When Vladislav Guzey planned a trip to Taiwan, he turned to AI — not just for itinerary planning, but for visa advice and more.
asserted
he → plan → advice
He had no reason to doubt it — he used AI all the time.
asserted
he → have → AI
So when AI told him that Russian passport holders didn’t need a visa to enter the country, he believed his bot.
Until he turned up at the airport and had to turn back around.
“I relied on that information,” Guzey told The Post.
asserted
Guzey → tell → information
The AI summary, he later realized, provided outdated search results — turning stale information into a confident answer with real consequences.
And he’s hardly alone.
asserted
he → realize → consequences
AI knows our darkest secrets.
asserted
AI → know → secrets
The embarrassing questions we should know the answer to.
asserted
we → know → answer
So, it’s inevitable that travelers would outsource the tedious act of itinerary planning to chatbots.
asserted
travelers → ’ → chatbots
But many are discovering that their confident answers aren’t necessarily correct.
asserted
answers → discover → ?
An alarming number, 43%, of AI-generated itineraries contained at least one verifiable problem, according to a recent study, and almost three-quarters of U.S. travelers (74%) have reportedly turned to AI during the planning stage of a trip, seeking hotel recommendations, what to do, and where to shop.
uncertain
quarters → generate → what
Half of travelers even use it while traveling for weather and flight updates.
asserted
Half → use → updates
But as more travelers turn to AI to map out vacations, travel advisors told The Post that outright mistakes are only part of the problem.
asserted
mistakes → turn → problem
It can struggle with personal preferences, pacing, and details that turn a list of recommendations into a truly memorable travel experience.
asserted
that → struggle → experience
ChatGPT can produce an itinerary in seconds, but it often lacks any individuality.
asserted
it → produce → individuality
That’s helping drive travelers back toward something a chatbot still can’t replicate: human judgment and consideration — a lesson Robyn Levin learned the hard way.
asserted
Levin → help → something
Levin thought she had found exactly what she was looking for when planning a complicated European vacation: a newly renovated hotel on the Baltic coast that ChatGPT assured her had been updated in 2024.
asserted
ChatGPT → think → 2024
Exactly what she had asked for when planning her trip.
asserted
she → ask → trip
Except the hotel hadn’t been renovated since 2008.
asserted
hotel → renovate → 2008
“Chat apologized, as it frequently does, and said, ‘I should have been more careful with the research,’” Levin told The Post.
asserted
Levin → apologize → Post
The six-continent traveler and luxury travel author had turned to AI for help planning, because why not?
asserted
traveler → turn → planning
ChatGPT suggested a hotel that hadn’t been renovated and a travel route that didn’t exist.
uncertain
that → suggest → hotel
Optimistic, Levin tried again and asked for a luxury Lake Como hotel near a train station.
asserted
Levin → try → station
ChatGPT enthusiastically recommended one it said was just a two-minute walk away.
asserted
it → recommend → one
To reach the hotel, Levin would actually need to take a ferry across the lake.
asserted
Levin → reach → lake
Andrew Hickey was on a trip in Italy and turned to ChatGPT to find a good restaurant near his hotel and confirm it was open.
“I walked over to find out that they are closed on Mondays.
asserted
they → turn → Mondays
Not just this particular Monday, but they have always been closed on Mondays,” the 46-year-old traveler from New Jersey told The Post.
asserted
traveler → close → Post
“When I went back to AI to tell them that this is inaccurate information, they replied, ‘my bad 😂.”
asserted
they → go → them
And while planning a trip near Banff, Hadis Hesabi — the co-founder and CTO of an AI trip-planning startup — ironically used AI and was recommended a scenic waterfall that sounded convincing enough to visit.
asserted
that → plan → waterfall
Hesabi told The Post the description had enough detail to seem plausible.
asserted
description → tell → detail
“We only caught it when we tried to verify the location before going,” she said, noting that the experience became a guide for how she built her own company.
“Rather than letting the language model invent geography, we built a knowledge graph of real places and geographic relationships for our planner, Xavier, to reason over,” Hesabi said, cautioning that you should always verify your itinerary against a map or authoritative source before you go.
asserted
you → catch → map
Travel advisors told The Post that botched itineraries are just one part of the problem.
asserted
itineraries → tell → problem
Even when AI gets the facts right, it can still suggest a trip with basic options, tourist traps, or even a long line.
asserted
it → get → options
But some travel agencies can cost thousands of dollars annually in memberships, while others take a commission on the trip they plan.
asserted
they → cost → trip
AI, meanwhile, can spit out an itinerary for free — an obvious draw for travelers who don’t want to pay a professional planner.
asserted
who → spit → planner
But you know what they say, cheap is cheap.
asserted
cheap → know → what
Amy Norman, founder and CEO of bespoke travel agency Nellie, said one client brought her an AI-generated Tokyo itinerary filled with legitimate suggestions — Shibuya, Meiji Shrine and other familiar landmarks.
asserted
client → say → suggestions
“It was the equivalent of telling a first-time visitor to New York to see Times Square and Central Park,” Norman told The Post.
asserted
Norman → tell → Post
“Our clients are looking for the opposite: access, originality and a trip that feels deeply personal.”
asserted
that → look → opposite
When one Nellie client traveled to London with young children obsessed with Harry Potter, Norman’s team didn’t stop at booking the obvious studio tour.
asserted
team → travel → tour
They arranged for actors to arrive in costume and in character, then led the family around London to locations connected to the films—turning the entire day into part of the fantasy.
asserted
actors → arrange → fantasy
A human planner is also considering things an itinerary generator can easily miss, the luxury travel planner said.
asserted
planner → consider → things
…and 22 more, not listed.
The Aporia
