“AI is planning travellers’ entire itinerary.”

🟢 This is TRUE. While AI has not significantly changed travellers’ preferences, it is changing how they discover and plan their trips.

AI is not changing what travellers seek, but how they discover and plan their trips. This is especially evident in China, where travellers increasingly favour interest-led, personalised itineraries over standard checklists, whether for family holidays, honeymoons, diving, wellness, or adventure.

Itinerary-planning behaviour has become a dominant use case, with roughly one in five enquiries asking the AI to design an entire trip rather than answer a single question.

Chinese travellers, especially younger ones, increasingly begin with an AI assistant such as DeepSeek, Doubao or the tools built into OTAs like Trip.com and Fliggy, rather than a keyword search or a long scroll through Xiaohongshu. They ask for a whole trip in one sentence: i.e. a four-day family break in Hokkaido, close to a ski slope, within a set budget.

Before we dive into understanding what the statement implies…

How much do travellers trust and follow the generated itinerary?

Before a majority of consumers are ready to allow an agent to book on their behalf, certain elements will need to be addressed, including perceived accuracy and relevance, which are major draws of AI in the first place. The more accurate and relevant the recommendations are, the more likely consumers are to make a booking. I think consumers will also want some level of transparency and human oversight, at least initially. (Read more from EMI)

Our recent survey of thousands of participants show that acceptance of travel itineraries depends more on itinerary quality than on whether recommendations come from GenAI or traditional platforms, with nearly identical satisfaction ratings. Tourists focus on the quality of recommendations, with operational feasibility being the most influential, followed by the credibility of destination information. When itineraries were perceived as highly feasible, 84% of participants indicated a willingness to commit to the itinerary, compared with only 50% when feasibility was low.

What about AI’s Past Record of Hallucination?

LLMs most commonly hallucinate when they lack precise information, are pushed to guess rather than admit uncertainty, or deal with obscure topics. They're also overly eager to tell the user what they want to hear rather than push back.

Two tourists in Peru paid to be dropped on a rural road in search of a "Sacred Canyon of Humantay" that doesn't exist, at an altitude where being lost is not a joke (that really happened by the way).

Around 37% of tourists now use AI for travel advice and trust it more than review sites, yet tourism research suggests 90% of AI-generated itineraries contain mistakes (CNN, 2026).

The first thing I teach in my Gen AI seminars is how to turn your chatbot into a one-click fact-checker, and it takes thirty seconds.

Put a standing instruction in your assistant's custom settings telling it that every factual claim must carry an inline link to the exact sentence in the source that supports it, using a text-fragment URL — the links ending in #:~:text=, which scroll to and highlight the relevant passage like this.

Now, let’s talk about what should be done to adapt to this behaviour change.

Is there anything a DMO should do about this change?

The biggest change that needs to happen is actually making sure the machines get to hear what the destinations are saying about themselves. We’re still designing websites for humans, but non-human traffic already exceeds human traffic. For years, there were two kinds of bots: Google and everything else, and most organisations sensibly blocked all but a few of the bots. But now those bots are critically important to influencing AI responses.

Before you change the way you position your destination, you need to ask the unglamorous question: whether you’re letting the bots in and whether or not they’re successfully reading what you already have. Too many organisations are using outdated bot-blocking rules that are handicapping their content by preventing the very visits that would help their AI visibility.

Build AI personas of your target visitors and use them to conceptualize new products and messaging strategies. (Proof on its accuracy). On-demand synthetic feedback is a game-changer for the early stages of testing new concepts. Give them your ideas and then test the survivors on real humans, because synthetic audiences are good for a first pass but poor at emotional truth.

Build Generative Engine Optimization. Chatbots are already describing your destination; the question is whether they're describing it correctly. The fix is technical but effective. Add an llms.txt file to your website and use software to convert the pages into AI-agent-ready markdown (.md) files. Publish structured, machine-readable destination data, and content with specific facts and named sources. As with SEO, doing these simple things will give you a leg up on the competition.

This new search approach is helping more niche, experience-led travel and smaller destinations that rarely surfaced in traditional search rankings.

Visibility for destinations now depends on whether an AI model can find, understand and trust your information. Tourism boards and hotels are rewriting content in clear, structured Chinese, keeping prices and availability current, and working with the platforms these assistants draw on. Striking imagery still matters on social media, but accurate, machine-readable detail is what gets a destination recommended.

How are DMOs acting on this new source of information?

For Tourism Fiji, this means translating individual interests into bookable journeys. In China, our AI ecosystem collaborating with TikTok and Fliggy, guides travellers from discovery through tailored itineraries to real-time booking. The pivotal shift from telling travellers what Fiji offers to helping them comprehend how Fiji meets their unique aspirations makes our marketing more responsive and measurable.

AI enables us to move beyond a one-size-fits-all image. We can now respond with greater precision, helping travellers discover the islands, cultures, and hidden corners of Fiji that resonate most personally with them.

We treat Incheon Easy as a real-time signal based on actual search keyword data. This lets us identify which attractions, routes, and seasons are underused by specific visitor groups, and run customised campaigns accordingly, rather than promoting the same popular sites to everyone. As a result, we can pursue more balanced destination management — for example, steering interest toward less crowded sites during peak periods.

Previously, visitor communication was largely one-way — we published information and measured interest through participation, redemptions, or view counts. With AI, communication has become genuinely interactive: visitors now describe, in their own words, exactly what they are looking for, letting us monitor actual search keywords and query content rather than counts alone.

What steps are OTAs taking to improve the precision of their AI tools and reinforce travellers' trust and safety?

An AI assistant is only as reliable as the data underneath it. A model that invents a hotel rate or offers a room that sold out an hour ago damages trust fast, and in travel the cost is a real person stranded at check-in. The leading travel distributors and tech providers are responding on several fronts. They are grounding AI answers in live inventory, so prices, availability and cancellation terms come directly from the supplier at the moment of the query rather than from the model's memory. They keep the final step transparent, showing the traveller the exact booking conditions before payment. And they are tightening security and data governance as AI agents begin to transact on users' behalf. At Dida, this is why we built Dida MCP. It connects AI assistants directly to verified, bookable hotel content, so a recommendation becomes an accurate reservation.

OTAs like Booking.com are improving precision by grounding AI responses in existing platform information, including verified inventory, property details, policies, availability and traveller reviews, rather than relying on open-ended information generation alone. This is supported by testing before deployment, ongoing performance monitoring, privacy and security safeguards, transparent communication about AI’s limitations, and access to human support and recourse. For example, Booking.com’s Property Q&A feature retrieves relevant information from property listings, reviews and photos to answer specific traveller questions through natural conversation, reducing comparison effort while keeping the underlying information available for verification. The approach is grounded on “assistance before autonomy”: AI can search, synthesize and explain, but travellers should retain meaningful choice and confirmation over important decisions.

“AI is ruining the planet.”

🟠 Well, this is not entirely wrong, but AI can also do a lot of good when deployed at the right place.

It is undeniable that the appearance of Gen AI and the mass adoption has led to a visible increase in use of resources.

There’s no stopping the use of AI, so might as well join it- but for the good forces.

How can AI immediately contribute to more responsible tourism outcomes?

Satellite imagery, sensor networks and audit data can identify deforestation, water stress, reef degradation and energy waste far earlier than manual inspection. EarthCheck uses AI to analyse audit data across its certified network and identify patterns that an individual assessor might miss.

Detection alone, however, does not drive conservation outcomes. A monitoring system that identifies a problem achieves little unless an operator, regulator or community commits resources to address it. The gap between "we now know" and "we now act" is where many environmental monitoring investments fail to deliver value. Closing that gap is ultimately a governance and funding challenge, not a technical one.

Much of the underlying data to measure tourism’s environmental impacts already exists, but finding it, bringing it together, and turning it into useful insights has traditionally been a manual and time-consuming process. AI can significantly reduce that burden.

For example, Sustainable Travel International’s Humí extracts data from existing tour itineraries, then applies our established methodology to calculate emissions from each component, removing much of the technical work involved in calculating a carbon footprint.

Destinations can then aggregate measurements across the tourism sector to identify trends, pinpoint the largest sources of impact, and understand where resources and interventions can have the greatest effect.

AI can help destinations to better anticipate and manage overtourism by continuously analysing diverse, dynamic tourism-related Big Data.

Advanced deep learning AI analytical methods can extract information from this Big Data, for example, visitor flows, booking patterns of accommodation and transportation, and social media activity about intentions. This information can be used to identify emerging hotspots and forecast demand, which can help destination managers to implement proactive measures before overcrowding occurs.

What about in the long term?

AI can help destinations move from measuring individual impacts to understanding where action is needed across the tourism system. By analyzing data across the sector, AI can identify common challenges and show where businesses can make changes themselves versus where barriers require destination-level action. For example, analyzing emissions data from Humí across a destination could reveal a need for better public transportation, infrastructure investments, or new incentives that make it easier for tourism businesses to decarbonize. We also see potential to extend Humí beyond carbon, using AI technology to measure other tourism impacts such as waste or overtourism.

AI can also help destinations shift from reacting to impacts to anticipating them. Predictive modeling can show how different tourism growth or development scenarios could affect natural resources, communities, and other sustainability priorities, helping destinations plan for those impacts before they occur.

AI works well as an early-warning and benchmarking layer within a sustainability strategy. It can track certification progress, flag emerging risks, and compare performance across a portfolio at a scale that would be difficult to manage manually.

Its limitation is that it cannot replace the slower, more complex work of changing the behaviour of staff, suppliers and travellers. Certification frameworks are most effective when data is paired with accountability through verified standards, independent audits, and leadership that treats findings as a mandate for action rather than a report.

Any destination or operator adopting AI for sustainability should consider not only what the tool measures, but also who is responsible for acting on the results.

Beyond mitigating overtourism, AI can support more balanced and sustainable tourism development by helping destinations understand visitor preferences while aligning tourism activities with local capacity, environmental limits, and community needs. AI-driven decision support systems and digital twins can model and evaluate the potential impacts. As a result, destinations can enhance visitor experiences, protect natural and cultural assets, reduce pressure on local communities, and promote more equitable distribution of tourism benefits across regions.

Visitor flow is a data question before it is an AI question, and there are two layers to it. Collection is usually fragmented across different systems with different owners. Turning all of that into a single source of truth is harder still, and takes time, money and expertise.

This is where governments should be investing: data architects and analysts, a single source of truth built properly, and used to ask better questions. AI then does what it does well, finding patterns across sources, heat mapping, and forecasting pressure points early enough to act. The same foundation is invaluable in a crisis, when the problem is never too much information but too little, too slowly.

“Travel and tourism is a people industry, AI can never.”

🟠 This is mostly TRUE. There are elements within this industry that the machine can never replace, but there are also some jobs at risk. Hear what the expert has to say.

Before we dive into understanding what are the impacts

How does the tourism industry perceive the impact of AI on job security and workforce planning?

I’m seeing more automation in customer service, reservations, marketing and administration, while robotics is moving into real-world environments. As an example, just this week Singapore’s LTA began trialling Olly, a humanoid robot at one of its MRT stations. Olly can interact with passengers, provide assistance and walk with them to a platform.

Now imagine the same technology in a hotel. A robot could greet guests, provide directions, answer routine questions, deliver items to rooms or assist with check-in, while AI handles much of the conversation behind it. We’re already seeing some of these functions deployed in hotels and I believe this will move much faster than many people expect. The combination of AI and robotics will reduce the need for people to perform repetitive and transactional work.

That doesn’t mean hotels will become people-free, but it does mean they may need fewer people doing basic tasks and more people focused on service, relationships and complex situations. Workforce planning needs to prepare for that reality now, or organisations will find themselves behind not only the competition, but market expectations.

What human-centred skills and capabilities will remain essential as AI becomes increasingly integrated into tourism workplaces?

As AI and robotics take over more routine work, I believe the value and need for genuine human skills will increase. A hotel robot may be perfectly capable of delivering extra towels to a room. AI can answer a guest’s question instantly. But what happens when that guest is upset because their room isn’t ready, their flight has been cancelled or something has gone seriously wrong? That is where human judgement, empathy and ownership still matter.

Hospitality is ultimately about how you make people feel, the impression you leave and not simply how efficiently you process their requests. Communication, emotional intelligence, relationship building, leadership, creativity and problem-solving will therefore become even more important. Cultural awareness will also remain critical in a region as diverse as Asia Pacific.

I believe adaptability will become one of the most valuable career skills and people cannot assume the job they start in will look the same five or ten years later. My advice is pretty simple and direct: don’t compete with AI at the things it does best. Learn to use it and become exceptional at the human skills it cannot easily replicate.

How do we maintain “authenticity” when AI is intertwined with this people-centric industry?

Authenticity has become a fashionable word, but travellers have always wanted authentic experiences, just as experiential travel was never new. It comes down to storytelling, managing expectations and being honest about what you are. Honesty is exactly what AI-generated imagery takes away.

A DMO can say plainly that it will not pay for AI-generated visuals, will not repost or allow tagging on synthetic imagery of the destination, and will ask for it to be removed. Where a creator's genuine work is beautiful, approve it, license it and pay them for access.

AI is a part of our daily lives now. Now what?

What’s the future of AI? What risk should we be watching out for? Can it be advanced for something better, bigger?

What role can AI play in supporting more responsible, sustainable and equitable outcomes across tourism?

AI can help travellers make choices that better reflect their needs and values, while enabling destinations to distribute demand more effectively and helping smaller tourism businesses reach relevant customers. It can surface accessible options, lesser-known destinations and experiences that benefit local communities, as well as give businesses clearer insight into changing traveller preferences. For example, Booking.com’s 2025 research found that 71% of travellers valued AI recommendations that help them avoid overcrowded destinations or peak times, while 60% wanted AI to highlight experiences that positively benefit local communities. AI is not inherently sustainable on its own. However, AI's contribution depends on reliable and inclusive data, clearly defined destination objectives and measurable outcomes, with communities and public authorities the key determinants of what responsible tourism growth should mean.

AI has the potential to, and in some ways already is, radically change accessibility and inclusion in the tourism industry. For travelers with accessibility challenges, AI can shift through poorly designed websites, make bookings on their behalf, and guide them to destinations that may be more accessible or inclusive. Once on the ground, AI can provide quick guidance to tackle all the tricky situations that travelers face, such as assisting with translation, navigating new places, and even figuring out what food you’ve just been served.

However, there are also risks involved when using AI in these ways—information and facts need to be verified, bias can be unintentionally integrated into recommendations, reducing inclusivity, and small businesses or emerging destinations can be excluded, impeding value capture from tourism. More importantly, AI has the potential to keep tourists from exploring destinations in authentic and unplanned ways—which, in my opinion, is the most beautiful part of a trip.

What happens when artificial intelligence becomes part of the journey?

From discovering where to go and planning what to do, to managing destinations and resources behind the scenes, AI is changing the way the world travels. Conversational tools can tailor experiences to individual travellers, while intelligent systems are helping tourism organisations make faster, smarter decisions.

But this transformation is about more than technology. As AI becomes increasingly embedded in tourism, it also challenges us to think differently about accessibility, creativity, resource use, algorithmic diversity, and who gets to benefit from these advances. Will AI make travel more personal and inclusive — or create new gaps between those who can adopt it and those who cannot?

The 2026 World Tourism Day Campaign, led by the Pacific Asia Travel Association (PATA), invites you to explore these questions from across the tourism ecosystem. Through diverse perspectives, emerging applications and real-world possibilities, the campaign looks beyond the excitement surrounding AI to consider what its growing role could mean for the people, businesses and destinations that make travel possible.

The theme, “Digital Agenda and Artificial Intelligence to Redesign Tourism,” explores this transformation through five core areas:

Meet the Experts

  • Andrew Chan

    Founder & CEO, ACI HR Solutions

  • C.A. Clark

    General Manager, Miles Partnership South Pacific

  • Professor Haiyan Song

    Chair Professor and Associate Dean, School of Hotel and Tourism Management, The Hong Kong Polytechnic University

  • Incheon Tourism Organization

  • Jessie F. McComb

    Senior Tourism Industry Specialist, Asian Development Bank

  • Matt Gibson

    CEO, UpThink

  • Mochamad Nalendra

    Founder & CEO, Wise Steps

  • Paloma Zapata

    CEO, Sustainable Travel International