Loading summary
Transcript41 lines
- [00:00]
A
AI is moving fast across the enterprise, but without visibility, it's just chaos. ServiceNow turns that chaos into control. With the AI control tower, you see all your AI in one place. To put AI to work for people, visit ServiceNow.com.
- [00:19]
B
Hey, what's news, listeners? It's Sunday, July 26th. I'm Luke Vargas for the Wall Street Journal, and this is what's News Sunday, the show where we tackle the big questions about the biggest stories in the news. And this week we've talked here about AI and automation, taking on driving. But what about flying? With the aviation industry facing an array of challenges, from backlog of planes to safety concerns and shortages of pilots, mechanics and air traffic controllers, maybe AI can come in handy. But if you thought turning over the wheel to AI was tricky, the margins for error in the skies are even tighter. Let's get right to it. I am a self professed avgeek, the nickname for plane enthusiasts. But I am no expert in aviation, which is why I've brought in my Journal colleague, aviation reporter Ben Katz, to supply that bit of expertise to the matter. Are you ready?
- [01:15]
C
I'm ready. Thank you for having me.
- [01:16]
B
You and I both spent the past week at Farnborough International Airshow outside of London, where the sound of jets screaming overhead frequently interrupted our interviews. I, for one, pretty much broke my recording equipment one point and definitely blew up my ears trying to record an F35 Lightning, which you're now hearing. But I'm glad I went to the show because I could hear firsthand about the role that folks in the industry think AI can play and the problems that it might be able to fix. Because, Ben, there are a lot of problems. I think frequent listeners of the show will be familiar with some of these. But if you could just sketch kind of a scene for us of what has broken down across a number of fronts in the aviation world.
- [01:56]
C
As you say, Luke, the industry is really battling with a number of things at the moment. We're dealing with record high, you know, fuel prices, which is weighing on airlines. A lot of disruption, a lot of geopolitics. But even at an industrial level, if we go back 30, 40 years, competition was something that was more rife back then than it is now. We used to have three, four different airframes, all kind of competing, giving airlines a lot more choice than they have now. At the moment, we have a duopoly, right? We have Airbus and Boeing that for the most part dominate the sale of big commercial aircraft to airlines. Part of that problem is just simply the technological hurdle and the associated cost of how much money it takes to develop a new aircraft. And that's really kept out new entrants. Bombardier is a good example. The now a220, that was an independent aircraft that, you know, fell into hard times and was snatched up by Airbus. That was an attempt to kind of break the duopoly. But the promise of AI maybe, you know, lowers the burden that threshold a little bit.
- [02:50]
B
We'll see though we should say even if for Boeing and Airbus they can't make enough planes, which is another issue.
- [02:55]
C
Absolutely. I mean, manufacturing delays is something that has been plaguing the industry since even before the pandemic. Coming out of the pandemic, suppliers have really been struggling to keep up to meet the aggressive ramp up schedules. A lot of frustration out there from airlines that they can't get their planes. And then of course there's just the fact that the planes that are being built, a lot of questions around quality of those aircraft that are coming out. Obviously Boeing in particular has faced a lot of challenges in that space over the last few years with the 737 Max and more recently with the 777X. And there's an irony here because it's the question of human labor and the people that are, you know, at airports managing ATC towers that are conducting the maintenance, that are doing the ground handling, that are making the planes that are flying the planes. I mean, it's endless. There are millions and millions and millions of people in this industry and we lost a lot of them during the pandemic.
- [03:46]
B
Okay, we're going to get to really all those challenges over the course of our conversation here. And let's now start to bring in some of the voices I spoke to at Farnborough to talk about how AI could help fix things. And let's start with competition. One of your first points, basically, as you said, the kinds of commercial aircraft in the skies now are more or less variants of the same design. Boeing and Airbus making most of them. Ben Sheath is with Siemens Industry Software and he told me he's actually hopeful AI could help get some new ideas off the ground.
- [04:17]
D
Somebody's still going to have to come up with an idea, but then they can validate and test and work that idea through many, many, many, many more times to get to some confidence that is going to work and iterate those designs at such great pace. And if I really wanted to be super optimistic about the world, what you think an aeroplane looks like will not be what it looks like maybe in 10 or 15 years. And I think that could be quite fun.
- [04:42]
B
Fun for him, probably. Fun for us. Can you argue with that cheap, fast iteration of ideas? Maybe that's like the perfect use case, is it not for AI though, I guess the proof will be in the pudding.
- [04:51]
C
It's a very appealing use case, absolutely. And one that everybody is really watching closely. Airbus embarked on this ambition to build a hydrogen powered aircraft. They said that it would fly by 2035 and most recently have said, well, no, that's not really going to happen.
- [05:06]
B
Lots of supersonic planes, lots of press releases and nothing to show for it.
- [05:10]
C
Exactly. And that's kind of been a really consistent challenge. I mean, even if you look at what's known as the EVTOL industry, right. Electric vertical take off and landing aircraft, I sometimes think of them as kind of electric helicopters, although the industry would probably push back at that. They do, they do that characterization. But what we've seen is kind of a lot of the stuff coming through and not really breaking through. What the promise is is that AI can produce a way of really testing and figuring out the designs on paper. You know, are there little tweaks that we can make, almost like a Formula one car? You know, every little bit of improvement to the airframe that makes it more, you know, aerodynamic over time, a huge boon in, in terms of fuel savings. The issue here, it really comes down to the remarkably high standard that the industry has in terms of safety. Right. The regulators are at the core of these kind of projects. And that again, to go back to EVTOLs, that is one of the massive hurdles in front of them is how do we certify a new type of aircraft.
- [06:10]
B
And those are safety thresholds no one wants to lower. Regardless of whether we'd like some new planes out there. Let's talk about another thing that could maybe help get new planemakers off the ground. Being able to build enough planes if they're lucky enough to get clearance for what they've designed. I spoke to, and I know you did as well, Tom o', Leary, the CEO of Jet Zero. This is an aerospace startup that has got a pretty futuristic looking, all wing aircraft, as they call it, a sort of triangular looking plane. They want to build them in North Carolina in addition to honing that design with the help of AI. I was told they're doing the same with their future factory.
- [06:43]
E
Just think about how a company like ours makes money by maximizing the throughput of that factory. And so your ability to ramp into your throughput is one of the most critical things now imagine if you're able to simulate production. You know, like you wouldn't even think of doing 100 hours of flight training before you got in the sim and tried it out. That's the way our factory is going to work. So we're actually working with Siemens to simulate our factory so that we've produced in a simulated environment, you know, thousands of aircraft before we ever produce the first physical unit.
- [07:20]
B
Ben, your thoughts? Is that overcoming another hurdle that's maybe sunk upstart planemakers before?
- [07:26]
C
I think it's difficult to underplay how important this element is. It's something that Airbus has been actually talking about a lot in the last couple years when they're thinking about their successor to what's known as the A320. That's their best selling narrow body aircraft that competes with the 737. Effectively. The industry is expecting this massive boom. Aviation is growing, it's projected to continue growing, and we simply cannot build aircraft fast enough. Airbus says that so much of the efficiencies that will come in the next generation single aisle aircraft will be simply on the factory floor, shifting from, you know, a model where we build one aircraft at a time and it goes, you know, relatively slowly through the production line to more of a Ford style car manufacturing facility, right, where we can be producing really at scale.
- [08:14]
B
All right, we've got to take a very short break, but when we come back, we'll look at whether AI can catch maintenance issues in increasingly complex planes like you mentioned there, Ben, whether it can keep pilots focused on the task at hand and address a shortage of air traffic controllers and more. Stick around.
- [08:30]
A
Over 60 years, Medline has demonstrated a track record of delivering consistent growth. Today, Medline is proudly NASDAQ listed and is the largest provider of medical surgical products and supply chain solutions serving all points of care. See how Medline makes healthcare run better@medline.com.
- [08:52]
B
Ben, let's talk about maintenance issues. This is really important to people in the subject of a lot of anxiety around air travel and specifically on this, how engineers can catch and fix problems more reliably, we should say this is very urgent. McKinsey has estimated there's going to be a shortage of around 60,000 aviation technicians in the US by 2029. Eric Swedberg is the America's EVP at Dassault Systems. He talked to me about how engineers can use AI. I mean, he sort of presented a vision here that he said could well be necessary given the direction plane design is going.
- [09:26]
F
The only way to unlock that is
- [09:27]
B
going to be through AI.
- [09:28]
F
I can actually display that full manufacturing center that builds that airplane wing. And I can say, show me why that broke. And I'm talking to the companion. The companion says, okay, it broke because of this. And the guy that was working on that last night, he did this change. And here's the things that you need to help me figure out how to go forward and move forward on it.
- [09:44]
B
Ben, what about this? AI systems getting more and more involved in helping diagnose problems and suggest fixes. And then humans ultimately, at the last step, there crucially being the ones to execute on that.
- [09:55]
C
This, again, is a really promising area. But I think the more interesting thing is on the diagnosis of problems, you're looking at huge amounts of engine data, of airframe data, and having an AI being able to see, oh, there was a little blip at this point in flight, And I can reference that to an incident that happened six years ago in this part of the world, et cetera, et cetera. That's where I see it happening. And once again, regulators sign off on their confidence in AI systems to do those diagnoses. That's the thing that frees up the human beings, because the humans can then come in, confirm if the diagnosis is accurate, and then take action.
- [10:31]
B
Let's see. And actually, what we're going to talk about next kind of builds on that as we go from one shortage to the next, which is pilots. The consulting firm Oliver Wyman has put that shortfall in North America alone at nearly 30,000 pilots by 2032. And one institution that is training the pilots of the future is Embry Riddle Aeronautical University. And to give students there better feedback about their flying and to reduce sort of rote tasks for flight instructors, they've developed some software with a company called navi. And assistant professor and director of flight research Andrew Schneider told me how it works, both kind of to give students better feedback on how they perform during a training session, but also to kind of extract from the chatter that's occurring in the cockpit between instructor and pilot, potential repair or safety issues.
- [11:19]
G
We found over time that, you know, when you're flying an airplane, you might talk about some quirky issue like a button sticking. You know, it's fairly inconsequential for the flight, but, you know, after you land, what about producing a report that says, hey, these are that you talked about. Would you like to forward these to maintenance?
- [11:35]
B
Ben, I know your reporting on aviation safety often traces right back to that cockpit environment. Whether pilots are able to keep up with all these Inputs coming to them and your maintenance. Related stories often go back to pilot generated reports about the things they're seeing. What do you make of how technology could maybe help pilots do their job better and take tasks off their plate?
- [11:55]
C
You've picked up on two, at least two kind of really critical points. But you know, the first one just to address the pilot shortage. I mean, there's pilot institutions, the academies are really, really struggling to keep up. Airlines have been begging the schools to help them get more pilots out and into the cockpits. You know, since the pandemic, when we saw so many retirements across the industry, we've seen these airlines park up planes, cancel specific regional routes that they're flying simply because they don't have enough people. If we can make the process more efficient of getting a pilot through the training procedure, that would go a huge way towards kind of filling that gap. The second thing is this idea of maintenance reports, right? So much of the challenges that the faa, the Federal Aviation Administration has is just simply not knowing what's happening. And that's partly because airlines or, you know, in the general aviation community from flight schools or private operators don't do the reporting that they necessarily are required to do. That kind of leaves the FAA slightly in the dark. So having these kind of automated systems that immediately pass on and collect data around, you know, any potential issues, can really help catch anything before it can escalate to a real issue.
- [12:58]
B
And even when reports are flagged, as you've discovered in your reporting, pilots are all describing things in their own sort of termin.
- [13:06]
C
For a project we worked on last year, we actually analyzed over 1 million reports that are made to the FAA. And you know, with each one there's different language that's used, there are different acronyms. You know, pilots are humans and they record things differently. We deployed this AI that we created in house, obviously highly checked and made sure that it was accurate to really help us parse through a lot of that language. And what we found was pretty critical and really revealing. And it's work that the FAA certainly is doing and something that AI can really be used to do in a way that ultimately will save lives.
- [13:36]
B
Ben, finally, we need to talk about air traffic control. There was a US government report late last year that found that the number of air traffic controllers had declined by 6% over the past decade. Meanwhile, there have been a 10% increase in flights that are relying on air traffic control systems. And that sort of already sounds like a mismatch, but just you wait, because Farnborough was really packed with vendors of all sorts pushing new cargo drones, air taxis, the EVTOLs you were talking about there, the autonomous military planes that are going to be operating in formation alongside crewed aircraft. All of this just crowding up the skies. And actually, while I was there, Honeywell Aerospace announced that they had been tapped to lead an EU project to integrate cargo drones and EVTOLs into European aerospace. And Michel Zavoshek told me why deterministic, as he called them, AI systems could be key to pulling that off.
- [14:26]
H
If you look at the advanced air mobility of the future, they've been flying very close to the buildings, very close, close to very populated areas. So it's like it's really key enabler. The idea is that we will define rules and deterministic ways on how to make the decisions quickly. So there will be a lot of data being evaluated on board of the aircraft. There will be ground part of the infrastructure which will be collecting the data and proposing the decisions, being vetted by a human. But in the future, some of the decisions, decisions will be really done by the computer.
- [15:00]
B
Ben, A decent amount of jargon there, but you can sort of punch through it and see that maybe AI is just going to need to play some sort of role in deconflicting airspace and presenting decisions to the humans that are still crucially involved in the system. Ben, this feels like a job ripe for an AI shakeup, not least because of the sort of mental, emotional stress that's put on people who are trying to do this life savings work.
- [15:26]
C
The topics that we've covered up until now are, you know, really critical, really important, but in a way you could almost characterize them as kind of nice to haves, right? Save money, improve efficiency, improve safety for atc. I mean, this is critical. Having AI having these systems is something that the industry doesn't just want to have. We're at a point where we need it and we need it really, really quickly. The US has really been struggling over the last few years, but even before then with massive shortages in air traffic controllers. And that's put huge amounts of pressure on the ATC workers who are working every day to just try and keep flights flying. If you go back just to earlier this year, if listeners recall the incident at laguardier Airport where you saw the collision of an Air Canada aircraft with a fire truck on the Runway. We don't have the final report yet, but there's a lot to suggest that the air traffic controller involved in the incident was just overwhelmed. He didn't have enough support. That's a huge line of inquiry that the investigators are looking into. So, you know, this idea of deconflicting airspace is not only necessary for airlines flying today as it is now, but also, as you very astutely mentioned, the fact that we are expecting tens of thousands of new aircraft, you know, these EVTOLs, drones occupying different levels of the skies. How do you figure that all out? You know, we certainly don't have the people. Is it time to turn to an AI robot?
- [16:50]
B
Let's see.
- [16:50]
H
Let's see.
- [16:51]
B
Ben Katz is a Journal aviation reporter. Ben, we couldn't have done this without you. Thank you so much.
- [16:55]
C
Thanks.
- [16:56]
B
And that's it for what's news for July 26th. Today's show was produced by Daniel Bach with supervising producers Sandra Kilhoff and Melanie Roy. I'm Luke Vargas. And we'll be back tomorrow morning with a brand new show. Until then, thanks for listening.
- [17:16]
G
Introducing Fidelity Trader Plus. With customizable tools and charts you can access across all your devices, try our most powerful trading trading platform yet@fidelity.com trader+ investing involves risk, including risk of loss. Fidelity Brokerage Services, LLC Member nyse, SIPC.