It’s hard to avoid discussions about AI these days. From smart home assistants to complex analytics systems in businesses, AI seems to have reached almost every corner of our lives. But how does this groundbreaking AI technology fit into the world of air traffic control, an arena where human judgment, skill, and safety are paramount?
AI technology has matured from an abstract concept to an applicable technology that we use in our everyday lives, says Carl Westin, researcher at the Department of Engineering and Natural Sciences at Linköping University.
In his research, he notes that air traffic control can have a safer and more user-friendly future thanks to AI systems that learn user behaviors and preferences.
– For example, AI can adapt its recommendations to an air traffic controller's personality and behavior. If two aircraft are on a collision course, AI can adapt the recommendations to the air traffic controller's preferences, such as separation distance and what type of action is needed.
Air traffic control is complex and requires extreme precision. Every movement in the airspace must be carefully considered, and there is no room for error.
“The most important task of the air traffic controller is to help pilots avoid flying into things. AI can be seen as an extension of the air traffic controller's capabilities, a tool that can identify and resolve conflicts faster and more effectively,” says Carl Westin.
Just like Spotify recommends music based on listening history, AI can recommend decisions or actions for air traffic controllers based on their past behaviors.
“By creating AI that understands and respects the unique characteristics of air traffic controllers, we can develop technology that truly improves safety and efficiency in air traffic control,” says Carl Westin.
The risks, he points out, are that only difficult and complex tasks can be left to the air traffic controller when AI fails, that the pleasure of the profession disappears when the air traffic controller becomes a passive observer, and that AI can be difficult to understand at the same time as the air traffic controller is expected to cooperate with and be responsible for the AI systems.
– AI and air traffic controllers should be considered colleagues rather than AI replacing air traffic controllers.
How does an air traffic controller's decision-making process differ from that of an AI?
– AI and humans differ in how they process information and make decisions. AI can handle huge amounts of data and does not forget details. Humans, on the other hand, have a limited memory capacity and make decisions based on a combination of current perception, past experiences, intuition and emotions.
Where AI is systematic in its decisions, humans often use a more heuristic approach. AI can help humans discover patterns in data that we might miss, while humans can help AI better understand and interact with us by teaching it about human behavior, needs, and emotions.
Carl Westin highlights the importance of understanding and respecting the unique needs and characteristics of air traffic controllers. But the question is how do you go about designing AI systems that are both innovative and consider human well-being and safety in air traffic control?
– There are many theories, design strategies and methods to use. We should think innovatively and explore what we can achieve technically. A classic example is the iPad. If we had only started from people's needs today, we might not have had the iPad.
Carl Westin falls silent for a moment – and then continues.
– That's why visionaries are needed and we must give space to crazy ideas and innovative ideas without limitations. But especially in high-risk environments, it is important that we reduce the effects of technical errors. During the design process, we need to ensure that the technology works and build experience on how it works and is received in a workplace before it is introduced. We can do this through simulations, rigorous testing, and evaluations with end users.
