AI Experience Design: Making AI People Trust
AI experience design decides whether people actually use the AI you built: what it shows them, when it hands control to a person, and how it proves it is right.
A capable model is not the same as a useful product. The gap between the two is almost always the experience: the screens, the handoffs, and the small signals that tell someone whether they can rely on what the AI just said. I design that layer for healthcare and enterprise teams.
Why AI experience design decides adoption
People will not act on AI they do not trust, and trust is what turns a capable model into something a team actually uses. When AI makes a high-stakes call unchecked and gets it wrong, the damage is not one bad output. It is every output afterward, because people quietly go back to the old way.
Good AI experience design prevents that. It keeps people in control where it matters, makes the AI's work easy to check, and lets the routine work run on its own.
Where my approach comes from
My work started in storytelling and UX, where I learned to translate human needs into clear experiences people actually use. At IDEO, that meant designing for people who could not afford for the technology to be wrong.
I was the lead interaction designer on Project BLAID, a shoulder-worn wearable that Toyota announced in 2016 to help blind and low-vision people find their way indoors, in places like malls and office buildings. It recognized landmarks like stairs, doors, restrooms, and escalators, and guided people to them.
On Project BLAID, I designed the navigation language and audio: how the wearable guided people through sound and vibration, with voice and button input. We worked directly with blind users on echolocation and on how much faster they process sound than sighted people, experimented with binaural audio as a guidance source, and built a simplified version that ran on donated iPhones so anyone could use it.
About a year before that, I worked on in-car driver safety explorations for an automaker, using directional speakers so only the driver heard certain cues. Both projects taught me what AI teams are learning now: guidance only works if people trust it in the moment they need it. I still begin with the person on the other end, then decide what the technology should do for them.
I have also taught that way of working. Nielsen Norman Group invited me onto their UX podcast to talk about storytelling and stakeholder buy-in, and I have taught a full-day certification course with them. More background is on the About page.
Four patterns I use in AI experience design
Keep people in control by stakes, not by habit
Keep a person in control for decisions that are high stakes or irreversible, anything regulated, and any case where the model is uncertain. Automate the routine, reversible work where being wrong is cheap. A review step that rubber-stamps is worse than none, so the AI has to surface its reasoning and confidence for the person to decide well. More in Human-in-the-Loop AI.
Show the sources
Teams will not act on outputs they cannot verify. Citations are a usability feature: discrete inline markers, an expandable evidence drawer, and a one-click jump to the source. Prioritize the references that matter and keep the rest a click away, so trust does not turn into overload. More in The Citation Pattern.
Put friction where it belongs
In sensitive workflows, removing every bit of friction increases risk. Better systems slow people down at exactly the right decision points and nowhere else. More in Friction Is a Feature.
Design past the chatbox
The most valuable AI in your product will not look like a chat window. It will look like work that finishes itself, inside structured screens built for the task. More in AI Experience Design Beyond the Chatbox.
AI experience design in healthcare and enterprise
These patterns matter most where the stakes are highest. My most recent work has been in healthcare and commerce: patient experience, self scheduling, and clinical workflow design in healthcare, and memberships, seller onboarding, and merchant tools at Walmart and Shopify. In enterprise settings the same principles apply to workflows where a wrong answer costs money, time, or a customer's confidence.
In both, the goal is the same: informed judgment at speed. The AI does the heavy lifting, the person makes the call that matters, and everyone can see why.
Frequently asked questions
What is AI experience design?
AI experience design is the design of how people interact with AI inside a product: what the AI shows, how it explains itself, when it hands control to a person, and how it earns enough trust that people use it.
How is AI experience design different from UX design?
It builds on UX design, but AI does not behave the same way every time. AI experience design adds patterns for uncertainty, sources and citations, human review at high-stakes moments, and recovery when the model gets something wrong.
How do you get people to trust AI?
Show the work. People trust AI they can check, so the experience should surface reasoning, confidence, and sources, and keep a person in control for decisions that are high stakes, regulated, or uncertain.
Does adding human review slow AI down?
Only if the review step is designed badly. Done well, the AI surfaces its reasoning so the person decides quickly and with good information. A small cost at the right points buys the trust that makes people use the system at all.