Travel website Expedia wants to get people to start their travel search on its site with AI instead of using an external search engine.

Expedia already uses AI for some customer service features and to help property owners describe their homes and hotels. The company hopes in the future that AI will help it recommend travel destinations to customers based on previous trips and bring more direct traffic to its site. It’s a long-term plan to shift the balance of power on the web — albeit one that’s still in its earliest stages for the company.

Rajesh Naidu, chief architect and head of data management at Expedia, says the goal is to get users started on their trips in one place. Expedia hopes to produce recommendations trained with its library of flight and hotel information and informed by users’ travel preferences. “By being able to train large language models on our data, this rich 70 petabytes’ worth of data we’ve gathered over the years, we can eventually recommend places to go and stay and do and continue to refine and personalize that,” Naidu tells The Verge in an interview.

According to Naidu, when people plan trips, they often start by going to a search engine to look for a destination. Only then do they visit services like Expedia to start booking travel and accommodation. There’s nothing inherently wrong with going to Google and typing “best vacation that isn’t cold and not that far from New York,” but Naidu believes there’s value in streamlining the travel planning process even more.

Travel is one of the examples AI evangelists use when talking about the power of AI assistants. Imagine needing to book a flight, so all you have to do is say your preferred flight time and airline, and the AI will book it for you. Expedia’s proposal sounds like a midpoint: AI can’t book it all for customers yet, but the goal is to get people to plan their travel in one place — you know, like travel agents before internet booking took their jobs. 

Expedia’s former chief financial officer Jeff Hurst testified during the US v. Google antitrust trial that even though the travel site increased its payments to Google for ad space, traffic never increased after the search engine began showing flights and hotels. Vertical search engines — sites that specialize in searching for information about one thing, like travel or restaurant reviews — have a long and troubled relationship with the search giant. And the more that Expedia cuts out the middleman in Google, the more revenue it can get. When I confessed that I began booking a recent vacation with a Google search before I found myself on Expedia to hunt for hotels, Naidu even jokingly pointed out that I gave my money to Google before Expedia. (I did not end up booking through Expedia.)

While all-in-one personalized travel planning is not available yet on Expedia, the service has long used machine learning algorithms to track flight prices and find hotels. It launched a Virtual Assistant tool in March 2020 to help customers with post-booking questions like refunds and confirming dates of stay, then announced a ChatGPT-powered customer service chatbot plug-in in April. The company also runs the vacation rental website Vrbo, which uses AI to help write descriptions of property owners’ homes and evaluate the quality of photos. 

Other travel companies have invested in AI, too — Airbnb bought AI-powered trip-planning startup Gameplanner.AI in November.