For AI to be useful, it has to meet the real world.
For this panel on the second day of the 2026 A World For Travel forum, Ada Xu, global government and international affairs at Fliggy/Alibaba, brought together practitioners for whom the reality of travel has to match expectations.
The speakers noted that generative AI should enable the creation of complex, personalised travel products, but making them happen — and the logistics they entail — are just as critical.
Generating a seamless itinerary in seconds is one thing; ensuring it is feasible down to the last metre is another.
The industry's main challenge is no longer content generation, but wiring AI into real-time mobility data to avoid logistical hallucinations.
The first facet of the 'generative illusion' is its apparent omniscience.
"There is a disconnect between the wonderful itinerary AI will give you and what is actually possible," said Myriam Younes, managing director for partnerships and advertising at Rome2Rio and Omio.
"AI will always give you an answer. It will not necessarily tell you 'I don't know'."
Connecting AI to live inventories.
The travel sector therefore identifies a missing piece: the ability to provide the necessary technical 'reality check'.
For now, that verification perpetuates fragmentation, as users craft their itinerary with an AI assistant such as ChatGPT but then have to confirm the feasibility and compatibility of those choices on third-party services by checking actual timetables and availability.
In many regions (Asia, Latin America), bus and ferry networks remain highly fragmented and lightly digitised, making the injection of real-time data complex but indispensable.
"Is the ferry you're considering actually running in November? Is there a train you should book when you arrive? Is the bus operating at the frequency you need in the low season?" Younes added.
"There is still a disconnect between the ideal itinerary and the real conditions of access, mobility and connectivity you need to complete your journey."
AI is not a strategy in itself.
According to Ada Xu, plugging AI into live inventories is precisely the model developed by Alibaba, which has integrated Fliggy's AI assistant with hotel and transport databases in real time.
When a user describes their ideal stay — for example, three days in Paris with the family and a view of the Eiffel Tower — the assistant does not stop at theoretical suggestions: it checks live prices, confirms availability for the exact dates and generates instant booking links.
"For us, it is about ensuring we provide relevant information in the way customers want to access it," said Alastair Crossley, global head of travel solutions at AXA Partners.
"It is all about the customer journey and how we use AI to improve our interactions with customers at every stage."
The panellists conclude that AI is not an end in itself but a facilitator.
"I would say AI is not a strategy in itself for us," Crossley said.
"It is a lever that enables us to improve the continuous cycle of benefits delivered to the customer."
Those who win tomorrow will not be the ones with the chattiest chatbots, but those who can match the ease of a conversational interface with the rigour of real-time B2B transport inventories and data.
