Travellers moved faster than the industry
Phocuswright found that 56% of US leisure travellers used AI for at least one trip in 2026, up from 43% nine months earlier. They called it the fastest behavioural shift the travel industry has seen in more than a decade. Skift Research puts familiarity with AI trip-planning tools at 62%, with usage up 124% year over year.
The businesses are behind. Mindtrip, working with Sabre, Sojern and MMGY Travel Intelligence, surveyed 1,006 American travellers and 103 destination marketing leaders in April. Among travellers, 54% had already used AI to plan a trip and 75% intend to use it again. Among the organisations selling those destinations, 22% offer AI trip planning on their own site.
So the tools work and people use them. Then look at what happened to trust over the same months.
Gallup surveyed 1,572 Americans aged 14 to 29 in February and March. The share who feel hopeful about AI fell to 18%, down from 27% a year earlier. Half of them use it weekly or daily anyway. Booking.com found that 91% of travellers hold concerns about AI, and 35% fully trust what it gives back.
Daily use and falling faith, in the same people, at the same time.
The part everyone gets wrong
The comfortable story is that young travellers can spot AI content and reject it.
The Yale Youth Poll tested that this spring with 3,429 registered voters, 2,008 of them aged 18 to 34. That younger group was confident it could tell a real photograph from a generated one. Then, as The Nation reported, they were shown two pairs of headshots, one pair from stock photography and one generated from a single-sentence prompt. 80% of them got it right about half the time. A coin flip.
They cannot see it. They suspect it.
That is a different thing, and it is harder to work with, because suspicion needs no evidence. It attaches to everything in view, including the parts that are real.
What that does to a list
A list of ten places used to be worth something. Someone had to go, or read, or ask around, and then write it down.
Now anyone can produce one in four seconds, for free, in any language, and it will be decent. When something becomes free, the value moves somewhere else. In travel it moved to the question that comes before the list: who are you when you travel, and where do you actually belong?
That question has never survived a search bar, and a prompt does not get you closer. You cannot type your instinct into a box, and most people cannot say it out loud either. Ask someone what kind of traveller they are and you get the version they perform, the one that sounds good at a dinner party.
Their swipes say something else.
How the reading works
You swipe on mood images. No questionnaire, no profile to fill in.
What counts is what you choose and how fast you choose it. Instinct arrives first, in under a second, and reasoning spends the rest of the time building a respectable explanation for the choice you already made. Those choices are scored against real places, and three destinations come back.
That reading is a calculation. No generative AI touches it, and nothing in your result is written by an AI model.
The mood images are a mix of photographs and images we generated ourselves. What each one has to do is precise, and generating is sometimes the only way to get it exactly right. Given what the Yale numbers say about telling the difference, the honest move is to say so rather than to claim otherwise.
Where AI does the work is on this side of the screen. It refines our code, polishes our copy, pressure-tests our arguments, and pushes back when a claim does not hold up. Technical wingman, editorial sparring partner.
This piece is the example. The line about Gen Z seeing through AI imagery sat in our own brand guidelines until the Yale data proved otherwise, so we cut it.
About the marks
None of this arrived with the current wave. In July 2024 I wrote a vision document for PIM, the Dutch platform for marketing innovation, with AI and sustainability running through it. It leaned on a Gartner forecast that by 2026, 80% of creative talent would use generative AI daily. That year is now.
One of its four pillars was ethics: AI in marketing only earns trust when the standards around it are strict enough to protect the people on the other end.
Gartner also expects that by 2027, 20% of brands will position themselves on the absence of AI in their product. That forecast is from 2024.
Two years later that pillar is law.
Since 2 August, the European Commission requires providers of generative AI serving the EU to mark what their systems produce in a machine-readable way, under Article 50 of the AI Act. Google DeepMind has been watermarking text from the Gemini app since May 2024. Anthropic started doing the same for Claude this month. The mark travels through copy and paste, and a full rewrite removes it.
Which makes the honest version of this note straightforward: parts of this text carry a mark, and they should.
We are a small team trying to read instinct in a world of dropdown menus. The AI tooling gets better every month and we will keep using it wherever it makes U-GO sharper. The day it starts assisting the swipe itself, you will read it here first.
Until then, the swipe stays yours.
People also ask
Does U-GO use AI to choose my destinations?
No. Your top-3 is a calculation based on what you swiped and how fast you swiped it, scored against real places. No generative AI touches that result and nothing in it is written by an AI model. AI helps build the product, not read you.
Are the images in the swipe AI generated?
Some of them are. The swipe is a mix of photographs and images we generated, because what decides is whether an image does exactly what the reading needs it to do. Your result is a separate matter: that is a calculation, and no AI model writes it.
Why do AI travel planners suggest the same places to everyone?
Because a prompt can only work with what you can put into words, and most people describe the traveller they perform rather than the one they are. A search bar has the same limit. Your instinct never makes it into the box, so what comes back is a good list for a generic person.