We Changed. The UI Didn't.
The shift to generative UI isn't a technology story. It's a human behaviour story.
A few weeks ago I needed a hotel. I didn’t open a browser, didn’t go to Booking.com, didn’t do any of the things I’d normally do. I opened ChatGPT, typed what I wanted, city, dates, rough vibe, and got back options with images, prices, and a direct link to book. Maybe ninety seconds total.
I didn’t think much of it at first. Then I did.
Because I’ve spent over a decade designing software. I know how much work goes into a product experience, the search filters, the comparison views, the photo carousels, the reviews, the map. Someone built all of that. And I skipped every single part of it without even noticing.
That’s when it landed. The interface didn’t fail me. I just didn’t need it anymore.
Here’s what I think we’re getting wrong about generative UI: we keep treating it as a technology story.
What can AI render? How do we design for unpredictable outputs? What happens to our design systems? These are real questions but they’re all responses to something that already happened, quietly, before most product teams noticed.
Human behaviour changed. Not because of AI. AI just made the new behaviour easier to act on.
I notice it in myself and the people around me. We used to search. Open a product, get oriented, use the filters, click through pages, make sense of whatever structure the design team decided on. Now we describe what we want and expect something to figure it out. We shifted from command-and-respond to express-and-receive, and we did it naturally, without anyone asking us to.
We never actually wanted to navigate. We wanted the answer. Navigation was just the tax we paid to get there. And now that the tax has been reduced, almost nobody is choosing to pay it voluntarily.
The question worth asking isn’t “what does generative UI change?” It’s “what does it tell us about what people always wanted?”
I kept thinking about what that ninety second booking actually meant for the products I wasn’t opening.
If a meaningful chunk of hotel bookings in the next few years starts with a ChatGPT prompt rather than opening Airbnb, what does Airbnb do?
This isn’t hypothetical anymore. Expedia and Booking.com already built apps that plug inventory directly into ChatGPT. Accor launched inside ChatGPT in January. The distribution channel already shifted. And the reason it shifted isn’t because the technology was impressive. It’s because users were ready for it immediately, because the behaviour was already there.
Airbnb’s CEO has said a general AI can give you a list of homes but can’t verify identities or handle global payments and trust infrastructure. Fair. But that’s an infrastructure answer. It doesn’t address the earlier problem, which is that the moment someone decides where to start looking is moving away from branded destinations and toward a prompt box. Not because AI told them to. Because it finally matches how they wanted to think about the problem.
Did your users already change, and are you the last to know?
The products that win won’t be the ones with the best UI. They’ll be the ones that understood what users actually wanted to accomplish and built around that, not around the interface they assumed users were willing to tolerate.
That’s always been good design. What’s changed is that users now have options. They don’t have to tolerate the interface anymore. And that tells us something important not about AI, but about how much patience people always had, and how little of it was ever really there.