Your Vacation Doesn't Need an AI Travel Agent

Your vacation doesn't need an AI Travel Agent. It might want one for the fun part — the dreaming, the mood-boarding, the “what if we just went to Lisbon” spiral at midnight. But the moment money changes hands, the chatbot should lose its vote. New research keeps arriving at the same awkward conclusion: travelers love letting AI plan the trip and quietly refuse to let it book the trip. That split is not a glitch in adoption. It is the correct instinct, and it is worth defending.

The numbers tell the story plainly. More than half of Gen Z has already used AI tools to plan a holiday, and TikTok has overtaken Google as the generation’s top source of travel inspiration, according to the Travel Brand Barometer data covered here this week. The inspiration economy has fully moved. The transaction economy has not — and the gap between the two is where the trouble lives.

The inspiration is real. The bookings are not.

Ask AI users what actually went wrong and the answers are sobering. A survey by travel concierge company Greetwell, reported by Stacker, found that more than half of travelers who used AI for trip planning hit at least one bad recommendation. Roughly one in six were sent to a place, tour, or activity that did not exist at all. More than a quarter got incorrect prices, hours, or details; a fifth were pointed at options that were fully booked or unavailable; one in ten were directed to a business that had permanently closed. A wrong address is an annoyance. Building a vacation around an event that never existed is a week of your life you do not get back.

Travelers seem to sense this. In the Greetwell data, about a third of respondents used AI to research or plan a trip but deliberately stopped before booking, while only fourteen percent let AI handle research and booking together. A separate Expedia Group survey from April found that just eight percent of travelers were comfortable letting AI handle actual bookings. Booking Holdings, the company behind Booking.com and Priceline, disclosed at an industry summit this month that fewer than one percent of its room nights are currently booked through AI channels, according to TravelersToday. The industry has spent years and fortunes building the AI Travel Agent. Customers keep walking past it at the register.

The trust gap is the feature, not a bug

It is tempting to treat this as a temporary problem that better models will solve. The evidence suggests otherwise. A peer-reviewed study published this year in the Journal of Consumer Behaviour ran controlled experiments with more than a thousand participants and found that AI hallucinations — false information delivered with total confidence — significantly reduced how accurate travelers judged the advice to be, which in turn reduced their willingness to follow AI-generated itineraries. The researchers noted that eliminating hallucinations in generative AI is currently impossible. That is not a roadmap problem. It is the technology describing itself honestly.

The stakes are not theoretical. When Air Canada’s AI chatbot quoted a passenger an incorrect fare, a court ordered the airline to pay damages. A consumer review analysis documented AI booking platforms confirming reservations that were still pending, canceling trips without consent, and charging stored cards to fix the AI’s own mistakes. Gareth Williams, the founder of Skyscanner, put it bluntly earlier this year: he has been struck by how negative the public is toward AI compared with people inside the industry. There is currently no consumer protection framework specifically governing AI-driven booking errors — no settled rule for who pays when the agent gets it wrong. Until that exists, the liability for every hallucinated fare quietly lands on you.

None of this means the tools are useless. Used as an inspiration engine, a chatbot is genuinely good at the part of travel planning that used to take forty tabs: surfacing neighborhoods you had never heard of, translating a vibe into an itinerary, drafting the skeleton of a trip in seconds. Deloitte’s travel research shows AI use for travel nearly doubled in a year. The smart move is to let it do that job and then take the job back — verifying hours, prices, and existence against real sources before a single card number is typed.

Draw the line before you tap pay

The parallel worth noticing is insurance. A Canadian survey this week found that three-quarters of Gen Z plan to travel this year, yet four in ten would hit the road without travel insurance, and nearly half trust travel health advice they see on social media. It is the same pattern in a different costume: maximum trust in the feed, minimum verification where it matters. An AI itinerary and a skipped insurance policy are both bets that nothing will go wrong, placed by people who have never had to collect.

So here is the opinion, stated plainly: let the machine dream, but keep the wallet human. Use the chatbot for the midnight “where should we go” spiral. Then do the unglamorous work — check the restaurant exists, check the price is real, check the hotel takes your booking, buy the insurance — on platforms where a human company is accountable when things break. The AI Travel Agent is a brilliant intern and a terrible signatory. Promote it accordingly, and your vacation might actually survive contact with reality. For more on how the generation is already traveling, see why Gen Z trusts AI with holiday planning more than Google and why Gen Z travels hard but skips the travel insurance.