A perspective piece from The Listening Market
There was a time when planning a holiday meant a notebook, a stack of dog-eared guidebooks, and an evening or two on the phone with a travel agent. Then came the internet, and the notebook was replaced by two dozen browser tabs — flights, hotels, restaurant reviews, weather forecasts, visa requirements, all open at once, none of them quite answering the question at hand. And now there is something else. A quiet, conversational presence that can draft a five-day itinerary for Lisbon in the time it takes to make a cup of tea. The age of AI holiday planning has arrived, and it is reshaping the rituals of travel in ways that are both remarkable and, at times, unsettling.
The shift is not hypothetical. According to research by Simon-Kucher, 42% of travellers used AI-powered tools such as ChatGPT or Copilot for itinerary planning in 2025, while a further 33% turned to them for destination research. A survey by the travel platform Klook, which polled 11,000 global users, found that a striking 91% of global travellers now rely on AI travel planners in some capacity. Marriott Bonvoy’s own research indicates that 79% of travellers plan to holiday as much or more in 2026 than they did the year before — and a growing share of them are using AI to do the planning. AI holiday planning is no longer a novelty being tested by curious early adopters. It has become a routine part of how people prepare to leave home.
The End of the Twenty-Tab Holiday
For anyone who has spent an evening cross-referencing hotel reviews across four different booking sites, the appeal is obvious. AI excels at the kind of repetitive, low-stakes cognitive work that makes travel planning feel like a chore rather than a pleasure. It can compare neighbourhoods, estimate travel times between attractions, suggest restaurants that fit a dietary preference, and draft a day-by-day schedule — all in a matter of seconds. The appeal, according to HUMAN Security’s 2026 survey of American travellers, is less about novelty and more about efficiency. Over half of respondents — 54% — said they were comfortable using AI to plan a vacation from start to finish, and the overwhelming majority identified time saved as the single biggest benefit.
The numbers support that sentiment. HUMAN Security’s broader research found that AI-driven internet traffic grew by 187% over the course of 2025, and that more than 95% of that growth was concentrated in just three industries: retail and e-commerce, streaming and media, and travel and hospitality. Travel, in other words, has become one of the clearest consumer use cases for AI — a category defined by research, comparison, and decision-making, which happens to be exactly the kind of work that large language models were built to accelerate.
What makes AI holiday planning feel different from the tools that came before it is its conversational nature. A search engine returns links; an AI assistant returns a draft. A traveller can describe a vague idea — a quiet beach town within three hours of an international airport, good food, not too expensive, suitable for a family with a toddler — and receive back something that resembles a real plan. It can be refined, challenged, and reimagined in the same chat. The friction between having an idea and seeing it take shape has been dramatically reduced, and that reduction in friction is changing behaviour at scale.
Personalisation, With Caveats
The promise of AI holiday planning is not merely speed but personalisation. A good AI tool can, in theory, learn from the way a traveller phrases a request — the emphasis on local food, the preference for slow mornings, the request to avoid anything that feels too touristy — and shape its recommendations accordingly. This is the dream that the travel industry has been chasing since the first loyalty programme: a planning experience that feels as though it were designed for one person, not a thousand.
In practice, the personalisation is real but imperfect. AI tools are trained on vast quantities of publicly available text, which means they tend to know a great deal about well-documented destinations and relatively little about the places that exist beyond the reach of enthusiastic travel bloggers. As Shyn Yee Ho, a board member at the non-profit Smiling Gecko Singapore and a director at tourism consultancy Horwath HTL, observed in an interview with CNBC, smaller and independent properties — particularly those in developing countries — risk being overlooked by AI tools simply because they lack the digital footprint to appear in the models’ training data. “Arguably,” she said, “they need the demand more than ever.”
There is a paradox here. AI has the potential to disperse tourists away from overcrowded hotspots and toward lesser-known destinations, which is exactly what many destinations need. But in its current form, it often does the opposite. Guy Llewellyn, an assistant professor at EHL Hospitality Business School Singapore, told CNBC that many AI systems are effectively trained on “the top 10 lists” of places to recommend, which means they tend to funnel travellers toward the same well-trodden destinations, exacerbating the very problem of overtourism they could theoretically help solve.
The Hallucination Problem
Then there is the matter of accuracy. Large language models are not databases; they are prediction engines, and sometimes their predictions are wrong. In the AI industry, these errors are called “hallucinations” — a gentle word for a phenomenon that can mean a traveller arriving at a restaurant that has been closed for two years, or following walking directions through a road that no longer exists. According to research cited by the BBC, a 2024 survey found that 37% of travellers who used AI to help plan their trips reported that the tools could not provide enough information, while 33% received entirely false suggestions. A separate analysis reported by Red River Ranch found that 90% of AI-generated travel itineraries contain at least one error, and over half suggest visiting at least one attraction that does not exist as described.
These are not abstract concerns. Leigh Rowan of the travel agency Savanti Travel recounted to CNBC the story of a client in Paris who arrived 35 minutes late for a business appointment after ChatGPT suggested a route that failed to account for road closures. A journey that should have taken ten minutes turned into forty-five. “They seem like they’re edge cases,” Rowan said, “but they’re actually very common.”
The trust gap is reflected in the data. A Booking.com report on consumer attitudes toward AI, cited by CNBC, found that 91% of respondents continue to have concerns about AI, with only 35% fully trusting its outputs. That figure sits in striking tension with the Klook survey’s finding that 91% of travellers are using AI planners. People are adopting the technology faster than they are trusting it — a pattern that says something profound about the modern relationship with automation. Convenience, it turns out, can outpace confidence.
Where People Draw the Line
The HUMAN Security survey reveals an important nuance in how travellers think about AI’s role. People are most comfortable with AI when it functions as a research assistant — helping to sort, compare, and discover — and markedly less comfortable when it begins to act as a decision-maker. The sharpest trust divide appears at the moment of booking. While travellers may be happy to let AI draft an itinerary, the majority still want a human checkpoint before any money changes hands. Planning support feels useful. Payment feels risky.
This instinct is not unreasonable. A holiday is, for most people, a significant expenditure of money and time, and the stakes of getting it wrong are high enough to justify a degree of caution. When the trip matters — a honeymoon, a once-in-a-lifetime journey, a family reunion in a distant city — most people still place greater trust in human judgment, whether their own or that of a professional travel advisor. AI can help them get to a better shortlist faster, but the final call remains, for now, a human one.
There are also things that AI simply cannot do well, at least not yet. Rowan pointed out that AI tools struggle with the real-world nuances that experienced travel planners account for instinctively: seasonal weather when suggesting outdoor experiences, travel fatigue after long-haul flights, multi-generational logistics, allergies, disabilities, and intolerances. And when things go wrong — when an airspace closes, when a strike cancels a fleet of trains, when a hurricane redirects an entire region’s travel plans — AI is not going to get anyone to the top of the queue for the next available seat. For the unpredictable, the empathetic, and the genuinely complicated, human judgment still holds the field.
The Road Ahead
None of this is likely to slow the adoption of AI in travel planning. The American Express 2026 Global Travel Trends Report found that 40% of global respondents plan to spend more on travel in 2026 than they did the year before, and 74% of Millennials and Gen Z describe travel as a “non-negotiable.” Amadeus, the travel technology company, reports a 64% year-on-year increase in AI usage in travel. The direction of travel, so to speak, is clear.
What will change is the quality of the tools. Llewellyn, of EHL Hospitality Business School, argues that the hospitality industry can reduce hallucinations and improve the accuracy of AI planners by structuring and opening its data — not necessarily in customer-facing ways, but through back-end channels that allow AI models to access factual, up-to-date information about opening hours, room availability, and local conditions. As more companies make their data available through application programming interfaces, and as AI models become better at distinguishing between verified information and conjecture, the frequency of errors should decline. “AI planners are going to happen,” Llewellyn told CNBC. “The first few iterations are going to be slow. They’re going to have some issues, but it’s going to be a really impactful thing for the industry.”
The picture that emerges from all of this is one of a technology in transition. AI holiday planning is not replacing the human imagination, and it is not replacing the joy of discovering a tucked-away café or a quiet beach that does not appear in any guidebook. What it is doing is absorbing the tedious, mechanical work that surrounds those discoveries — the comparisons, the calculations, the logistics — and leaving, at least in theory, more space for the parts of travel that matter.
Whether that promise is fulfilled will depend on how the tools evolve, how the industry adapts, and how travellers themselves learn to use them. The technology is already in the hands of millions of people. The question is no longer whether AI will change the way people plan holidays. It already has. The question is whether that change will be shaped deliberately, or simply allowed to take its own course — and whether the holidays that emerge on the other side will still feel as though they belong to the people taking them.
