AI Travel Planning: A Great Co-Pilot, a Terrible Captain
|
|
Tempo di lettura 4 min
Here's a confession that might surprise you: some of our clients found us through AI. They asked a chatbot where to swim with sperm whales or ski untouched powder, and our name came up. So this won't be a rant about how AI travel planning is ruining travel: we believe AI travel planning is genuinely good at some things.
But lately the internet is full of posts promising that the right prompts will turn ChatGPT into a travel agent. That's where we draw a line, because there's a difference between planning a trip and pulling one off. AI travel planning is a brilliant co-pilot. It makes a terrible captain.
What AI travel planning is genuinely good at
Let's be fair, because the tools have earned it. For the early, messy stage of a trip, the dreaming, AI travel planning is fantastic. Ask it to brainstorm destinations for a free two weeks in February and it'll surface options you hadn't considered. Ask it to draft a rough day-by-day, summarise visa rules, or compare three regions, and you'll have a useful first sketch in seconds.
This is real value. Many now treat AI travel planning as another member of the team, handing it the admin nobody enjoys so the humans can focus on what matters. Used this way, as a research assistant, an idea generator, a first draft, AI travel planning saves time and genuinely helps.
The trouble starts when people mistake that first draft for the finished trip.
AI travel planning is a great co-pilot. Just don't let it fly the plane.
Where AI travel planning falls apart
AI travel planning is confidently, plausibly wrong far more often than you'd think, and it never knows when it's guessing.
The examples are real, and occasionally alarming. AI travel planning tools have invented entire towns: travellers in Peru were sent to a place that doesn't exist, complete with fabricated historical sites, and ended up stranded in remote terrain with no infrastructure. Others have been pointed toward a fictional Eiffel Tower replica in Beijing. On the milder end, these tools routinely recommend restaurants that closed years ago, quote museum hours that changed after the pandemic, and describe bus routes that no longer run.
None of this is a rare glitch. One 2025 analysis found that general chatbots invented venues 15–20% of the time; even dedicated travel planners still got it wrong 3–5% of the time. And a peer-reviewed 2026 study reached a conclusion worth sitting with: eliminating these hallucinations is, for now, impossible; and worse, people tend to follow the fabricated itineraries anyway, because a confident, personalised-looking plan simply feels trustworthy.
That's the real trap of AI travel planning. It doesn't fail obviously. It fails convincingly.
Information vs Connections
But even when the information is flawless, AI travel planning hits a wall it cannot climb: and this is the part that matters most.
AI travel planning works only from what has already been published. It scrapes the open internet: blogs, reviews, listicles... the same widely available information everyone else can see. That's information. What it doesn't have, and can't have, is connections.
We've run expeditions in these places. We know which local operator actually delivers and which one looks polished online but cuts corners in person. We know the skipper to call, the permit to pull, the guide whose network gets you where the crowds never reach. We know that last week the whales moved forty kilometres north, because someone we trust was in the water and told us. No prompt gets you that, because none of it was ever written down. It lives in relationships, built over years, on the ground.
AI travel planning can find you information. It cannot introduce you to people. And on a real expedition to a remote corner of the world, the people are the trip.
The 11pm test
There's one more thing, and it's the simplest of all. This is where AI travel planning runs out of road entirely. When something goes wrong (and on adventurous travel, occasionally it does) you cannot call a chatbot.
Your transfer doesn't show at 11pm in a country whose language you don't speak. A flight cancels and unravels a week of tightly connected logistics. The weather turns and the whole plan has to be rebuilt on the fly. AI travel planning handed you a document; it can't pick up the phone, it can't rebook your bush flight, and it can't reassure you at midnight. A human who has actually been there can, and will.
It's telling that demand for human travel specialists is rising, not falling, in the middle of the AI boom. In Britain, around a third of travellers still book through a person. In the US, the main professional body's membership has jumped from 190,000 to over 310,000 in two years. People are voting with their trips: and they're voting for humans, precisely because the tools got so good at the easy part.
Use AI to dream. Use people to go.
So here's our honest take. Fire up ChatGPT. Brainstorm, compare, sketch the rough shape of your next adventure... as a starting point, AI travel planning is a wonderful place to dream.
Then, when you're ready to actually go somewhere wild, somewhere the itinerary can't be scraped off a blog, where the magic depends on who you know and who has your back at midnight, talk to people who have been there.
AI travel planning is a great co-pilot. Just don't let it fly the plane.
