September 10, 2017
Recently, I was invited to visit the Deutsche Flugsicherung (DFS; German Air Traffic Control), to exchange thoughts on future business scenarios, the rise of drones, and innovation for the year 2040 and beyond.
In preparation for the meeting, I started thinking about how a service like this is usually fully hidden from most of the users who benefit from it. I mean, how often do you think about the tower and its controllers while at the airport, or during your flights? Where is the user experience of such services? Yes, it is mostly for pilots, and so forth, but if you think a bit further, the experience that we as privates enjoy is arriving at our destination in one piece (or on time ;-)). Not much interaction needed, right?
While strolling through the DFS campus on a perfect summer day, I was shown the different buildings for all the special activities. I was given a crash course in the realities of the work of air traffic controllers. Learning that only the best are selected, before being put through intensive, long and continuously updated training programs.
Air traffic controllers must make lightning fast decisions, according to long learned routines. I got the impression that this is almost like executing algorithms. Over the course of their work life, their way of interacting with challenges gets so entrenched that they develop a sort of super power for making incredibly quick decisions based on predefined standards.
So, how much does the work of air traffic controllers differ from what is increasingly being undertaken by modern bots? Wouldn’t you think the question is when, not if, AI/machine learning will become a substitute for their work? Personally, I think it is exactly this type of work that an AI could do better than a human. We already have self-driving trains, and cars and airplanes are on the way. Most railroad switches are controlled by subroutines, so why shouldn’t the skies be?
Thinking about the future of work, AI, and humanity itself, I have three thoughts for how we need to design the future of air traffic control intelligently (not just cut the current approach and implement new technological solutions):
1. For at least one or two generations, we need to build trust in AI and its fail proof capabilities. For the moment, it just feels good knowing there is a human being on the other side watching over you.
2. The work that air traffic controllers do is amazing, and almost superhuman. It is a craft that takes years to master. If it vanishes with the blind application of a new distruptive technology, without the realization that we will still need human inputs for years to come, it would take years to redevelop the human institutions we currently have, e.g. the facilities from training to executing. Think what would happen in the case of a glitch in an automated system, even for a minute or two, you would need to have these highly trained people be able to jump in immediately—this reality also holds for self-driving cars and planes.
3. It is a valuable and human friendly job. By human friendly I mean challenging, satisfying, and relevant. In short: it’s the type of work we may want to keep, because it utilizes humans at their best.
So, how should we innovate a perfectly operating human system (there have been no air traffic control accidents in Germany for decades)? Where should we set priorities? Decreasing costs, or driving efforts towards new technologies alone?
Although the German air traffic control is owned by the German Government, they know the future lies in new business models and new services. Innovation will increase revenues coming from alternatives approaches to controlling the skies. Changes will happen, and after my visit I am very confident that these can be for the better.