AI more human → patients demand transparency it's NOT. Disclose it's AI → it must prove it HAS human qualities. 1M interactions taught us technology doesn't replace humanity. It scaffolds it. What surprises you most? 👇

Patients care most about how well medical AI performs, more than FDA approval, brand-name certification, or even having a doctor in the room. 1️⃣ A conjoint survey of 3,000 US adults tested what drives patient trust in and preference for AI-assisted medical encounters. 2️⃣ AI performing better than a specialist had the largest effect, increasing the probability of choosing that visit by 32.5%. 3️⃣ AI performing at specialist level still boosted visit preference by 24.8%, far outpacing any governance mechanism. 4️⃣ Having a clinician present increased preference by 18.4%, roughly equal to AI performing at the level of a general practitioner (19.1%). 5️⃣ FDA approval and Mayo Clinic certification each increased preference by 11.1%, with no difference between the two. 6️⃣ Local hospital certification had a smaller effect (7.8%), despite being the governance layer most relevant to real-world performance variation. 7️⃣ Disclosing that AI was trained on representative US population data boosted preference by 11.9%, but labeling data as biased had no effect compared to providing no data information at all. 8️⃣ Open-ended responses confirmed the pattern: 25.7% of respondents mentioned AI performance, 22.7% mentioned clinician presence. 9️⃣ Women showed the same preference patterns as men but reported lower overall trust in AI-involved visits. 🔟 The scenario was limited to a moderate-risk dermatology diagnosis (rash), so findings may not generalize to higher-stakes clinical decisions. ✍🏻 Ana Bracic, Kayte Spector-Bagdady, JD, MBe, Sophie Towle, Rina Zhang, Cornelius A. James, W. Nicholson Price II. Factors for Patient Trust and Acceptance of Medical Artificial Intelligence. JAMA Network Open. 2026. DOI: 10.1001/jamanetworkopen.2026.0815

🛎️ NEW RESEARCH 🛎️ AI in health is here – but public trust, organizational governance, and policy isn’t keeping pace. Last year, I published a commentary (https://lnkd.in/ezYebyEz) on how states were converging and diverging on Health AI requirements. Building on that work, I’ve been lucky to lead new research, funded by the California Health Care Foundation, delivered by NORC at the University of Chicago to 1500 patients across the US, and supported by our Policy Workgroup, and prior workshops of 150+ clinicians, patients, and vendors. Our findings reveal a stark gap between AI adoption and confidence: 💊 75% of people are using AI, yet only 13% feel very comfortable with it 💊 51% say AI makes them trust healthcare less; only 12% say it increases trust 💊 93% report at least one concern about AI in healthcare 💊 80%+ say trust would increase with clear accountability measures Some surprising themes emerged: 💻 Concerns focus less on AI itself, more on governance, accountability, and patient protection. 💻 Use, comfort, and trust are only modestly linked and highly context-dependent. 💻 Data commercialization concerns outweigh concerns about algorithmic bias. Notably 12% report never having considered AI bias at all.  💻 Patients fear ‘humans-out-the-loop’ scenarios– critical in the current Agentic AI debate 💻 Disclosure that AI is involved in care is crucial, but does not alone build trust; in some cases, it reduces trust. 💻 No single institution is seen as a trusted overseer. Instead, people favor multi-layered governance across independent non-profits, health systems, and federal regulators. As states race to regulate health AI and health systems scale adoption, this research offers a foundation for evidence-based governance grounded in real public behaviour and preferences. Huge thanks to our partners and participants for making this work possible.  We hope these findings inform industry leaders, including policymakers, deployers of health AI, those responsible for AI adoption and governance, and more. Coalition for Health AI (CHAI) is committed to gathering consensus and developing tools for patients and providers that set a gold standard for transparency. (And as ever, thank you to Ann Li who made this look so beautiful) David Blumenthal Karandeep Singh Suchi Saria Ramin Bastani Syed Mohiuddin Brian Anderson, MD Randall Rutta Grace Cordovano, PhD, BCPA Lily Liu Daniel Yang, MD Lauren Kahre, MPH Amy Zolotow

Patients love health answers from AI... until they know who wrote them. In one study, patients rated ChatGPT’s answers as more empathetic and higher quality than doctors’. But in other studies, as soon as patients knew the advice came from a chatbot, their trust dropped—even when the content was identical. This mismatch is showing up across healthcare research. 📄 In a 2023 JAMA Internal Medicine study, patients preferred AI-generated responses to doctors’ answers on empathy and quality—when the AI wasn’t labeled as such. 📄 But in a 2024 study from Nature Medicine, when participants were told that health advice came from a chatbot, they rated it as less trustworthy than the same advice from a human. 📄 Another study found that people were less likely to follow medical guidance when it was attributed to an AI—even when the advice was medically accurate. Why the shift? Because we don’t just judge messages by what they say. We judge them by who we believe is speaking—and whether we think they care. So when do patients prefer AI? 🤖 For routine tasks, clear answers, and low-stakes questions 🤖 When the advice feels nonjudgmental and easy to understand When do they still want a human? 🤝 For emotionally charged issues 🤝 When the situation calls for warmth, nuance, or reassurance 🤝 And especially when they know a chatbot is behind the screen Of course, this could change in a year - in fact, it's changing right now. But the constant is trust. Whether it’s delivered by a human or a machine, patients need to feel safe, understood, and respected.

💡 Your patients are already using AI to research their health. The question is whether you're ready for that. Gallup just released a fascinating new survey — out today — that should be required reading for anyone building or buying healthcare solutions right now. 📊 One in four Americans has used AI for health information or advice. 59% use it to research before a doctor visit. 56% use it after. The AI-informed patient is not a future trend. It's the current reality. ⚠️ But here's the tension: only 33% of recent AI health users say they trust the information they're getting. Only 4% strongly trust it. And yet an estimated 14 million adults say AI advice led them to skip a provider visit entirely in the past 30 days. People are making real healthcare decisions on information they don't fully trust — often because cost, access, or feeling dismissed by a provider left them with no better option. 🤔 That last point is the one I keep coming back to. 21% of AI health users turned to AI because they felt dismissed or ignored by a provider in the past. AI isn't just filling an information gap. In many cases it's filling a trust gap that the healthcare system created. 🔮 Trust in AI-generated health information will grow. The models are getting better fast, and as they do, the 33% who trust it today will become a much larger number. That's not a guess — it's the same curve we've seen with every consumer technology that started with skepticism and ended up embedded in daily life. 🏥 What this means for care delivery organizations is significant. The AI-powered consumer is changing the nature of the first interaction. Patients are arriving more informed, more opinionated, and sometimes more confused than ever. The healthcare companies that will win are the ones that design for that reality — that meet the AI-informed patient where they are, build on what they've already learned, and earn trust rather than assuming it. The Gallup article is linked in the comments. Worth reading in full. 👇 I'm curious how others in healthcare are thinking about this shift — are you changing how you engage patients or members in response to AI-driven self-research? And for those of you using AI tools for your own health questions, what's your experience been? Would love to hear both perspectives in the comments. #DigitalHealth #AIinHealthcare #HealthcareInnovation #PatientExperience #HealthTech

AI is racing ahead. Public trust isn’t keeping up. New global data from Pew Research Center paints a clear picture: people know about AI, but many are uneasy—and who they trust to regulate it varies widely. What stands out • Awareness is high, but not deep: a third have heard “a lot” about AI, nearly half “a little,” and only a small share nothing at all. Awareness strongly correlates with national income. • Concern outweighs excitement: roughly one in three are more concerned than excited, while only one in six are mostly excited. Older adults and women are more likely to express concern. • Trust to regulate AI is local first: most people trust their own country, followed by the EU, the U.S., and then China. • Generational gap: younger adults are more aware and more optimistic; older adults tend to be more cautious. Heavy internet users are also more positive. Why this matters for healthcare • Adoption hinges on trust. If patients and clinicians are uneasy, even the best AI won’t scale. Regulatory trust favors the EU and domestic systems in many markets, so aligning with clear, transparent guardrails is a strategic advantage—not a checkbox. • Equity risk is real. Lower awareness tracks with lower income and less education. If we don’t invest in AI literacy and explainability, we risk widening digital health disparities right when we promise precision care for all. • Design for the skeptic. Older adults and many women report higher concern. Build workflows and communication that demonstrate safety, reliability, and accountability at the point of care—not just in a white paper. • Localize governance. Health systems that operationalize responsible AI with clear data policies, bias monitoring, and audit trails will win trust faster than those waiting for “global consensus.” My take Trust is now a clinical feature. Organizations that treat governance, transparency, and AI literacy as core product capabilities will move from pilots to impact. Those that don’t will stall in the “concern” zone, no matter how advanced the model. At GE HealthCare, our Responsible AI Principles ensure trust in every solution: Safety, Validity & Reliability, Security & Resiliency, Accountability & Transparency, Explainability & Interpretability, Privacy-Enhanced, and Fairness with Bias Managed. Enjoy the Sunday-Read. Your move How are you measuring and building trust in your AI-enabled care pathways—among clinicians, patients, and regulators? Which single change would boost confidence the most in your organization?

There’s a quiet shift happening in healthcare, and it’s happening before a patient ever talks to a doctor – millions of times a day. AI is no longer just supporting care. It’s becoming the new digital front door. For many patients, the care journey now starts with a question – and more often than not, that question is being answered by AI. In fact, a KFF study released this week found about 1 in 3 adults turn to AI chatbots for physical or mental health advice: rivaling social media as an information source. That means AI is already shaping decisions for patients about where they go and what they do next, before a clinician is ever involved. That’s worth paying closer attention to. Not because AI isn’t a net positive for healthcare. It is. It has real potential to simplify complexity and help people navigate care more easily. But it also introduces a different kind of risk – overly confident answers to nuanced questions in moments where clinical judgment matters most. We’re beginning to see how this plays out: situations where serious conditions are under-triaged. Others where patients are directed to higher levels of care than necessary. Guidance that works well in straightforward cases, but becomes less reliable in more complex, real-world scenarios. The conversation can’t just be about where AI should play a role in care. It already does. The more important question is what happens next. Where should AI guide and where should it step aside? How do we ensure there is a clear and clinically sound path from digital interaction to provider-led care? Because in healthcare, good AI isn’t defined by how much it can handle independently. It’s defined by how well it supports the right decisions at the right time. The future isn’t AI or clinicians. It’s both – working together with clear guardrails, thoughtful escalation, and accountability built into how care begins. As organizations continue to invest in new technology, the opportunity isn’t just improving access. It’s designing a safer, more consistent path to care. That’s where trust is built – and where the next phase of healthcare will be defined. #AIinHealthcare #virtualcare #digitalhealth #clinicalquality #healthcareinnovation

🗣️ In my new STAT op-ed as a Public Voices Fellow on technology in the public interest with The OpEd Project supported by MacArthur Foundation, I examine how the AI industry's push into the health care sector is deepening a trust crisis: https://lnkd.in/eiZzPq74 ⬇️ The pandemic response was associated with a steep 30-point decline in trust in U.S. health care providers and hospitals. That erosion is continuing as health systems move quickly to adopt AI tools that remain largely untested in real-world clinical settings and with little to no input from patients and communities. 🫀 This moment builds on longstanding mistrust rooted in medical and systemic racism experienced by Black communities and other communities of color. 👩🏿⚕️ Even at the clinical level, the effects are already visible. A LinkedIn colleague who is a physician recently shared that the use of ambient AI during patient visits led some individuals to hesitate before disclosing sensitive information. When patients can't share important information with their health care providers, that carries consequences for care, communication, and outcomes. 🗣️ In the op-ed, I call on health systems to move at the speed of trust, a concept advanced by adrienne maree brown, rather than the speed of investment. 🌱 This approach centers community in decisions about whether, when, and how AI is integrated into care. Trustworthiness is cultivated through participation, transparency, and accountability. RWJF Ford Foundation Doris Duke Foundation Lumina Foundation Kapor Foundation Mellon Foundation Omidyar Network The David and Lucile Packard Foundation Siegel Family Endowment