With the rise of artificial intelligence tools, planning trips has never seemed easier. With just a few clicks, applications offer personalized itineraries, must-see activities, and even practical tips to ensure nothing is missed. However, behind this apparent efficiency lies a growing risk: the potential for vacations to take a disastrous turn. Increasingly, travelers are reporting bewildering experiences involving fictitious locations, inaccurate schedules, and absurd recommendations.
When AI Creates Imaginary Destinations
When two tourists in Peru presented their hiking itinerary to Miguel Angel Gongora Meza, an experienced guide, he couldn’t believe his eyes. They were eager to explore the “sacred canyon of Humantay.” The only problem? That place simply does not exist.
“They showed me a screenshot filled with evocative and convincing descriptions, but it was all false,” Gongora Meza told the BBC. “There is no sacred canyon of Humantay. The name mixes two real locations that have no connection to the description.” The couple had already spent nearly $160 only to find themselves on an isolated road, without a guide, and without a true destination.
A mistake like this could have serious consequences. In the Peruvian Andes, where conditions can be extreme, getting lost can be fatal. “This kind of misinformation is risky here in Peru. At 4,000 meters above sea level, with no mobile signal and limited oxygen, an imaginary destination can literally be fatal,” warns the guide.
“Hallucinations” with Very Real Consequences
Generative AI tools, such as ChatGPT, Microsoft Copilot, or Google Gemini, are increasingly appealing to travelers. A recent survey found that one in three international tourists already use them. Dedicated apps like Wonderplan or Layla promise optimized itineraries designed like mathematical equations. However, these systems do not truly understand geography or logistical subtleties. Dana Yao encountered this in Japan. ChatGPT had assured her that the last cable car from Mount Misen descended at 5:30 PM. In reality, it had closed well before that. “We ended up stuck at the top of the mountain,” she recounted to the BBC.
Some errors are amusing, such as when Layla claimed there was an Eiffel Tower in Beijing. However, others have more serious implications: suggestions for attractions just before they close, “short” hikes that extend for miles, or unrealistic itineraries, like the case reported by the Huffington Post, where AI had omitted an essential helicopter ride in Alaska while ignoring the physical constraints of travelers exhausted from 24 hours without sleep.
At times, AI fails on crucial details. Judy Gauthier from Go City mentioned that AI had recommended a visit to Laurel Falls in the Smoky Mountains only for her to find out on Facebook that the trail had been closed for 18 months. Another time, when asking whether swimming was possible in San Sebastián, Spain, the AI warned her about shark attacks, confusing the beach with the city aquarium.
Why AI Fails at Planning Trips
These mistakes stem from how language models operate. “AI doesn’t distinguish between travel advice, itineraries, or recipes,” explains Rayid Ghani, a machine learning professor at Carnegie Mellon. “It manipulates words without genuine understanding. It generates text that appears credible, but can be completely false.”
Planning a trip is particularly complex. Transport schedules, weather conditions, local events—all are in constant flux. Dr. Niusha Shafiabady, a computational intelligence expert, highlights that “flight availability, pricing, weather, and traffic change in real-time, making it difficult for AI to stay current.” Even with accurate data, AI struggles. Research indicates that OpenAI’s most advanced model fails 90% of the time when it comes to planning complex trips.
Jay Stevens, CEO of Wayfairer Travel, tested ChatGPT to organize a skiing trip in Japan. Although the AI recommended stunning onsens (hot springs), it didn’t take bus schedules into account, which didn’t match reality. He explains: “At 11:11 AM, it was snowing, I was soaked, and I was starting to get cold.” He ultimately had to take a taxi, realizing the limitations of AI in complex logistical contexts.
The Importance of the Human Element
Human travel agents consider numerous details, such as airport security times, local holidays, or family preferences to avoid unnecessary trips. Travel consultant Nolan Burris sums up the situation well: “AI is excellent for generating ideas, but entrusting it 100% with travel planning is a risky gamble.”
Itineraries generated by AI are more useful as sources of inspiration than as definitive plans. Experts recommend verifying each suggestion: check schedules on official websites, confirm information with local groups, or use Google Maps for distances. Allowing some extra time is also essential.
Ultimately, AI cannot defy the laws of physics. It might suggest finishing a museum tour just before your flight’s departure, but it won’t be able to teleport you to the boarding gate in time. These digital tools are more akin to enthusiastic interns than travel experts. They can provide interesting ideas, but it’s up to you to distinguish what is feasible from what is purely fantasy. Additionally, here’s how a war between artificial intelligence and humanity will end.
