Getty Images; Rebecca Zisser/BI
When I booked my vacation this year, I spent about a month tracking flights and weighing various layover scenarios, and eventually used a year's worth of hoarded credit card points plus about $280 out of pocket to get a flight in an economy seat to Croatia.
Could this all have been easier and cheaper if I'd used AI? It's possible some artificial intelligence played a role in what I booked and paid. AI has seeped into every corner of the travel industry: It's pricing flights, examining dents on rental cars with eagle-eyed precision, blocking suspicious-looking Airbnb bookings, and parsing the air in hotel rooms for signs of sneaky smokers. For travel companies, AI is the vigilant cost-cutter. For travelers, however, it's a more ambivalent tagalong: AI could be a personalized travel agent, or it could be a restrictive chaperone monitoring every part of your trip and sometimes hitting you with higher costs.
There is already a lack of transparency for people to understand the costs of their trip, and that only gets murkier once LLMs are involved internally at companies. "It's the question of how are they doing it behind the scenes efficiently?" Ari Lightman, a professor at Carnegie Mellon University's Heinz College of Information Systems and Public Policy, says of travel companies using AI. "Are they doing it in your best interests or are they doing it in their best interests? Sometimes, they don't match."
That was the question facing Delta this summer, when the airline told investors that it has begun using generative AI to determine prices of about 3% of its US flights, and planned to expand that to cover about a fifth of routes by the end of the year. There was immediate pushback, including a letter from US senators expressing concern that the plan would lead to price increases "up to each individual's 'pain point.'" Delta clarified later that it has not and does not plan to use an individual pricing model based on customers' personal data, but using AI to set prices based on aggregate purchasing data for specific routes and flights and to forecast demand — the tech "recommends pricing adjustments in both directions to enhance market competitiveness and drive sales, benefiting both our customers and our business," according to the company.
AI isn't just acting like the middleman for price bargaining — it's also checking to make sure travelers behave.
It's an update to dynamic pricing, a longstanding part of the travel industry where prices rise and fall based on factors like booking date and demand. But it's clear this move is just the beginning of a more efficient pricing tool for Delta: "The more data it has, and the more cases we give it, the more it learns, and we're really excited about it," Delta president Glen Hauenstein told investors in July.
Customers already detest dynamic pricing. Ticketmaster, rideshare apps, and even Wendy's have irked many with erratic costs. Travelers can try to track price changes with Google Flights, but ultimately can't know they're getting the best deal when they pull the trigger. There are some winners, and some very frustrated losers (a friend and I just booked the same flight within minutes of one another last month — he paid $84, checked bag included, I paid $103 for no bag). Dynamic pricing with gen AI means those shifts can happen much quicker, says Richie Karaburun, a professor at New York University's SPS Tisch Center of Hospitality. Now, customers might see them flip "not on a daily basis, not hourly, but on a minutes basis," he says.
AI isn't just acting like the middleman for price bargaining — it's also checking to make sure travelers behave. Car rental company Hertz deployed UVeye, an AI vehicle scanner to check for dings and scrapes on its cars. Emily Spencer, a Hertz spokesperson, tells me in an email that the old way of doing inspections "caused confusion and frustration," and "was manual, subjective, and inconsistent." The new system brings "precision, objectivity, and transparency to the process," Spencer says. "Our goal through this initiative is to enhance the safety, quality, and reliability of our fleet and to create a more consistent rental experience for our customers." Hertz has scanned nearly 1 million cars so far, and more than 97% had no billable damage. Hertz did not answer questions I asked about the amount of times the tech has wrongfully flagged damage, but says its customer support team deals with disputes.
Airbnb, meanwhile, has been using AI in recent years to block bookings that may be for parties, looking at a user's age, length of the booking, and how close they live to the property — all data points the company says can align with those who do throw parties. By the end of 2024, the company reported a 54% decrease in the rate of all party reports in the US since 2020. That could stop people from damaging a host's property, or it might block a legitimate stay. I asked Airbnb for data on the number of booking attempts incorrectly flagged as parties, but the company did not provide it (although Airbnb does say people who have been blocked incorrectly can reach out to the company's support team to try to fix a reservation).
Travelers who want to hack the system back are using AI to try to save time and money. There's Mindtrip for personalized travel itineraries, Gondola, which pulls your reward points from your email and tries to find the best travel deals, and companies like Kayak that use ChatGPT as a virtual assistant. Expedia has been using personalized recommendations for years, but it now has an AI customer service agent handling 143 million conversations each year, and another tool that crafts vacation itineraries from Instagram reels — making that social media idealized vacation closer to a reality.
But the virtual travel agent is yet to take flight: In 2024, 47% of people 34 and younger said they felt comfortable using AI to plan a trip, the most of any age group, according to a YouGov survey of 700 adults. In 2025, that fell to 34% for adults 24 and younger, and 38% for people ages 25 to 34, according to a July iteration of the survey. SEO Travel, a marketing consultant company in England, generated 100 travel itineraries with ChatGPT last year and found that 90% of them had at least one error — including attractions that are permanently closed or suggestions to visit places outside of their business hours. Last year, I reported on a walking tour-focused chatbot by asking for recommendations on a day around New York City with my dog, and it gave me complicated itineraries that involved long walks and subway rides that criss-crossed back and forth between boroughs, and the chatbot repeatedly suggested activities that aren't dog friendly — like a spin class. I am not the only traveler led astray, there are reports of people traveling to Peru to hike mountains that don't exist or getting duped by generative AI videos of fake places.
Like Hauenstein of Delta said, the AI models are only as good as what they've learned. That gap might hurt travelers in the short-term. "You need foundational data for any of these LLMs to work or any AI, and how do you provide the most clean, standardized, complete, accurate set of data for the models to train off of?" says Andy Frawley, CEO of Data Axle, a data marketing company. But if those data sets do improve, there's potential for AI to work for both travelers and travel companies by helping to connect the right ones together. "If I have a trip planned and I've got a $5,000 budget, who can capture most of the $5,000 budget?" Frawley says. "That's where I think using AI to build that experiential part of the trip is where in the short term there's going to be a big difference."
AI might make your trip more expensive, or it might find you a deal. It's becoming harder to know if you're on the winning or the losing side of the coin. When companies push AI into their systems in the name of efficiency, they run the risk of going too far. What's good for the bottom line isn't always good for customer morale. And travel purchases are tied to our emotions. Away from home, travelers are particularly vulnerable when flights are delayed or hotels overbooked: it's not just a missed experience, but an issue that can strand them. A bad experience "has ripple effects associated with the entire vacation," says Lightman. A long line for one attraction might mean you miss the next, a rental car snafu could make you late to a wedding. A bad vacation memory can be just as impactful as a good one.
Amanda Hoover is a senior correspondent at Business Insider covering the tech industry. She writes about the biggest tech companies and trends.
