By Lisa Turner, EAA Lifetime 509911
This piece originally ran in Lisa’s Airworthy column in the May 2024 issue of EAA Sport Aviation magazine.
Discussions of artificial intelligence have filled the news so you may feel overwhelmed. Much of what we are hearing about AI revolves around generative AI, or large language models (ChatGPT, Gemini, Copilot). These systems have been trained using billions of data points and then wrapped in enough computing power to deliver analysis based on what they know and what you ask them to do.
Heralded as the next wonderful thing, we’re also being warned of its possible dangers. At first, it appears simple, like Garmin’s Autoland. Smart systems can analyze huge amounts of information and act based on programmed algorithms. AI takes it further by extrapolating from the data to choose the best course of action.
It’s a remarkable and welcome innovation. For all its simplicity for the user — press a button — the unit and all that goes into it is complex. As it helps us make decisions, it can also scare us for the same reason. It can make its own decisions.
If you drive a car or truck with “advanced driver assistance” features, you know it’s limited in “intelligent” responses. You might experience some responses that don’t deliver what you were expecting. If the car in front of you gets into the left turn lane with you reasonably close behind, your car may think the turning vehicle is stopping and will react by applying emergency braking. If the smart system wasn’t on, you’d know what the other car was doing, and you’d smoothly pass to their right, saving the brakes for another day.
While the driver assistance features may be helpful — as in stop-and-go driving — it’s not intelligent; it’s just a programmed response. Even Tesla’s Autopilot gets it wrong from time to time — which can, and has, resulted in disastrous consequences.
Intelligent driver assistance systems in cars and trucks are developing, but they are not fully here yet. We tend to have faith in the systems and often rely on them so much that we fail to give full attention to what is developing around us. We fiddle with screens and settings and have a hard time multitasking. I know that piloting an aircraft is the essence of multitasking, so having systems to help, from autopilot to intuitive map interfaces, can be great. But so far, these helpers are not intelligent. Even Autoland, as amazing as it is, is built on a precise set of algorithms using a dataset.
My point here is that until we can get into our airplane, punch in our destination, and sit back reading the newspaper, we will need to be active pilots. In our wonderful small airplanes or large vintage beasts, the thought of giving up any of the flying to AI goes against the grain. The whole point of flying them is the enjoyment of being a pilot.
Just as we announce who “has the airplane,” we need to be active participants and not allow ourselves to become distracted. This is hard to do when we have multiple fancy gadgets in our airplanes (and in our cars). Artificial intelligence will be welcomed when it reduces our workload and error possibilities and does this correctly every time. Right now, we are more distracted than ever before — in our cars and in our airplanes — with multiple gadgets and displays. Distracted piloting and distracted driving are two large accident categories, and they carry over into distracted maintenance.
Where does AI fit in our world of small aircraft maintenance? While AI is arriving for large aircraft and aircraft operators as smart data-driven software systems, these are mostly unaffordable for those of us with a J-3 Cub, Stearman, or homebuilt.
Smart systems for large operators include predictive maintenance using data from sensors on the airplane, autonomous inspections using drones on preset paths to examine aircraft component areas and flag anomalies, documentation and scheduling software, and performance monitoring on a large scale.
How can we “little guys” use these concepts to make maintenance easier? While these are on a vastly different scale than the “big guys,” we can set up systems to help us manage the health of our airplanes. That’s the good news. The bad news is that these systems take some work on our part.
While I’d love to say that AI will give us a solution for our small aircraft maintenance where we can stand back and let robots do the work, we’re not there yet and won’t be there soon. Where we will really reap the benefits of smart systems will be in the tools we employ in the processes.
Performance monitoring. Attention to temperatures and pressures is an important function in the air that will help us recognize impending problems with the engine and other systems. This is a perfect application for smart systems. If you have flown with an array of instrumentation in a glass panel, you have seen the amazing functionality it can deliver. It won’t be long before AI will integrate all your in-flight data in alerts for action both in the air and on the ground.
Data collection and logging. Systems that log data in the air are a great help in diagnosing impending mechanical failures and predicting component life. Combining engine monitor data with performance data and then overlaying AI will give the pilot and the mechanic helpful advice on keeping things in top shape. While this may not help you with your Stearman right now, affordable electronics that can integrate with existing systems are in our future.
Error proofing. I’ve saved the best for last: checklists. You’ve seen opinions recently on the use of checklists in the aviation world. From Steve Krog’s great article in this year’s February edition of EAA Sport Aviation to other magazine commentators, it’s receiving renewed attention because of its indispensability. What I keep hearing is that checklists won’t help you if you don’t use them. Of course, we know this. In all of our early training as pilots, we were surrounded by the rigor of using checklists — from our preflight and engine start to our landing and shutdown.
Where smart systems can help is protecting us against human error. It’s normal to err and forget. Why not give these tasks to a smart system? I hope that, eventually, we will all be able to afford smart systems that can recognize when we miss an item on the list. We’re still seeing instances where pilots forget landing gear, even with the advanced safety systems currently in place.
An example of checklist improvement is the ForeFlight checklist app that has a single-tap emergency button that immediately takes you to your aircraft’s emergency checklist procedures, saving you time and sparing you the frazzled and frantic search for the right page in your lap as you try to keep flying the airplane.
While there are many good talking/voice-responsive piloting checklists, there are few designed for a single owner who is wrenching on their airplane. Rather than spending a lot of time here suggesting what you should use, I will tell you that the time spent on researching a voice checklist app that you can modify to include your maintenance checklist items will be time well spent.
It may feel like overkill to use this level of preparation for your maintenance work — “getting ready to get ready” — but it will improve your ability to save time and not forget things. Who hasn’t started a job only to stop what you’re doing to go retrieve a tool? While AI may not help us on this one until we have our robot fetcher, having a standardized process will. Before I begin an annual or condition inspection, there’s an entire checklist I go through that allows me to get right down to business and not have distractions or interruptions.
So, where does AI leave us as small aircraft owners and mechanics? While we can employ smart systems to automate some of our piloting — from autopilot to engine parameters to gear-up warnings — we face more work on the ground, making sure we’re doing the right things. The less distracted we are, the better we will perform.
Use smart systems for procedures and checklists and keep refining them to fit your specific situation. If you decide to use only the written lists for your airplane, take the time to review these and add what you think will make them better, including printing them in a larger font with selective color with labeled tabs.
Get training on complex systems and what to do if something goes wrong. We already know that pilot error contributes to accidents, but machinery and electronics can also fail. If we understand the methods of operation, we can react to a failure in the system.
If you have an electronic flight instrument system/glass panel in your airplane, read the manuals thoroughly and make sure the sensors and settings are calibrated for optimum performance. On homebuilts, I encounter owners who don’t know how the electronics were set up — or if they ever were. They just want to get in the air. In certified aircraft panel retrofits, I’ve encountered situations where owners didn’t know if the systems were calibrated or not.
Reduce distractions. When flying, practice a scan that includes looking outside. Don’t allow yourself to be pulled into a rabbit hole of self-absorbed activity. When performing maintenance on your airplane, set aside dedicated quiet time. Assemble your manuals and checklists in one place, and add what the builders group or the type club has recommended for improvements. Make a preparation for maintenance checklist. There is no reason you cannot use the talking checklist apps for maintenance lists.
AI for flight or maintenance is anything that helps us handle critical tasks without distracting us and getting it right every time. It is “error-proofing” in its highest form. That’s the good news. The catch is that in the small aircraft maintenance world today, this is still you.
Lisa Turner, EAA Lifetime 509911, is a manufacturing engineer, A&P mechanic, EAA technical counselor and flight advisor, and former designated airworthiness representative. She built and flew a Pulsar XP and Kolb Mark III, and is researching her next homebuilt project. Lisa’s third book, Dream Take Flight, details her Pulsar flying adventures and life lessons. Write to Lisa at Lisa@DreamTakeFlight.com and learn more at DreamTakeFlight.com.