In April, United Airlines hit a huge pocket of public relations turbulence after a passenger was forcibly removed from one of its partners' airplanes. The incident raised questions about blindly following procedures, passenger rights, and United's executive leadership.
Here's another question it raised: Could artificial intelligence (AI) have prevented the embarrassing drama from even happening?
AI and machine learning are already impacting many areas of business, such as marketing, as well as most industries, including retail. The travel industry in particular "is ripe for AI interventions," says Param Singh, Associate Professor of Business Technologies at Carnegie Mellon University's Tepper School of Business. From chatbots to robotic bellhops, here are seven ways AI could impact business travel in the months, and years, ahead.
1. Fewer overbooking dramas
On April 9, 2017, a paying passenger was dragged off United Express Flight 3411, from Chicago to Louisville, Ky. Four seats on the full flight were needed to accommodate airline crew members, as USA Today and others reported.
After no volunteers came forward, four passengers were selected by computer. Passengers were chosen on the basis of frequent-flier status, fare type, and connecting flight options. Three passengers eventually deplaned willingly in exchange for travel vouchers. A fourth, physician David Dao, refused, was removed against his will—and became an unwilling member of the viral video hall of fame.
AI could have helped United avoid the high-profile drama in several ways, says Henry H. Harteveldt, president and travel industry analyst of Atmosphere Research Group. In theory at least, AI could have provided an early warning to the airline's crew scheduling or planning application about a potential staffing problem on the horizon, giving the airline more time to address the issue, he explains.
Also, on the day of the flight, AI might have enabled the airline to identify the passengers most agreeable to changing their travel plans based on their profile data, Harteveldt says. Younger passengers, for instance, would potentially have more flexibility and greater interest in travel vouchers, vs. a physician like Dao, aged 69, who was anxious to return to his practice in Kentucky.
2. More personalized service
Some of the AI interventions are already happening, with chatbots for booking (such as GuestU and SnapTravel), personal travel assistants (such as Mezi), and AI to help human agents with travel planning (notably Lola), Singh says.
"Most of the AI interventions right now are what we could call machine-learning-driven," Singh explains. "With large amount of personal data, sophisticated algorithms are able to predict your needs and recommend appropriate solutions. These are, at a core level, automating the functions that people perform."
But with the next wave of applications, "we'll start seeing major improvements in the business travel experience," Singh says. "This wave will be AI interventions built on cognitive computing. These systems will have the ability to understand, learn and reason through the enormous data and then provide solutions that a human agent won't be able to provide on their own. These systems would provide value-added services and experiences, which would cognitively not be possible for the average employee in the travel industry."
The travel industry can use AI and machine learning "to learn about the habits and preferences of its frequent fliers and guests, to provide more personalized experiences," says Sumit Gupta, VP of HPC, AI and analytics at IBM. "Imagine the day when you can sit down in your seat and the flight attendant already knows just how you like your gin and tonic. Then, you're greeted at the hotel desk by name because of visual recognition software. And the Yankees game is already playing on the TV when I arrive in my room."
Wayne Thompson, chief data scientist at analytics software developer SAS, paints the following picture of AI-assisted business travel in the future:
"Let's say you have an important customer briefing in Los Angeles," Thompson explains. "You've already received a text that your flight is on time. Monday morning is one of the busiest times at the airport, and naturally you're running late. You start to worry about finding a spot to park in the packed airport garage, but then your navigation system uses image detection to direct you to the best open spot. Using convolutional networks, the computer can analyze photos of the parking lot in real time and detect images with a 6 percent error rate, which is better than the human eye."
Once you pass through airport security, "you're back on track timewise and decide to get a coffee and something to read," Thompson continues. "While approaching the book store, you're notified of special promotions based on your reading history. Then, at checkout you receive a coupon for gardening and classic car magazines, based on a recommendation system that knows these are your hobbies.
"Now you're starting to wonder why your co-worker hasn't arrived at the gate. She receives a warning that she was in the wrong terminal and gets instructions on the quickest route to the correct gate. Location services have long been used to route planes. Now, they can also be leveraged to better move passengers along and help assure that flights are on time."
Once you're in the air, you use the airplane's Wi-Fi to tweet something like: "RDU > LAX leaving on time. No complaints here. First leg of this busy travel day could have been ugly but was not." Using entity extraction and sentiment analysis software, the tweet is interpreted as positive, so the airline responds: "Thanks! We hope the rest of your day goes as smoothly. Should be sunny and 80 in LA when you arrive."
3. Smarter apps and chatbots
Many developers are already using AI and machine learning to enhance the traveler's experience via apps. For example, based on information Kayak has learned about you and what you've told the app/web service, your preferred hotel brands will be at the top of your Kayak search results. Location and context-aware data will alert you if, say, you're on a trip to Paris and rain is in the forecast. "You'd get an alert, telling you if you want to see the Eiffel Tower, go now," says Kayak CTO Giorgos Zacharia.
The Lola app, released in 2016, offers AI-based chatbot functionality along with a staff of human travel agents. "We're trying to create superhuman travel consultants who are AI-powered and can handle more trips per hour than a regular travel agent can," Lola CEO and co-founder Paul English told Skift. "They can make dramatically better recommendations than normal travel agents."
Also in 2016, 12 Radisson Blu Hotels in the U.K. began offering guests access to "Edward," an interactive, SMS-based service to answer guest questions about hotel amenities, directions, and receive guest feedback, Forbes reported.
4. Better customer service
Hilton Worldwide contact centers have AI and machine learning help from Mattersight, a behavioral routing software service, in hopes of creating a better customer experience.
"When a business customer calls into a hotel, airline or cruise line using our technology, Mattersight matches their data and analyzes their personality and behavior traits in less than five seconds," notes Andy Traba, Vice President of Behavioral & Data Science at Mattersight Corporation. "Tone, tempo, grammar, and syntax are all fed into an algorithm. That algorithm mines data from billions of customer calls to quickly pair the traveler with a call center agent who is best suited for their personality and current behavior."
For example, a caller traveling internationally who's distraught about a lost reservation "would likely be routed to a different agent than someone who calls up to check room availability," Traba says.