Busakorn Pongparnit | Moment | Getty Images
The ability of ChatGPT to seize upon a specific, verifiable application in a focused business sector could go a long way to underscoring it as a monetizable asset. Current use cases for generative artificial intelligence may seem limited and a bit ad-hoc, but a big explosion in use cases over the next year is the expectation, according to researchers, with customer service a prime target. That makes the travel sector, already once remade by the dawn of the internet, a likely place for rapid adoption.
Already, generative AI is being experimented with inside the travel sector, albeit with mixed results. One profound limitation of ChatGPT’s application in travel to date is that its data does not extend beyond September 2021. That is a problem for now, particularly in the world of travel where information needs to be current to be useful – but only for now.
The online travel agencies are already today in mid-reshaping from the inside by AI, and even if travelers are not aware of it, recommendations offered and decisions being made influenced by the technology.
“Booking.com has been using AI and machine learning for over a decade,” said Glenn Fogel, CEO of Booking Holdings, parent company to Booking.com. “It is ingrained in the customer journey at every step on our platform.”
That includes personalized recommendations for trips, and machine translation in more than 40 languages and dialects.
How AI becomes your personal travel assistant
Now top travel executives are thinking through the implications of the latest iteration with the new race between Microsoft-funded OpenAI’s ChatGPT and Google’s Bard, among other early examples of public generate AIs. The large language models that are at the heart of generative AI provide, “interesting possibilities in particular for itinerary building and question answering,” Fogel said. But the primary consideration, he said, isn’t so much to replace human interaction.
“Travel is fundamentally about connecting people and communities, and that human connection will always play a crucial role in the travel experience,” Fogel said.
But humans don’t possess AI’s ability to analyze vast databases and that will add new levels of value and trip support for travelers, with human interaction continuing to play a critical role in shaping the overall trip experience. “The innovation happening in travel should be all about making the human interaction between travelers and supplier partners even richer, while creating efficiencies at scale,” he said.
Booking Holding’s Kayak platform recently announced its official integration of ChatGPT, to gradually expand to more users, in a blog post – written by ChatGPT and edited by Kayak staff. It described ChatGPT “as a virtual travel assistant,” allowing for more conversational interactions with Kayak’s search engine.
“By simply typing in natural language queries, like ‘Where can I fly to from NYC for under $500 in April’, users will receive personalized recommendations based on their search criteria and Kayak’s historical travel data,” the blog noted.
The AI’s ability understand and analyze natural language allows for more personalized recommendations, too. “If someone asks, ‘I’m looking for a hotel in New York City that’s close to Central Park’, ChatGPT can understand the traveler’s specific needs and preferences and ask Kayak to provide tailored recommendations based on that information,” according to the blog.
Airport stress and anxiety is another focus
This role for AI as a trip assistant will follow the traveler through the experience including the ability to intercede quickly when travel plans are disrupted.
“The potential for AI to remove friction, surface value, predict potential problems and intervene with real-time solutions if your travels go awry, is something that continues to drive our teams,” Fogel said. “We ultimately want to recreate and even exceed the ease and personal touch of the days of a traditional travel agent through the power and use of cutting-edge technology, and AI is central to that,” he said.
There may be not better example today of travel plan disruption than the airport experience. That has made Matt Breed, chief information officer for Seattle-Tacoma International Airport, an early adopter of AI, with improving the customer experience at the airport among his top goals.
“There has traditionally been a lot of anxiety associated with travel,” Breed said, largely due to the stacking up of unknowns that accompany air travel, and airports can be a particularly fraught experience, he added. “How long will it take me to get through security? Will I have time to get something to eat? What shopping items are available near my gate?”
Breed predicts that at least some of the anxiety can be eliminated or mitigated by focused use of AI and generative AI tools like ChatGPT.
He envisions AI as an itinerary analyzer, pulling in multiple sources of real-time information and helping to optimize the passenger journey – in effect, a travel personal assistant.
“Matt, traffic to the airport today is a little heavy. I would suggest leaving at 5: 25 a.m., which given projected TSA pre-check wait times should leave you with time to grab Starbucks near gate N15. Should I place your order to have it ready when you arrive?,” the AI might be able to suggest.
And Breed is betting this hypothetical, bringing the technology into the customer’s pocket and personalizes its use, will soon be reality. “That type of scenario is not very far off, especially with the real time data feeds that are currently exposed or will be emerging in the near future,” Breed said.
Behind the scenes at Seattle-Tacoma, AI is also playing a bigger role in operations, where there is also a lot of room to improve. The airport is using the Assaia Apron AI system for surface decision-making, and while it may not directly interact with the traveler, it is customer focused.
“Being able to make incremental adjustments to our operations based on suggestions from the AI really can have a large impact on how efficiently we park aircraft and turn them around, which helps avoid those delays that travelers really hate,” Breed said.
Scheduling at the airport is always a challenge – staffing, baggage handling, parking the airplanes. AI can play a large role in helping to optimize operations in the future, especially as it gains the ability to view data across multiple systems, in real time. “One of the most impressive things that I have seen out of GPT-4 is its ability to provide structure and organization around data that at first glance seems very unstructured and chaotic,” Breed said.
Online booking sites and apps already big winners
Amid all of the ChatGPT buzz, travel analysts caution that it is easy to miss the context of an industry in which new technology has been a constant, but also often overstated at time of introduction.
“The travel industry is an industry of buzzwords,” said Max Starkov, who has been a hospitality and online travel industry consultant for thirty years. “When blockchain appeared, the industry proclaimed that travel would never be the same. The same happened when the first delivery and waiter robots came into being. The metaverse? ‘This is how people will travel in the future!'”
Instead of viewing new technology in revolutionary terms, Starkov suggests using a more boring word for the latest: tools.
“All of these technologies are simply advanced tools, enabling smart operators and vendors to better serve their customers and increase market share in the process,” he said.
He does think the biggest winners will be the OTAs like Booking Holdings, which have already implemented the plug-ins to ChatGPT, such as Kayak, Expedia, Trip.com, and are working on integrating ChatGPT into their chatbots and virtual agents on websites to enable itinerary building capabilities.
“Adding AI as a planning tool will help the OTAs to shortcut the digital customer journey and almost immediately transport travelers from the planning to the booking phase of the customer journey,” Starkov said.
In other words, the trip from suggestion to booking gets shorter.
“The OTAs are well prepared to make bookable any AI-suggested itinerary with their 2.5 million multi-room accommodation establishments, 6 million vacation rentals, 600 airlines, 250,000 local experiences, etc.”
And while human travel agents have a role to play today and few are willing to speak about the AI as anything but a complement to human workers, Starkov was more blunt. “The losers will be traditional travel agencies, tour operators, independent hotels and restaurants, which do not have the financial, human and technological capabilities to take advantage of ChatGPT,” he said.
He says the proof is already there.
“The OTAs have already proved two of the main uses of ChatGPT in travel: trip planning and customer service,” Starkov said.
For example, Expedia, Trip.com and Kayak.com were “super-fast” to implement plug-ins to ChatGPT to make ChatGPT suggestions bookable. Expedia and Trip.com have already integrated ChatGPT on their websites and mobile apps to enable trip planning/itinerary building capabilities.
“Things will move quickly from here,” he said. “I expect within weeks and months, smart airlines add trip planning, language translation and customer service chatbots, powered by generative AI.”
Echoing Breed’s view, he thinks more airports will be adding informational and customer service chatbots to their websites and mobile apps to handle passenger inquiries, provide airline schedules, provide directions, shopping and dining information, and issue alerts for traffic and schedule disruptions.
But Starkov isn’t as certain about which parts of the travel experience AI actually makes easier.
ChatGPT has defined its own best uses cases in travel as enhanced chatbot, virtual assistants, translator of content, and marketing and website copywriting. The AI said in the blog post written for Kayak, “We’ve decided that the robots aren’t yet ready to take over the world … but they are ready to help people search for travel.”
None of that is guaranteed to dispel many of the core anxieties and frustrations that accompany travel, but it may make the waiting shorter, the apparent chaos more orderly, and the frustrations of unknowns fewer. And, if it doesn’t always manage to do that, at least it can order a cappuccino for you while you wait.