When you board a United Airlines flight, the gate agents, flight attendants, and others are coordinating in chat rooms to ensure the plane departs on time, coordinating many tasks that you, as a passenger, may never notice. Is there still space for my carry-on baggage? Did the caterer bring the missing orange juice? Is there a way for the family to sit together?
If your flight is delayed, you'll receive an explanatory message via text and in the United Airlines app. In most cases, that message is generated by AI. Meanwhile, in offices around the world, dispatchers examine this real-time data to ensure that crews can legally fly planes without violating FAA regulations. And just a few weeks ago, United Airlines introduced an AI customer service chatbot.
Jason Birnbaum, who became United's CIO in 2022, manages a team of more than 1,500 employees and about 2,000 contractors responsible for all the technology that makes this happen.
“The things I love about our business are also the things you hate about this business,” he told me in a recent interview. “I worked in the appliance business at GE for many years. It could be down for a day or so, but I don't think anyone would notice. They'd say, 'Okay, the dishwasher is rolling off the line. But it wasn't newsworthy. Now, when something happens to him, even for 15 minutes, not only is it reported on social media, but news trucks head to the airport. ”
Prior to joining United, Mr. Birnbaum spent 16 years at GE, rising from technology manager to CIO of GE Consumer and Industrial, based in Budapest. In 2009 he became CI of GE Healthcare Global Supply Chain. He joined United in 2015 as senior vice president of digital technology, where he led projects such as ConnectionSaver, one of United's first AI/ML-based services to proactively hold flights when connectivity is poor. (which saved me $12 a few hours at SFO last week).
I wanted to talk with Birnbaum about how he and other CIOs at global companies are thinking about using AI. That's one area of innovation that airlines are focusing on. But before we talk about AI, United Airlines is still in the process of moving its services to the cloud. If there's one trend in cloud computing today, it's that everyone is trying to optimize their cloud infrastructure and reduce their spending.
United Continental Airlines YR202 3490 (CAL) 737-800 BSI interior.Image credit: United
“We're starting to see companies and startups saying, 'How do we optimize the cloud, how do we manage the cloud?' A lot of people are saying, 'We have a lot of data. We're looking at questions like, “But can we better preserve it?'' Or, 'We have a lot of new applications. Can you help us monitor them better?' Because all the tools we had before don't work anymore,” he said. The era of digital transformation is probably over and we are now entering the era of cloud optimization, he said.
United Airlines itself is focused on the cloud, specifically AWS as its preferred cloud provider. Not surprisingly, United Airlines is also looking at ways to optimize its cloud usage, both from a cost and reliability standpoint. Like many companies going through this process, this also means considering developer productivity and adding automation and his DevOps practices to the mix. We have an established presence [in the cloud]But now we're trying to continue optimizing the market as well,” Birnbaum said.
But it's also about reliability. Like all airlines, United Airlines still operates many legacy systems that are still functional. “Frankly, we are being very careful as we move forward on this journey to not disrupt surgery or create any self-inflicted wounds,” he said.
United Airlines has already migrated and decommissioned many legacy systems, and that process is still ongoing. For example, the company plans to shut down large Unisys-based systems later this year. But Birnbaum also believes United will continue to use on-premises systems. “I just want to be in the best place for the application and the user experience, whether it's for performance, privacy or security reasons,” he said.
What the company isn't trying to build, however, is some kind of comprehensive United Platform to run all its systems on. But the day-to-day operations of airlines are far too complex for that, Birnbaum said. For example, some platforms manage reservations, ticketing, and baggage tracking, while others handle crew assignments.
An employee at the United Airlines Station Operations Center at Newark Liberty International Airport in Newark, New Jersey.Image credit: Angus Mordaunt/Bloomberg via Getty Images
These systems must work together in near real-time if something goes wrong. That's why United Airlines is betting on its One Cloud provider. “I don't think he'll have one platform,” Birnbaum said. “I think we're going to get very good at connecting things and allowing applications to talk to each other.”
In practice, this means teams can now see, for example, when caterers have stepped off a plane and who has checked in for a flight. Ground teams and flight attendants can also see everything through an internal chat app.
Every flight has an AI story
While all of this work is still in progress, United Airlines is also thinking about how to best leverage AI.
One of the stories we often hear about AI/ML in large enterprises is that ChatGPT didn't necessarily change the way engineers thought about AI/ML, but suddenly it became a boardroom discussion. That also applies to United.
“We have had a pretty mature AI practice,” Birnbaum said when I asked him when he realized generative AI was something the team needed to pay attention to. “We've built a lot of functionality to manage models and do things like tuning. So the good news for us is that we had already made a pretty big investment in this functionality. What Has it changed? [when ChatGPT arrived] It wasn't that I had to take it seriously. That was who was asking about it. All of a sudden, the CEO and the board say, “We want to know more about this.'' ”
Birnbaum said United is pretty bullish on AI. “I think there are a lot of different examples in the travel industry where AI can be leveraged for both customers and employees.” One example is United Airlines' “Every Flight Tells a Story.”
Not so long ago, it was common to receive notifications when your flight was delayed, but there was no detailed information about it. Maybe your flight arrived late. Maybe there was a maintenance problem. A few years ago, United Airlines began using agents to create short notices explaining delays and sending them via app or text message. We currently get data from chat apps and other sources, and the majority of these messages are written by AI.
Similarly, United Airlines is also looking at using generative AI to summarize flight information so its operations team can quickly understand what's going on.
United Airlines flight information board. Image credit: Jim Vondruska/Getty Images
Just a few weeks ago, United Airlines also fully transitioned its chat system on United.com to an AI agent. In my own testing, the system still felt quite limited, but Birnbaum said that's just the beginning.
Air Canada famously used AI bots that sometimes gave incorrect answers, but Birnbaum said he's not too worried about that. From a technical perspective, the bot leverages United Airlines' knowledge base and should be able to control hallucinations. “But for me [the Air Canada incident] This wasn't a technology failure, it was a customer service failure. Because, I don't often comment on it, but I can say that today human agents are also giving wrong answers. We just have to deal with it and move on. I think we are well prepared for that situation,” Birnbaum said.
Later this year, United also plans to introduce a tool currently called “Get Me Close.” In many cases, if a delay occurs, customers will be willing to change their plans and move to a nearby airport. When my flight to Berlin was canceled, I had United Airlines reschedule me to Amsterdam (not very close, but close enough to get on the train and host the keynote the next morning). distance).
“Our mobile tools are great, and they are, but when people have conversations with humans, those interactions are typically focused on building optionality. “We're running late,'' and someone might say, “Can you send me to Philadelphia instead of New York?'' Could you please take me closer? We believe that interaction is a great use case for AI. ”
AI for the pilot?
After creating a system that automatically writes delay “stories” into apps, Birnbaum's team is now looking at where the same generative AI technology could be used. One area is the short briefing pilots typically receive before takeoff.
“I actually had a pilot come up to me and say, 'One of the things that some pilots are good at is they get on that speaker and say, 'Hey, welcome, guys going to Las Vegas, so-and-so.' By the way. ' And he said, “Some pilots are introverts. Could you give us an AI engine that helps us generate announcements about where we're going on the plane so we can make very appropriate announcements about what's going on?” Why not?’ And I thought that was a great use case. ”
After all, one of the main drivers of airline customer satisfaction is actually the interaction with pilots. A few years ago, United started focusing on Net Promoter Score, asking pilots to make delay announcements while standing in front of the cabin, for example. It makes sense for airlines to consider how they can improve these important interactions, but they also want to continue to allow pilots to go off-script if they can.
Another area where generative AI could help pilots is summarizing complex technical documents. But, as Birnbaum rightly pointed out, everything that involves a pilot flying an airplane is highly structured and regulated, so it will be a while before airlines start operating anything. Dew.