The internet gave travellers information. Artificial intelligence has given them something more seductive: instant expertise. A few prompts can produce a ten-day itinerary through Japan, complete with boutique hotels, hidden restaurants and recommendations that sound remarkably personal. The result can be so convincing that travellers may never question where the advice came from. But persuasion and trust are not the same thing. In a revealing experiment, consumers responded more favourably to recommendations attributed to professional travel experts than to AI-generated suggestions. The difference raises a fascinating question for the travel industry: what makes people believe the person behind a recommendation?
The experiment that challenged AI’s appeal
In March 2025, researchers Young Eun Park of Sookmyung Women’s University and Hyunsang Son of the University of New Mexico published an intriguing study in PLOS One.
They wanted to understand how consumers respond when travel recommendations come from three different sources: an AI system, crowdsourced opinions or a professional travel expert.
In their first experiment, 450 participants evaluated travel advertisements featuring recommendations attributed to these sources.
The result was revealing.
On a seven-point scale, advertisements featuring professional travel-expert recommendations received an average rating of 5.24. Crowdsourced recommendations scored 5.06, while AI-generated suggestions received 4.74.
The effect was statistically significant, although modest. Importantly, the experiment measured attitudes towards advertisements, not actual bookings.
But what made consumers respond less favourably to AI?
The researchers uncovered a psychological reaction that technology companies might find uncomfortable.
Consumers perceived AI recommendations as relatively unfamiliar, sometimes even eerie. That discomfort was associated with a perception that introducing AI might benefit the travel company more than the traveller.
Suddenly, the question was no longer whether AI could produce a convincing recommendation. It was whether consumers believed the technology was being deployed in their interests.
A business trip is not a honeymoon
The researchers then introduced another variable: the purpose of travel.
In a second experiment involving 551 participants, they compared reactions to recommendations for business trips and romantic getaways.
And the pattern changed.
For business travel, AI-generated recommendations attracted more favourable responses. For romantic travel, crowdsourced recommendations performed better.
Consider the psychology.
A business traveller may primarily want efficiency, convenient transport connections, reliable accommodation and minimal disruption. These are relatively structured requirements that lend themselves to systematic comparison.
A honeymoon or romantic getaway is different. The consumer may be seeking intimacy, atmosphere, discovery or an experience that cannot easily be reduced to a checklist.
Here, the opinions of other travellers may carry greater emotional relevance.
For travel curators, the finding offers a more sophisticated way to understand customers. Consumers do not necessarily have a permanent preference for human or machine-generated advice. Their expectations can change according to the emotional purpose of the journey.
And that purpose is not always obvious from the destination or budget.
The closer the departure, the greater the value of credibility
Another experiment, published in Humanities and Social Sciences Communications in June 2026, revealed a further twist.
Researchers Fangxuan Sam Li and Qianqian Su conducted three experiments involving 708 participants, investigating whether the timing of a journey influenced responses to human and AI travel recommendations.
When participants considered travelling in the near future, human recommendations were more persuasive. For journeys further away, AI recommendations could become more attractive.
The researchers identified perceived credibility as a mechanism behind these differences.
Think about what changes psychologically as departure approaches.
A holiday planned for next year is still an abstract possibility. There is time to reconsider the destination, compare hotels and experiment with itineraries.
A journey scheduled for next week feels considerably more concrete. Mistakes have immediate consequences. An unsuitable hotel, a poorly timed connection or an overlooked travel requirement can become an expensive problem.
The study suggests that the distance between planning and departure can alter how consumers evaluate advice.
For travel agencies, this opens up a particularly interesting commercial question: does the perceived value of professional reassurance increase as customers approach the point of commitment?
Why more convincing AI is not necessarily more trusted
There is a temptation to believe that making AI more sophisticated, conversational and human-like will naturally make consumers trust it more.
The evidence is less straightforward.
A 2026 meta-analysis published in Tourism Management synthesised 46 experimental studies involving 27,612 participants.
Across the studies, researchers found an overall preference for human service providers, particularly in consumers’ intentions to choose a service.
But that preference was not universal.
It became stronger in certain settings, including luxury and higher-stakes tourism experiences. In more standardised services, the difference could narrow or even reverse.
The researchers also found that highly human-like AI could sometimes intensify consumer aversion rather than reduce it.
The implication is fascinating. Making a machine resemble a person does not necessarily reproduce the psychological advantages associated with human interaction.
Consumers appear to distinguish between technological capability and the reassurance they derive from particular forms of human service.
The business of knowing what advice means
Travel agencies have traditionally demonstrated their expertise through destination knowledge, supplier relationships, itinerary design and logistical coordination.
AI has made portions of that knowledge easier to access. It has also created an abundance of recommendations that can appear equally persuasive.
Yet a recommendation is not simply information. It is a judgement about suitability.
A five-star hotel may be objectively luxurious but inappropriate for a family seeking privacy. A destination celebrated for its nightlife may disappoint someone looking for cultural immersion. An ambitious itinerary may look impressive while leaving little room to enjoy the journey.
None of these examples establishes that human consultants will consistently outperform AI. What they illustrate is the difference between identifying attractive options and judging their suitability within a particular customer’s circumstances.
The experiments also suggest that the way expertise is communicated deserves greater attention.
A consultant who can explain why a particular destination, hotel or experience is suitable may provide something beyond another recommendation: a reason for the customer to feel confident about the decision.
That confidence cannot simply be assumed. It must be supported by reliable advice, relevant experience and the ability to recognise when a recommendation is unsuitable.
Expertise is becoming a question of trust
The travel industry is entering an environment in which producing an itinerary is becoming progressively easier.
The harder challenge may be establishing why one recommendation deserves greater confidence than another.
Research does not suggest that consumers invariably prefer human travel agents. It suggests something more interesting: their preferences are shaped by the purpose of travel, perceived credibility, emotional expectations and the consequences of making a wrong decision.
For professional travel curators, that changes the competitive question.
The issue is no longer whether a consultant can provide more information than an AI assistant. It is whether the consultant can demonstrate judgement that the customer recognises as relevant, reliable and worth paying for.
AI can make a recommendation sound authoritative. But the research suggests that sounding like an expert and being trusted as one remain two different things.