As an example, a study carried out by Rubén Sanz, co-founder of BurgosTravel.com confirms the results:
“just in this month of June we have had 51 requests to review plans proposed by ChatGPT, Gemini, Claude… which are those we have used in the analysis. In 4 of them there were errors incompatible with real operations, in 12 the flight proposal was not the most optimal”.
AI creates a good starting point and saves time. However, the response is generated from news, travel blogs (updated or very old), and personal tips—where you don't know if it is sponsored content—and it delivers it without verifying its usefulness.
Travellers are often unaware of the additional costs; the biggest problem comes with poor advice on connections between different modes of transport.
For example, when looking for the cheapest possible flight, we find that it openly recommends booking flights without connections, between different airlines, that do not respect minimum transit times, or with airport changes at the passenger's own expense…
“for me this is a major failure; the AI, to save you 50€, can get you into a mess where, for example, you might have to buy a completely new ticket to return home (without the AI taking responsibility). Historically, we have recorded that, on average, 20% of trips have some unforeseen incident due to external factors”.
Three errors detected that can make your trip a disaster:
- Impossible Itineraries:
“The most worrying thing we see is when it cuts it too fine with public transport… The AI tends to state its answer with certainty, but it does not take into account that timetables can vary… and that, in certain places, results in you having no way to arrive or return. 8% of the cases in June had this problem. In one of the analysed itineraries created by an AI, to fit the travel days, it suggested the client sleep on Komodo Island to make the most of the morning for water activities, which would be a good idea if we didn't take into account that there is no accommodation there, and that Komodo dragons, which can be lethal, also inhabit it”.
- Generic Information:
In all cases, the AI gives the broadest possible answer, such as:“The essential places you must visit in Japan are Tokyo, Kyoto, and Osaka, combining the great modern city, ancient temples, and traditional food”.
But in the words of the travel planner: “the profile of a person who wants to travel to Japan to go to the Ghibli Studios park will require a completely different trip to a person who wants to go to see the museums on the art island of Naoshima”.
- Lack of support for unforeseen events:
“Let's be realistic, in all trips, unexpected problems can occur: a typhoon in Japan, torrential rain in Asia, a cancelled flight in Africa, a strike in America…
It is in these cases that the Human Factor becomes indispensable: that a knowledgeable person helps you and indicates how you need to proceed.
At the start of the instability in the Middle East, we relocated our clients and they returned at no extra cost, but in the media, passengers were constantly appearing stranded at different destinations, paying thousands of euros for plane tickets to return…”
Designing long-haul trips is complex; having travelled a lot in Europe does not make you a great traveller, as unforeseen events here are usually easy to resolve.
“For getaways, AI platforms are a great ally; 60% of young people (Millennials and Generation Z) already use them for all their short trips. But for more complex destinations, such as trips to Africa or Asia, validation from an advisor who has actually been to the destination is needed, rather than just giving information from a third party...
The increase in the use of these tools does not reduce the role of agencies, but is modifying their relationship with clients: AI plans ideas, and the expert organises the reality”.
