Healthcare AI can monitor patients continuously, reach out proactively and personalize care based on a patient's clinical history. Yet most patients still don't trust it to do any of those things. While the technology is capable, adoption of AI is stalling because many healthcare organizations are not yet ready to deploy it in ways patients trust. 

The barriers are both organizational (data infrastructure, systems integration) and human. Even when AI has access to a patient's clinical history, interactions often feel generic rather than personal, damaging trust at the organizational level, not just the individual one. When patients encounter healthcare AI today, it's typically through a chatbot or symptom checker with a disclaimer at the top and a path to a human agent if the system falls short. Disclaimers and escalation paths are the right last mile. The problem is when organizations treat them as the whole journey. What closes the gap on patient trust is an architecture that works today and also scales into the anticipatory, personalized, ambient experiences that healthcare AI can bring.

Give Patients What They Want

Our research into the future of AI interfaces surveyed experienced AI users across five scenarios, including healthcare, and found that 89% want a deeply personal AI partnership that knows their context, preferences and history from session to session. In healthcare specifically, that expectation connects to a long-standing tension the sector has never fully resolved. Patients want greater control over their health data. The sustained engagement on platforms like Epic’s MyChart suggests that patients will invest in digital health tools when they feel genuine ownership over what's shared and how it's used. In healthcare AI, data control is a trust signal, not just a compliance requirement.

What makes healthcare distinct from the other contexts in our research is a specific combination of preferences. In personal finance, travel and career development, users want to initiate. In health, they step back: biometric data was a top-selected trigger, while the instinct to type a prompt hit its lowest point across all five scenarios. Patients want the AI to notice and respond rather than waiting to be asked. They were also least willing to hand over control, with the highest proportion of any scenario wanting humans to make the final call. In practice, patients want AI that surfaces meaningful insights while leaving decisions in their hands. Build for that, and you've built for where anticipatory healthcare AI is heading. For healthcare organizations, these findings translate into concrete design decisions.

The Trust Architecture for AI in Healthcare

Our research points to three user experience decisions that shape how organizations can close the adoption gap today while building for what's coming.

1. Transparency ahead of action. In the healthcare scenario, 50% of participants selected "Announce plan" as a priority, meaning they want to know what the AI intends to do before it does it. Nearly as many selected "Show thought process," suggesting patients want to understand not just what the AI plans, but how it arrived there. In a clinical context, that means being clear about what data informed a recommendation and where the limits of that recommendation are.
   
   
2. Human-at-the-lever oversight. Unlike traditional human-in-the-loop designs where oversight happens at defined checkpoints, human-at-the-lever means patients hold control throughout. Our research showed the most distributed agency preferences of any scenario in healthcare: patients want to define what the AI can access upfront through explicit consent and data governance, monitor as the system executes, and retain a clear ability to pause or override at any point.
   
   
3. Personalization as the foundation. Generic AI in healthcare erodes trust, because a patient managing a chronic condition with a complex medication history won't trust a system that treats them as a first-time user. Meaningful personalization means understanding clinical history, communication preferences, and cultural and linguistic context. These features give patients genuine control over what the system knows and how it uses that information. The more patients feel ownership over the relationship, the less trust has to be rebuilt through disclaimers after the fact. 

Broken Trust Doesn't Come Back Easily

In most consumer categories, trust in AI can be rebuilt through repeated positive interactions. The stakes of a poor healthcare experience are correspondingly higher. When a patient encounters a wrong answer, a recommendation that misses their history, or an interaction that feels generic at a moment when it shouldn't, the damage extends well beyond that individual experience.

A poll our company ran found that 88% of patients have already seen AI make a mistake, and asking an AI "Are you sure?" doesn't reliably surface a better answer. These experiences chip away at patient confidence. By the time patients begin pulling away, rebuilding trust becomes far more difficult. It just means the trust gap is getting wider before they have a chance to address it.

What to Ask Your Healthcare AI Implementation Partner

The right implementation partner makes the difference between a healthcare AI deployment that earns patient trust and one that has to rebuild it. When evaluating partners, ask:

●      Does your partner understand the healthcare-specific trust requirements around clinical accuracy, data privacy, patient data ownership and regulatory compliance?

●      Does your partner design transparency and oversight mechanisms that work in today's interfaces and scale into ambient, voice and agentic environments?

●      Does your partner demonstrate how human-at-the-lever principles are built into the system's architecture, not layered on afterward as a UI fix?

What your partner builds now should still be working for your organization and its patients as healthcare AI evolves.

The organizations positioned to deliver the next generation of healthcare AI aren't treating today's interface and tomorrow's as separate investments. The human-centered UX decisions that build patient confidence now are the same ones that power what's coming.

Bobby Brown is Vice President of Healthcare at TELUS Digital, where he leads the organization's healthcare growth, transformation, and go-to-market strategy across health plans, pharmacy benefit managers (PBMs), providers, retail pharmacy, and digital health organizations. With more than 35 years of experience in healthcare, customer experience, and business process outsourcing, he specializes in driving large-scale operational transformation, digital innovation, and strategic partnerships for some of the industry's most complex organizations.