The return of the (robot) travel agent
Illustration by Erik CarterI was recently sitting in a hot tub with a friend — a glaciologist who studies how quickly the southern ice is sliding into the sea — when she mentioned that she had recently planned her...
I was recently sitting in a hot tub with a friend — a glaciologist who studies how quickly the southern ice is sliding into the sea — when she mentioned that she had recently planned her honeymoon using ChatGPT. Our fellow bathers burst into laughter. “You’d done it, too?” This, apparently, is the present state of get-togethers among friends in their early 30s: six people and three AI-assisted honeymoons between them.
My friend is a pro at arranging helicopters and snow cat brigades to remote wedges of ice. But she was overloaded with decisions about charger plates and floral arrangements, and had put the task of arranging a 12-day trip to Tasmania on her husband-to-be. A statistician, he was using ChatGPT to answer questions at work, following the advice of a mentor who told him he’d better make a habit of it. So he asked it for an itinerary that would emphasize the couple’s love of nature, adventure, and (it being a honeymoon) luxury. A specific request: time for at least one lengthy trail run.
“I was almost embarrassed,” he told me later. “It felt inauthentic asking a chatbot to plan my honeymoon.” What did it mean, exactly, to leave a major life experience to the inscrutable statistical mean of an internet’s worth of information? He was holding onto an ideal of himself, he said, as a seasoned traveler that “just inherently knows what to do and how to find things.” But also, the things the AI found looked pretty good. The itinerary portioned out the island’s attractions — mountains, coastline, wineries, an interesting town — on a logical route. It dutifully set aside time for multiple day-long runs (rewarded, in one case, by a couple’s massage at a luxury lodge). He asked for fewer hops between hotels, and the plan was gradually refined. Three hours later, he’d booked three hotels.
The travel industry is often a gateway for new technology to go from early adoption to mass use. Edie Cohen, a travel agent with more than 50 years of experience (starting when she was 14, with a temp job during the 1966 New York City school strike), recalls phone calls and reservations written on index cards giving way to teletype and then fax. Fixed-price family flight bundles became dynamic pricing algorithms. The work grew faster and more automated. “Boy, who knew what was coming?” Cohen tells me.
The internet turned Cohen’s profession into the premier case study for worker automation. Who needs a neighborhood travel agent when an online travel agency, like Priceline or Travelocity, lets you compare prices and make the booking yourself? If your business was buying people airline tickets, you didn’t make it. The airlines no longer felt a need to give out commissions for bookings, so mom-and-pop travel agents had to raise their fees. Most people balked.
What the internet didn’t have a tightly packaged answer for was travel planning. Perhaps you called up Cohen, who specializes in deciphering what it is that you actually want from leaving your home. Or you just borrowed a Lonely Planet guidebook from the library to gather inspiration. Those resources remain, in some form, but the internet is seductive. A world of information demands something be done with it.
“I was almost embarrassed. It felt inauthentic asking a chatbot to plan my honeymoon.”
So now you consult Reddit or Wikitravel or TikTok or Instagram or TripAdvisor in search of ideas. You rely on Google reviews or sift through a mix of online magazine articles and travel blogs, most of which are probably trying to sell you something. A friend told me about a co-worker who would regularly pay five locals on TaskRabbit to dish out unique dashes of local flavor wherever he’s headed. When I recently booked a hotel in Tokyo, I spent a few weeks exploring neighborhoods on Google Street View, peering into the pixelated windows of stand-up bars and ramen shops. What was I looking for? If I knew that, answering would certainly spoil the experience.
Jeremy, a remote finance worker from California who travels for eight to nine months of the year, reminds me that, according to Dutch psychologists, people tend to be happiest during the planning stages of a trip. This may have something to do with daydreams amid drudgery, he suggests, speaking over the clattering din of a cafe in Buenos Aires, Argentina. Then you actually have to do the thing. This is one of the most depressing scientific findings I have encountered.
And yet, planning has also become a chore, Jeremy admits. It could be all the internet clutter or because the stakes of planning seem to grow higher the more information that is available. (He had to be careful recently when consulting blogs for the best Patagonia hikes, to avoid looking at any photos, he tells me, which would ruin the experience.) Using AI shortcuts here and there had become a salve, Jeremy says. As a test, he asked ChatGPT for scuba spots in Indonesia, a country he knows well. He was surprised when it gave him indie spots the bloggers always miss. “I was like, ‘Oh shit, this is actually pretty useful,’” he says. “Those travel influencers are going to go out of business.”
Madison Rolley, a TikTok travel influencer based in Nashville, is inclined to agree — at least for a certain kind of poster. Like Jeremy, she’s relishing no longer having to scroll through the life stories of travel bloggers. And like my friend’s fiancé, she’s familiar with the small crisis of authenticity that such shortcutting triggers. She compares it to the liberation of using ChatGPT to automate her emails. Maybe we’ve become overly enamored with that sort of labor, convincing ourselves that the slog of DIY planning is what an authentic travel requires. “We love the struggle,” she tells me. Maybe it is all, truly, just logistics.
Rolley foresees a healthy business helping people make the best use of tools like ChatGPT to make their travel more efficient. People do need help. The internet is full of testimonials from journalists and bloggers having rather bad days designed by AI. The software invents hotels or suggests long-closed restaurants or gives implausible routes. Some of that can be managed by rephrasing your queries and asking the AI to correct its errors, explains Jaideep Patil, a developer of Forge My Trip, a tool that helps make those refinements. Time should also improve the AI systems as they come to rely less on dated training data and access more up-to-date information from the internet.
Then again, the internet might just be the problem. “I don’t think AI really gets it,” Cohen, the travel agent, tells me. In the end, ChatGPT is spitting out the same advertisements that make every resort look wonderful, she thinks. The same reviews that decide what is “luxury” or “drab” based on anonymous reviews and SEO phrases, the same blogs that send tourists straight into tourist traps. (And this is all before the internet gets flooded with AI-generated content itself.) Cohen’s peers made it this far by asking travelers the right questions about who they are and what travel actually means. That seems to be all that people are really looking for, whether they ask her or Google or an AI chatbot.
Maybe so. Another friend from the hot tub, similarly overwhelmed with wedding planning, told me he had in fact tried out a human travel agent. It seemed like a promising idea, but their tastes didn’t mesh. Instead, he and his partner went with not much of a plan at all, and would wake up in Seoul or Kyoto and consult their phones. He would sleuth on Google Maps and TripAdvisor, and she would consult ChatGPT, or “Chatty” as she calls it. They’d pick a plan when their exploratory journeys converged.
So had AI planned their honeymoon? I wondered. They looked at me with different answers in their eyes. Maybe I was getting too hung up on the logistics. Once they stepped out the hotel lobby into the sticky summer air, a day as newly married travelers in a foreign city stretching ahead of them, the answer seemed hardly to matter at all.