Episode 437
Failure Is Inevitable
In September 2004, a switch at Los Angeles air traffic control that needed a reset every thirty days was forgotten, and controllers lost radio contact with hundreds of planes. Mark Atherley, flying in from Canada, diverted to San Francisco. A long-haul captain and founder of the research firm Ancieo, he reads last Friday's nationwide ground stop the way pilots are trained to: the system caught the fault and stopped itself. He defends old, simple cockpit computers, young pilots, and a 20-year-old co-pilot whose final minutes deserve better than the commentary. In aviation, failure is assumed. That is why it so rarely wins.
--
(00:00) - Intro/Welcome
(02:06) - Why aviation doesn't hunt for blame
(06:24) - Diverting during the 2004 Los Angeles outage
(09:32) - Flying through other countries' airspace
(13:39) - Politics, investment and China's upgrade
(16:21) - Could AI break air traffic control?
(23:04) - Floppy disks and old flight computers
(27:24) - Autopilot is not AI
(31:17) - Will planes fly themselves?
(35:19) - Training a new generation of pilots
(40:54) - The Learjet crash near West Palm Beach
(44:34) - The Flying Cousin-stop
--
Jacob Shapiro Site: jacobshapiro.com
Jacob Shapiro LinkedIn: linkedin.com/in/jacob-l-s-a9337416
Jacob Twitter: x.com/JacobShap
Jacob Shapiro Substack: jashap.substack.com/subscribe
--
The Jacob Shapiro Show is produced and edited by Audiographies LLC. More information at audiographies.com
--
Jacob Shapiro is a speaker, consultant, author, and researcher covering global politics and affairs, economics, markets, technology, history, and culture. He speaks to audiences of all sizes around the world, helps global multinationals make strategic decisions about political risks and opportunities, and works directly with investors to grow and protect their assets in today’s volatile global environment. His insights help audiences across industries like finance, agriculture, and energy make sense of the world.
--
Transcript
Hello listeners. Welcome to the podcast
Jacob Shapiro:Hello, listeners. Welcome back to another episode of the Jacob Shapiro Podcast. Joining me on the show today is Mark Atherley. Mark is the guy that I text whenever I have questions about aerospace or what's going on in the world. He's aside from being a part-time pilot, he's the founder of Ancieo. This is one of the few episodes that I came in with questions that I was worried about, and I actually walked I walked out of the episode feeling better than when I walked in. Mark wants to be known as the Flying Cousin, so by the powers vested in me, Mark is now hereby dubbed the Flying Cousin. I hope you enjoy the conversation with him. Take care of the people that you love. Sorry for all of you who are trapped in travel hell. I hope you are out of it by the time you hear this podcast. Cheers, and see you out there. I'm so glad I wasn't traveling last week, and I, I've been traveling nonstop for the six weeks before that. I mean, the week before all this stuff happened in the United States, I was in Geneva, and then I went straight from Geneva to Mexico City, and then from Mexico City to home like a week before on Friday. And I was just imagining what it would be like to be stuck in the Mexico City airport after a week away from my family trying to get home, and then suddenly, you know, the FAA is gonna shut down their airspace. I mean, it's, what? The third time since nine eleven this has happened. There's a whole politicization of this conversation, pointing fingers about, well, whose fault was this? Is it they're updating you know, the wrong infrastructure? Is it this AI thing that came through? The lack of accountability from the top, from the, you know, transportation secretary. They had Mayor Pete on doing all these other things. So you're the person I go to when I want to actually understand what's going on when it comes to air travel and not just what the media wants me to think about these things. So honestly, I just wanted to get a sense of kinda how somebody who's actually doing this job, who's a part of this space, looks at what's happening in the United States. Is this a uniquely American problem? Was this weird? Does this happen everywhere? Are they really using floppy disks, and is there a good reason for them to be using floppy disks? Like sorta, you know all over the place. Start wherever you want based on that stream of consciousness question.
Mark Atherley:Well, it, it's funny actually because I-- ever since you sent me the message, I've been thinking about it, and there is so much to unpack about the whole process and the whole issue. And it's led me down some real rabbit holes, and so I'm gonna try and sort of keep a cogent train of thought going first of all. I think the first thing to say is I haven't got a clue what happened, and I don't think anybody does. And that then leads us into a thing that makes aviation so safe is the acknowledge-- you know, the acknowledgement that, that things go wrong. People often quote that lovely thing back from Gene Kranz about, you know, failure is not an option. Well, in aviation, failure is almost inevitable, and whenever somebody says that something's not gonna happen, as a pilot, your first thought is, "Well, of course it's gonna happen, and it's gonna happen to me, and it'll happen to other things, and it'll happen before you know it." It's almost like the Titanic, you know, the unsinkable ship.
Jacob Shapiro:Mm-hmm.
Mark Atherley:Which is one of the things that we've learned in aviation about, you know, building in redundancy, et cetera. Some things that leapt off the, you know, if you like, the social media page at me, first of all, was the desire to hold somebody accountable.
Jacob Shapiro:Mm.
Mark Atherley:And we don't really do that in aviation in the same way that other people do in other industries, because we've learned that the first thing you wanna do is the system needs to learn from failure, and from mistakes, and from error, and from sometimes just what my mother would have called Sod's Law, things just going wrong because of the, you know, because the inevitability of things going wrong. So without boring anybody about the intricacies of, Annex 13, I mean, people can look it up, et cetera, and all the rest of it, and there's various different ways you know, of doing it. In aviation, we don't look to apportion blame as the first thing that we're trying to achieve. What we're trying to do is we're trying to learn from the process, and we know that failure's gonna happen. We know that errors are gonna happen. We know that faults are gonna occur. But trying to apportion blame, all it does is it brings out the process whereby people will stop reporting things. With an open reporting culture, you'll have people who will put their hand up and say, "I got this bit wrong. I didn't mean to get it wrong, but I was tired. I didn't mean to get it wrong, but I was up all night with my kids." I got… Or, "I had a lack of knowledge," or this, or that, or all the various different things. Or there might be something where, for example, there's a problem with the actual infrastructure of the system. And that, that's where we often learn because it- because we're not looking to apportion blame. Yes, it's not a, it's not a carte blanche. Reckless, reckless- Yeah … behavior won't be tolerated. Neither would, you know Anything where that sort of requires, you know, punishment in that sense. But we do try to, but we do try to give people an open culture. And I have said open culture about 15 times in that process, but you could see that it's actually, it's part of the whole thing.
Jacob Shapiro:Yeah, no, and there, and there's so much to break apart there because, you know, at least, you know, according to the reporting we've got so far, I mean, it's Monday, November 16th. It sounds like they're saying that they were updating old antiquated parts of communication and radar systems, which is something that we want to happen. You want these things to be updated. They can't promise that it's not gonna happen again because obviously, to your point, it's happened once, so maybe it's gonna happen other times. Some were trying to ascribe it to a new AI system that was being rolled out. The, you know, at least the government is saying that's not what happened. There's no indication of cyber attack, although I'm gonna pick your brain about that as well. Well, but maybe like, you know, you're the man in the box, to quote Maverick. So if you're-- if you are the pilot and this happens, which is, you know, intercom- intermittent communication and even radar outage, like how serious would that have been? Like if you had been in the air over the United States at the time, like what would that have felt like to you? Would it have been a disaster? Would it have been, "Ah, like I know how to do this," but it's… Yeah.
Mark Atherley:Yeah. So the fact that there was actually a ground stop was a good thing because it shows the system works. And I was in the air when something like this happened back in 2004, and it tend- and my own personal experience in the actual incident itself was I think it was in September 2004. There was an issue in Los Angeles air traffic control, and when we were coming across the Canadian border, we were told that, and I'd never heard the phrase before, and I think I'd been flying at that point for about 15 years that, Los Angeles air traffic control was ATC zero, and we didn't, we didn't know what that meant. We said, "Sorry, what does that mean?" And they said, "Well, we don't know either because it hasn't happened forever." And what had actually happened was there was a part of the substructure, the infrastructure had a, effectively it had a manual clock, a switch, something like that, and it had to be reset every 30 days, and it had been forgotten because of human error, that thing that we try to sort of, you know, that we're aware of in aviation all the time. And it hadn't been set for 49 days, and when it got to the 49th day, it just said, "Well, sorry, I've had enough now," and switched itself off. We had to divert into San Francisco 'cause we were given sufficient warning. We were given, we had enough time to make a plan, which was again, that sort of takes you down the whole thing about decision-making, AI, all that kind of stuff that you probably wanna talk about as well. But we had time and we always were spending all of our time asking the what if question. What if, you know, we have an engine failure? What if there's the weather there? What if there's air traffic control issues, et cetera, and all the rest. So we diverted in San Francisco, put fuel on board the aircraft, made apologies to the passengers, and off we went when they managed to resolve it a f- couple of hours later. And we were s- we were actually quite fortunate because pilots and crew have got very con- tightly controlled flying hours, and we were within that time because we had the extra pilot on board to give us the rest.
Mark Atherley:Yeah, we diverted in San Francisco. There was one issue, I believe with a business jet and a cargo 757, but again, the redundancy's built into the whole process, the system whereby at the v- you know, even before Once air traffic control had actually got itself into a situation that it could no longer give guidance to those two particular aircraft, their own onboard TCAS systems, which is Traffic Collision Avoidance Systems, those systems actually gave those kept them out of harm's way, in fact.
Jacob Shapiro:Hmm.
Mark Atherley:So no, not a disaster, not great. Particularly long day out for a lot of people involved, et cetera, and all the rest of it. But, you know, it happens, and that's why w- that's why we try to learn as an industry, you know, from failure. We have to be aware of the fact that things go wrong. So yeah, it does happen.
Jacob Shapiro:Well, and I think one of the interesting things, and you're sort of already talking about it, you're talking about the open culture of the industry. And you, as the listeners can talk can probably surmise, are not American. You're flying all over the world. You're flying into, I assume, communicating with different air traffic- Yeah controllers from different countries and experiencing different infrastructure. Do you have a sense that there is this open culture that crosses borders, that there's a sort of shared project on this? Or do you have the sense, like when you fly into a particular country, "Oh, the infrastructure here is really bad," or, "These folks don't have that culture, and I'm more nervous when I'm in this airspace because I don't feel like I have that kind of access"?
Mark Atherley:So again, I'm gonna go back to ICAO and my admiration for that part of the United Nations. So every b- every country that is signed up to, ICAO and the United Nations, et cetera, has to abide by the s- by the internationally recognized standards. So no, I don't feel nervous in any particular airspace. And but I feel sometimes by natural national culture, I feel more comfortable perhaps. You know, whenever, you know, you're coming into UK tr- air traffic control areas and, you know, you s- and, you know, their call sign is London. It's lovely London calling. It's very relaxed, et cetera, and all the rest of it. I would imagine I'd be exactly the same if I was French and flying into Parisian air traffic control. There is an old joke, however, about an American air tra- American pilot captain who was leaving he was leaving Sydney as he was coming towards the end of his on his final retirement flight, and he said to it, just before he was about to be handed off to another area, he said, "I'd just like to thank you for being the second-best air traffic control service in the world. You know, thank you for looking after me over the last 20 or 30 years." And they were very pleased with that, and then eventually they had to ask the question, "Well, who's first?" And he said, "The rest of the world. Everyone else." So yeah, there's a national side of it. But no, I feel and I know that a lot of, a lot of my colleagues feel comfortable that as long as the country itself is applying the same international standards and same cultures that we- you should f- you should feel comfortable in all of them. I think, you know, some countries you know, in the old days you would have called a third world, but in the developing world won't have the same infrastructure, but that is getting better. When I used to fly over Africa in the '90s, we would have to use high-frequency radio which I'm sure, you know, some of the listeners have probably heard similar and, you know, with the static and the white noise and all the rest of it, and we would have to use that, and sometimes you'd be flying over a particular country, and it would be difficult to get, …
Mark Atherley:it would be difficult to get in contact with air traffic control in that particular area. But even those countries now have got a thing called CPDLC, which I won't-- Controller Pilot Data Link Controller, I think it is the a- is the full definition of the acronym. But that's basically like a text service between the pilots and the controllers, and you can talk, communicate quite easily with each other via satellite communications. So that's opening up a lot of the more remote spaces that there are. So, no, I don't-- And it's, I'm sure the question people might ask is this a problem unique to the United States? No, not at all. The UK's had its issues. We've had groundings due to flight planning computers failing, glitches in software. Everyone has it. You know, it, it's the nature of complex industries that, you know, you will have these issues, which again, which is the thing that reinforces the need for an open culture so that we learn from these things, not out there looking to apportion blame because the public, you know, wants something to be done immediately. That's not the way to deal with these complex situations.
Jacob Shapiro:Yeah, it's funny. I came into this maybe worried about the status of, US infrastructure, but I come out of it actually more worried about the political culture, because to your point what's happened in the last couple of days has had nothing to do with, okay, and what was the fix and what were we doing and why were we updating this and why was this a good idea and how can we fix this going forward? It was immediately politicization. Yeah. And if you were for this administration, you were saying one thing, and if you were anti this administration, you were saying another thing. You could see the Democrats lining up to take political points. You could see the Republicans lining up saying, "That's why we need more money. Give us more money because we need to update these things." I mean, a very politicized environment for these things.
Mark Atherley:I think-- I mean, I obviously as a Brit, I wouldn't want to tread on a particular political landmine. But what I would say that it's a societal issue in that there are-- people feel that, you know, we live in a world of twenty-four-hour news, social media, et cetera. People are always looking for a, an instant answer. And they don't come, you know they- Air traffic control upgrades are required because of the huge volume of traffic that there is now, you know, and the increase. And it-- and we're-- And India, for example, is only just really beginning to sort of, you know, take flight. China's out there. Even in Europe, you know, th- so supposedly, you know, with the aging demographic, tra-- you know, the traffic levels are sh- are increasing at a huge level, and that's gonna continue. So there is a, there is a need. What hasn't been done and where the-- where if you want to apportion blame to anybody, it's over the-- it's governments over the last sort of 20, 30 years that have squandered the opportunity to invest over a s- longer timescale. What they're, what they're trying to do often is they're trying to invest in the short term, and these-- the projects for these things are, are-- and the timescales are just too long, you know? They're too complex. It requires change management to, you know, to be done with cool, clear head. China, the big bugbear, have managed to upgrade their system, you know arguably to a very high standard from where they were, again, 20, 30 years ago. And their airport build, their airspace modifications and all that sort of stuff has been, you know, truly impressive. But it is a different system. It's obviously, you know, it works within five-year centralized plans and the large-- And the, also the Chinese military hold primacy over Chinese airspace, which is different from pretty much every other- Yeah every other world, every other country, sorry.
Jacob Shapiro:I think one of the reasons at least that, that folks are-- have this level of anxiety about this particular thing is because we've been hearing now about all the different ways that AI could break things, whether it's from the Anthropics or from, you know, the freezer hacks and everything else. And so they hear, "Oh, an AI system was rolled out, and we're switching from copper to fiber optic in part because we have all this data and we wanna transmit more data over long distances. And could this be the one that the AI system was the one that broke?" It turns out, no, actually, this was to your point, human error. We were replacing copper with fiber optic. This is not the one, but I think it does raise the fear about, okay, but could AI do something like this? Could AI shut down the system? Could AI get into one of these radar systems or into one of these old antiquated things and have a sort of hugging face type incident? Is that-- And I know that you're in the AI world too, so I wanna say very clearly to the listeners that's not what happened with this one, even though that was what a lot of the fear was originally. The original reporting was, is this a hack? Is this an AI thing? No, like it was actually much more boring than that. But I do think it raises the question of, okay, like what if? Like is that the sort of thing that we should be prepared for next? Yes or no?
Mark Atherley:Well, so yeah, agreed with you. This is most likely gonna end up being a switching problem in a piece of equipment which is completely AI-proof effectively because it's so old and it's effectively almost certainly got basically an on/off process. Um-
Jacob Shapiro:Mm-hmm
Mark Atherley:So I think there's a lot to unpack with AI and aviation. The UK CAA and the EASA, the European regulator, have both recently released documents, papers on the subject. And without boring people too much, they have-- I mean, I think the EASA is one that actually is probably the more straightforward to sort of explain. You have effectively three levels of automa- automation. You have level one, which is assisting a human being in decision-making. Level two, which is where it augments it. So you know, you have one which is a bit effectively like Google Search. Then you have the other one, which is like ChatGPT, which is level two, where it's giving you options, et cetera, and will do an, a degree of work. And then eventually you have level three, where the whole process is automated. I personally don't think that level three is gonna happen anytime within my working career in aviation. Level one is ef- is effectively already there in the sense that it… But it sits on the side of aviation. It gives options. And level two is somewhere that we would probably be striving towards in the next I don't know, maybe two, three years. A good use of Level Two is generating the options. So if we go back to my diversion when Los Angeles went into air traffic control zero, Level Two would've been in the ops center for the airline at the time. And what it would've done is it would've… It could've effectively rehearsed the scenario again and again, and it would've asked the what if. So that's a sort of Monte Carlo sort of, you know simulation sort of process. You can then look at, for example, fusion disruption technology gives you an insight into something where you'd look at the point where the system actually starts to break down.
Mark Atherley:Mm-hmm. And the models in that are where… Have taken-- have come from within a fusion reactor. They, what they were trying to do is they were trying to forecast into the millisecond bef- just literally just where the plasma starts to break down and destroys the reactor. You can use the same effectual logic and m- and mathematics to predict when a complex system is at that point of breaking. Now, the other day in with the FAA, we got nowhere near it because they saw the problem, and they put a ground stop in. So, so the system never got to a point of shutdown. It didn't you know, of catastrophic shutdown or an issue. And that, again, that gives me huge faith in how things work within aviation. So I think we'll see Level Two. L- And Level Two, to my, from my perspective, doesn't require the sort of agentic influence on systems whereby it would be taking decisions on its own. It would be presenting options, and it would be giving you the ability to run a simulation and therefore predict it. If you can predict something, you can plan for it, and that's a big thing in aviation where you're looking at things like, for example, discretionary fuel, where pilots are given the ability via law and regulation to carry an amount of fuel at their discretion f- when they're asking themselves the what if question. What if the weather comes into JFK on that particular time of our arrival? So I'm gonna carry 20, 30 minutes of fuel or something. Or what if air traffic control gives us a delay or a reroute? What if I'm flying over Africa and there's more convective weather in the center of the continent than forecast, et cetera? So there's an information gap, and that information gap is the thing that pilots tend to carry discretionary fuel for. Now on a, on an aircraft, you know, a long-haul flight, that could be two, three, four, five tons. I mean, old jumbo pilots used to count the lines in the weather forecast, and for each line, they'd add a ton of fuel You know, so we don't live in that world anymore.
Mark Atherley:But if you can fill the information gap, that's a big advantage that AI will give and will help with effective and more economic flying. And of course, you know, that reduces the carbon side of things as well. So I'm not concerned about AI infiltrating air traffic control at this particular point, largely because air traffic control is so, so old, antiquated, and basic and analog and not tuned into, not tuned into the same degree that, look, for example, my Mac is right now. You know, Swanick, the UK Air Traffic Control Center, was started in 1996, you know. And, you know, even the systems that are sitting there where they're, where they're talking about using AI to help predict traffic flows and to pr- and present different trajectories, they're, all they're doing is they're giving options to human air traffic controllers and pilots. And that's a good thing
Jacob Shapiro:Yeah. It's funny you probably saw that, that social media post that went viral. I reposted it myself because it was like Saturday night- Mm … and I re- I retweeted it about how they're still using floppy disks in US air traffic control, and I actually got to what you just said after the fact. I was thinking, "You know, I actually like that they're using floppy disks." It's good that this is analog. I think I would be more nervous actually if this wasn't analog because there are physical things there. But it's also very jarring. I mean, that feels very old school, so I don't-- I mean, kind of what you said about things being antiquated and analog, like that seems good, but it also doesn't seem like it's the direction that we're going. I mean, they're clipping-- You're not gonna be able to have a landline telephone in the United States in the next five years. They're getting rid of all of them. So it al- it almost seems like we're gonna eventually get to the point where we're gonna have a system that nobody's gonna know how to use, or at least the only people that will know how to do it are a very rarefied group of folks who are baptized in the analog technology that nobody else uses before. Am I kinda off on that, or do you-- Are you using floppy disks when you're in the air? I can't get the floppy disk metaphor out of my head.
Mark Atherley:Well, in a, in effect, yeah. There are aircraft flying around with so- One of the, one of the sort of misunderstandings about flying is that you press the button and you go, and the autopilot does the whole thing, et cetera. The amount of times I've heard that is the amount of times I've, you know, I've probably got well, I was about to say something rude, but I won't. But I think the point is you hear it a lot, and the pilots do two things. First of all, they fly the aircraft in the short term and in the longer term. In the short term, they use the flight controls, and then they also use parts of the autopilot system that offload workload. So for example, rather than turning it through a control column or a control yoke or something, eventually that before too long when the workload builds up, they'll be turning the aircraft through a heading knob. So they're literally just turning left, right, around the compass wheel, et cetera, and all the rest of it. They'll do that with speed. They'll do that with altitude. In the longer term, and by that I mean more than a couple of minutes, the aircraft to get the most efficient profile will be flown through a thing called flight management computer or system. Now, this is the point. Those systems, those computers are really basic. I mean, we're talking you know, some of them have been around for 50 years on a s- on an old 737, that kind of thing. I mean, they've been upgraded in the MAX and the various other things, and even on an Airbus 320, you know, some of these things are old. And they are upgraded, of course, to make them compliant with the world that they find themselves in, but the actual basic infrastructure of them is incredibly old. And that's a good thing because it's reliable. It's a very reliable system. The jet engine is actually incredibly simple really. We… I remember as a student pilot learning about it. You know, the process of a jet engine isn't complex. It's suck, squeeze, bang, and blow, which again sounds, you know, f- rude and something you might say about an actress as an bishop.
Mark Atherley:But the point being it's a simple process. It's not difficult. What becomes more advanced are the materials, et cetera, those incremental improvements like ceramic blades within the engines or the computers that help make it more efficient, that stuff becomes more advanced. No, it's just that, that there's very little in aviation where the technology is at, is, if you like, at the cutting edge in a safety critical process because it needs to be reliable and robust and resilient, and that's the other thing that… You know, another word that you, people might wanna take away from this whole process is that the system was resilient. If you'd have had an air traffic control system that had a cascading failure like this looked like it would have become- And there was no form of stop or ability to stop it, then I would be concerned. But there was. They managed to stop it. They managed to… There was a, there was an inbuilt resilience in the system, human and technological, and that gives me an enormous amount of faith. We didn't get to a point where aircraft were taking off without proper supervision and control.
Jacob Shapiro:Can I ask you a slightly ignorant question based on what you just said? Mm. 'cause you were talking about the autopilot.
Mark Atherley:Mm.
Jacob Shapiro:What is the difference between autopilot and AI? Is autopilot an early form of AI? Like how should I think about that?
Mark Atherley:No. No, no. So I mean, I remember when I started to learn how to fly, I flew an aircraft that didn't have an autopilot, then I flew aircraft that had a more advanced autopilot, and by that it basically could keep the wings level and it could make the engines go louder or smaller and all the rest of it. Do very, very basic things. An autopilot is just a system that basically executes either two things, either pilot con- pilot inputs or via direct controls, like for example, the heading knob, that kind of stuff, or via programming through a very simple computer, the flight management computer or the flight management system. There's no… There's algorithms within those processes, but there's no no LLMs involved in this kind of stuff. You know, there's no-- I mean, you and I have done enough digging, you know, to know that there's no cognition within AI anyway. You know, you… What you have is a set of very, very complex algorithms. There's no consciousness. It gives you a fabulous impression of consciousness and it will say, "Hey, Mark, you know, you're the…" You know, if it's ChatGPT, "You're the greatest pilot in the world." Thank you very much. If you're Claude, it'll tell you that you're a complete moron. You know- … they both seem to have their own particular styles. But no, an autopilot is has no cognitive functions and has no, no predictive functions and beyond the fact that it will say, for example, if you're flying at 37,000 feet at Mach speed of Mach decimal eight four, you need to start your descent at 200 miles because of the headwinds, because of the… It, it's just an algorithm. It's just a recipe, effectively. There's no, there's no real AI involved and we are a long way from that. I know Airbus have done a lot of work. They had a project called, I think it was called SPIC, which was Single Pilot in Command, in the cruise. They did a lot of work on that. Airlines have done a lot of work on reducing the amount of reserve pilots on longer flights. I think the general consensus is that we're gonna carry on in the way that we are for quite a lot bit longer.
Mark Atherley:Two pilots is the way to go, and three pilots is better, and four pilots is even better than that, you know? And it's- Mm-hmm … they all joke about why do you fly an aircraft with four engines? Because I couldn't find one with six. It's resilience and redundancy. That's… it's just written into the ethos and code of the industry. It, you know, and it's why I've had 30 plus years of it and why I would, you know, if I carried on, you know, for much longer, I would feel comfortable because as long as those processes are in it, as long as that's part of the creed, everything will be fine, effectively. The one thing I would say is I remember about, probably about 10, 15 years ago There was a shocking statistic whereby if the amount of accidents that we had, the rate of accidents, if it continued in the same way and the air traffic amount was going to generate to the forecast amounts, we would have an accident once a week, and we're nowhere near that.
Jacob Shapiro:Hmm.
Mark Atherley:We've got better, and we'll continue to get better. And again, because of the things like just culture, decision-making, you know, we haven't even started on how we train pilots to make decisions. I did a, I did a piece for the great British think tank that I think you're fa- now familiar with- Mm-hmm and I used as an example h- pilot decision-making processes on tax policy, which I thought was a bit cheeky, but the reality was it actually made a, made a-- it made the point that you need to allocate your decision-making process in better ways than we do right now, which is almost entirely reactionary.
Jacob Shapiro:Hmm. I wanted to get to you about, I wanted to get to you about pilots, so you're anticipating me. But before we get there, I just wanted to make plain I think what you said, which is, you know, we're self-driving cars are close. I have the hope of maybe my daughters will never drive a car. They'll only be doing self-driving cars. We'll see. So, you know, truckers, the trucker industry is worried about this. Sounds like you're not worried about self-driving or self-piloting planes at any point soon. Am I reading you right in that assumption?
Mark Atherley:Yeah. Yeah, you are because I-- again, it goes back to level one, level two, level three of AI if you s- according to EASA. So for me, I am an AI optimist. I don't, I don't forecast this great singularity that's gonna come where we, you know, certainly not within my lifetime or, you know, maybe I'll be wrong, and if I am. The Earl of Lord will come and get me. But- … for me, the real benefit of AI effectively is it gives-- it's a superhuman power to the human It allows humans to predict and forecast and make decisions to a degree that they-- that was just impossible before because there's the limit of how the actual brain, that front part of the brain that can only deal with about seven things. AI can deal with a million things at one time, but it-- what it doesn't have is it doesn't have instinct, and it doesn't have experience. It can examine the knowledge which is written down, and it's learning data all the time about all of us, et cetera, and all the rest of it. But it's literally, from my perspective, impossible that it will ever pick… be able to pick what goes on in and simulate what goes on in a tired pilot's brain, in a pilot that's been flying for thirty, forty years, but in the sense that… So somebody I'm close to the other day was describing how they were flying into an airport, and I'll deliberately de-identify all of that stuff for obvious reasons. They were flying behind a particular airline, which was from another country. Now, perfectly safe airline. All good. All fine. No problems. I'd fly with them anytime, particularly. And because of the pattern of air traffic control that they were on, but because of the conditions that they were in at the time, they were a captain, and they said to their first officer, "This is gonna end up in a go-around, a missed approach, and we're gonna have to abandon the landing because we're gonna get too close behind the other aircraft."
Mark Atherley:And they could see the pattern. It was the pattern recognition that they could see, et cetera, and all the rest of it, and that was built of X amount of years, X amount of experience. Lo and behold, that's exactly what happened. Now, AI can emulate that, but it's incredibly difficult in reality for it to actually generate that and for it to assign that, that, you know, that instinctual level of knowledge, almost sort of, you know, muscle memory kinda stuff. So yes, you can make an aircraft which under the right conditions can take off, can land, f- cruise, can land in all sorts of conditions without any human input, but it can't make those decisions about, for example, which way it's gonna fly around convective weather, et cetera, as efficiently and as a safety, safely as a human being would do. It cannot look at a particular carrier ahead of it and say, "Right, well, I happen to know that I'm gonna go around off this," et cetera. What it does is it gives a much more efficient and wider picture to allow the human being to make those decisions. And particularly in aviation, I'm comfortable that because of the way we train pilots, that there is a real possibility of demonstrating to the greater society and other industries how AI can be beneficial. Yeah. Well, that- Rather something to be feared.
Jacob Shapiro:No, it's a perfect segue because I wanted to ask about this too, and it's so refreshing because I feel like so often I hear that AI is a challenge for education because it's taking away critical thinking, because you can go and just get the answer. And what you said earlier was, well, AI actually allows us to produce all of these different scenarios that we wouldn't, we wouldn't have produced before. So now we have more data to feed to the pilots, and the pilots have more, information than they have before. How are pilots trained? Are you worried at all that pilots will be trained if you're interacting with young pilots, do you find young pilots who are getting to the point of being a pilot where they've been using AI too much, where the critical thinking that goes into it has atrophied, where if they lost the AI in the sky, they wouldn't know what the heck they were doing? Is the way that the system, the apprentice-ish, the apprentice-type system works, that it wouldn't get to that? Talk to me about the training system please.
Mark Atherley:Well, I'm smiling because, A, you should really ask my wife about the training thing. She's the, she's the training expert. Okay. Mm-hmm. But I'm also smiling because I was asked a very similar question at a conference in Istanbul when we were looking at, AI within aviation. And the question came from the floor, which was basically, you know-- And, well, it wasn't really a question, it was almost like a statement, which is basically they wanted this individual wanted to have older stick and rudder pilots, et cetera, because they felt that, that Gen Z, et cetera, don't have those skills. And my reply was, "That's utter rubbish." I fly with-- I mean, I should, you know, I should say I'm an, you know, I'm an older pilot. I regularly fly now with people that have been haven't been alive as long as I've been flying. I'm a part-time captain, so I fly twice a month on long-haul flights. That's about as little as you can do and stay legal, et cetera. But so I f- I don't have the exposure that others do. But from my own personal experience, young pilots of all ages, if that's not an oxymoron, but certainly all genders and all persuasions are just great. They're fantastic. … their stick and rudder skills are superb because we've, again, as an industry, we've learned how to train them well. We've l- we, you know, w- Their decision-making is top-notch. They are really, really… You, you-- If you get a young person early on and you s- you know, a bit like in the military, and they are a blank sheet of paper, and you can literally say, "Right, before the bad habits set in, this is how to make decisions in a critical time space."
Jacob Shapiro:Hmm.
Mark Atherley:And you get them going through those processes like TDODA or BRANN, NETS, all these various different acronyms. Aviation's full of acronyms, et cetera. But you can start off really early, and they're great at it. They're just so comfortable in running these sort of complex… They, they-- The habits haven't had time to-- bad habits haven't had time to bed in. Hmm. So they're very good. And as I said to the person at the conference, they're really, really good at challenging an old fart like me, which is great, which is what I want. I don't… Again, the open culture, open cockpit sort of process is you… And, you know, there's a political point perhaps to be made about yes men, et cetera, in today's, you know, governments. You want people who are gonna speak truth to power, and you want-- And accidents have happened because there's been a power gradient within the flight deck when you've had a very old senior captain and a much younger first officer who's been afraid to actually challenge the older pilot, so much so that they've flown into the ground because there hasn't been a challenge. Now we've learned from that, from those incidents and we've taken it back, and nowadays you have these younger pilots who are much more comfortable challenging us, and they keep us on the straight and narrow, and it's great. So I'm all for it, you know? AI doesn't really impinge on the day-to-day operation of a flight in the sense that apart from perhaps, you know, on a mobile device or something. So there isn't, there isn't really an exposure of that. Oh, have I lost you? I lost you there for a second. Um-
Jacob Shapiro:There but, That there has been within the industry, phases where there's been an over reliance on automation.
Mark Atherley:And again, we learned from that, which was where pilots were getting into them situations where there was a bit of mode confusion about what the f- the aircraft was doing. There was this famous phrase, "What's it doing now?" But again, we learned from that. We, we've learned that mode confusion is a risk, so we mitigate the risk, and that takes you on to threat error management, which is another key part of the whole process. You know, medicine has learnt a huge amount from aviation. There's a great book, The Checklist Manifesto, and I've forgotten who, the name of the author, but he's a surgeon, and he talks about how medicine has learnt from things like checklists. I had a, I had an operation, and I watched, and 'cause they knew I was a pilot, they were very proud of the pre-op process, which they said, you know, all the checks and balances and the checklist that they used, and they said, "Look, we do this." And they even use simulated operations now, and they throw simulated failings and faults and stuff with pa- with the patient. So yeah, no, I am not-- That's a very roundabout way of saying no, not even remotely concerned. They're great.
Jacob Shapiro:I rarely walk away from podcast conversations feeling better about something, but you, you've achieved it. I do-- Before we get out of here, though, I, you wanted to talk about the Learjet crash in West Palm Beach that happened earlier this month, and, you know, you were telling me about what your take is gonna be on this, and I-- you should make it, 'cause I don't think this- Yeah … this is certainly not a perspective I've heard, and I think it's worth sharing.
Mark Atherley:Yeah. So you know, we're both dads of daughters, and I feel that a lot of the, commentary has been wildly unfair. And, The young 20-year-old female pilot did an incredible job. Whatever the actual end result of the accident report, whatever they actually were-- wherever they actually find the root cause of the accident, I was really struck by how she kept the information coming the whole way down. And there must have been a point during that process in those last few seconds when she knew that it w- she was gonna die, but she didn't stop. She kept the, she kept the information going the whole way down. I mean, that's test pilot stuff. That's 1950s test pilot stuff. You know, incredible character, professionalism, everything about it, you know. And, you know, I thought about it and I you know, Chuck Yeager would've been really proud of her
Jacob Shapiro:Is there a particular reason or that private flying seems to have more of these incidents than sort of your main commercial aircraft? I mean, it's, you know, I've done a little bit of reading on this and sound like, sounded like she was on the hardest possible plane to fly of any plane. When I was reading about it was like, "Well, why is anybody flying these planes if they're so hard?" But-
Mark Atherley:Yeah. Well, no, there is no… Yes, statistically there will be parts of aviation that are more dangerous than other. It doesn't take the, you know, to be-- you don't have to be an expert to appreciate that being a crop duster pilot is more hazardous than being a 747 captain. I wouldn't say that business jet VIP operations are more dangerous by nature at all. They were stuck in a particular difficult scenario in the fact that they were surrounded by very active convective weather. And it is part of the, I suppose, part of the problem of aviation at the moment that we've had-- We do occasionally have incidents where weather becomes unpredictable, and that's just the nature of weather. No, I wouldn't say business jet flying is inherently more risky at all. I just-- It just really struck me that her professionalism was, for somebody of that age and that experience level, was incredible. And I think, you know, going back to our original point of apportioning blame, et cetera, all these various other things, it-- there's a lot to learn from aviation just culture, not, and how we make our decisions and how we approach our investigations of looking for the answer rather than the blame. There's a lot to be learned by other industries, I would suggest.
Jacob Shapiro:Yeah. Well, and to your point, a lot to learn about then how to face adversity when it comes- Yeah … even when the outcome is in front of you. Um-
Mark Atherley:Yeah so.
Jacob Shapiro:Mark, thank you for taking the time in. This was great. Pleasure. I hope you'll come back again soon.
Mark Atherley:Well, I've decided I want to be known as the flying cousin, which makes me sound like a-
Jacob Shapiro:Done. Done.
Mark Atherley:Yeah. Yes.
Jacob Shapiro:It shall be so. I have this power.
Mark Atherley:Yeah. Yeah. It makes me sound like I'm gonna be on a trapeze with a handlebar mustache in the '50s, but that's fine. I don't mind.
Jacob Shapiro:I so knight you the flying cousin. This is wonderful. Thank you for listening to the Jacob Shapiro Podcast. If you enjoyed today's conversation, you can find more episodes, essays, a link to our Substack, and tons more analysis at jacobshapiro.com. And if you ever want to reach out directly with questions, feedback, guest ideas, if you want to inquire about booking me for an event or any of my consulting services, you can email me at jacob@jacobshapiro.com. Also, if you found this episode valuable or interesting, please consider subscribing, leaving a review or a rating, or sharing it with someone you think might appreciate it. Take good care of the people that you love. Cheers, and we will see you out there.
Outro:I pledge to myself on this day to try to be kind every way. To every person, big and small, I will help them if they fall. Thank you for listening to the podcast.
