A roundup highlights multiple instances where AI travel guidance produced incorrect or invented travel information, including fake attractions and visa mistakes.
The Story
A roundup of AI travel-planning failures describes how chatbot-generated recommendations can mislead travellers, sometimes with consequences ranging from inconvenient detours to missed travel plans.
In Peru, two hikers preparing for a trek were warned by a local travel-agency owner just as they set out that the destination they were relying on did not exist: the “Sacred Canyon of Humantay.” The pair had believed they found a real bucket-list site and paid an extra $160, but the guidance they followed would have left them about 4,000 metres above sea level with no oxygen supply and no phone signal to alert rescuers.
The roundup also cites a case involving an Australian marketing strategist and author, Mark Pollard, who trusted ChatGPT’s assurances about visa requirements for entry to Chile. The mistake reportedly left him stranded in Mexico City after he was unable to board his flight to Chile and missed a conference he was scheduled to attend, and Australia’s Department of Foreign Affairs and Trade later warned travellers about the risks of following AI travel responses without verification.
A separate itinerary example involves a Singaporean traveller who used Notion AI to build a 12-day trip plan for Vietnam. The AI scheduled a “quick day trip” to Ba Na Hills while she was based in Hanoi, even though the round-trip distance cited in the roundup was about 1,400 kilometres.
In Lisbon, Portugal, a travel writer described requesting magic-themed recommendations and being told to try the Palácio do Mago, a place associated with local magicians. The roundup says this advice was a hallucination; it adds that the chatbot conceded it was incorrect after being pressed for further details about the venue.
Other chatbot failures described in the roundup include Layla AI Trip Planner placing the Eiffel Tower in Beijing, China, and a restaurant recommendation that the roundup says had “relocated across the water to Greenland.”
The roundup also points to errors involving travel-related services. It notes that Air Canada was ordered to pay compensation after its own AI chatbot told a customer he was eligible for a refund related to the airline’s discounted bereavement fare, despite the fare terms not supporting the refund claim.
As a countermeasure, the roundup says booking platforms including Airbnb and Expedia now display dedicated disclaimer pages stating that machine-generated responses may be inaccurate and prone to mistakes.
The roundup concludes that travel is difficult for AI because chatbots can generate persuasive-sounding answers using information that may be misconstrued or outdated, and they may also invent details when they cannot distinguish a valid response. It emphasizes the need for travellers to independently check key parts of any itinerary, including accommodations, transport, and sensitive requirements such as travel insurance and visa policies.
← More stories