✎ New Chat
⌖ Search Chat
ℹ︎ PATA Report
✦ Online Course
Notebooks
—————————————
Disclaimer
The content on this page are 100% from our experts and their great minds.
Consume with reassurance that these information are accurate and fact-checked.
Where do we begin?
“Digital Agenda and Artificial Intelligence to Redesign Tourism“
Welcome! This page gathers expert perspectives from across the travel sector to answer our biggest questions about AI in this people-centric industry. Explore our topics below to build your own verdict on AI.
This page discusses 5 key topics:
AI & Consumer Travel Planning Behaviour: How is consumer using AI to plan their trips? How should businesses & destinations adapt?
AI & DMO: What changes should destination management organisations make to catch up with AI era?
AI & Sustainability: Can AI be used for good things that’s not just business optimisation?
AI & Tourism Workforce: Is AI really going to takeover this people-centric industry? How should the workforce prepare themselves?
Click on the topic above or the notebooks in the chat panel to invite the experts to the chat. ➡️
✎ New Chat
⌖ Search Chat
ℹ︎ PATA Report
✦ Online Course
Notebooks
Recents
—————————————
Disclaimer
The content on this page are 100% from our experts and their great minds.
Consume with reassurance that these information are accurate and fact-checked.
Starting a conversation on AI & Travel Planning Behaviour…
Vincent Zheng, Tourism Fiji
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.
Mikyung Je, Incheon Tourism Organisation
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.
Florian Bois, Dida Travel
Chinese travellers, especially younger ones, increasingly begin with an AI assistant or the tools built into OTAs, 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.
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. Operational feasibility is 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. ()
Stephen Dutton, Euromonitor International
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. (More insights from EMI)
Matt Gibson, UpThink
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, stranded at a high altitude where being lost is not a joke (Source).
A couple on Japan's Mount Misen were given the wrong closing time by an AI tool and found the ropeway shut in the dark (Source).
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.
Sarah Mathews, e-Tourism Frontier
Personalisation and privacy are often treated as the same thing. Judging by how comfortably most people shop on Amazon or use Google, the day-to-day concern seems smaller than the debate suggests, although anyone who has rejected cookies and tracking should have that respect absolutely.
Done properly, personalisation is not about an individual at all. It is audience segmentation. I am a solo traveller, a dog owner, based in Asia, drawn to history and culture, wanting luxury without ultra-luxury and with no patience for anything cheesy. That is a persona, and I am not the only one who fits it.
Trust is won or lost through governance at individual business level: being clear about what you hold, why, and what you will not do with it.
Trip.com’s AI products are built on verified inventories, authorised proprietary user insights, and human oversight. TripGenie uses a hybrid AI model that combines general-purpose LLMs with our own fine-tuned models. It cites verified data sources and draws on live inventory to deliver accurate, accountable answers, while seamlessly escalating complex queries to live agents.
Trip.Planner, our one-stop AI travel hub, leverages real-time transport data, flexible planning features, and Trip.com’s over 20 million verified geo-tagged global points of interest to create practical, up-to-date itineraries that are easy to book. Expert vetting is available when needed, and expert-vetted recommendations can also be accessed via the in-app AI chat.
Both tools embed human-in-the-loop safeguards and operate under strict data-protection and privacy governance are continuously refined through testing, monitoring and user feedback.
An AI assistant is only as reliable as the data underneath it. 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.