Balancing AI and human factors in ATM

“AI is a tool,” said Oren Etzioni, technical director of the Allen Institute for Artificial Intelligence. “The choice about how it gets deployed is ours.”

We know artificial intelligence (AI) has the potential to streamline processes and save valuable time, money, and resources in the air traffic management (ATM) industry. For example, a study by EUROCONTROL revealed that it could boost predictability and efficiency by between 20 and 30%. A recent SESAR event explored the use of AI in taxiway inspection, runway monitoring, flow management, dynamic sectorization, and more.

But what we often overlook is how we can use AI to support human factors in ATMs—not to enable more work or cut costs but to improve the safety and well-being of controllers, aircrew, and passengers.

In this article, we will explore human factors in ATM. We’ll discuss how AI can help, its limitations, and how to strike the right balance to deliver a ‘service first, safety always’ approach.

An introduction to human factors in ATM

Human Factors is the term used to describe elements of the human condition that can impact an air traffic controller’s (ATCO) ability to perform their job safely and efficiently. Factors like stress, fatigue, ergonomics, workload, and training can impact professionals in every industry. But for ATCOs, they can lead to safety risks.

As an industry, we must consider human factors and build a culture that prioritizes safety and minimizes risk. It is vital to put the well-being of ATCOs and other ATM professionals at the heart of decision-making. From providing comprehensive training and ergonomic control rooms to promoting a work-life balance, there are many steps we take to create a safe and comfortable environment for real people with real lives.

As individuals, we tend to push this aside saying it’s the employer’s job to take care of us. It’s only recently that more frontline workers have started taking more care for themselves and start paying more attention to stress and fatigue.

But humans are, by nature, unpredictable. Errors are always a possibility. This raises the question, then: to what extent can ATCOs rely on AI to support human factors?

Let’s explore.

Can AI prevent errors caused by human factors?

Unlike humans, AI can’t get stressed. AI doesn’t get tired, uncomfortable, or overwhelmed. AI is fast and always available.

So, how can it help?

• Planning and decision-making: AI can help planner controllers and executive controllers decide what actions should be given to the air traffic controllers. It can access all the data it needs to help decide what capacity is required for the day ahead and plan when to open and close sectors.
• Data checking and interpretation: Controllers have to check data frequently. This includes Mid-term Conflict Detection (MTCD), routes, levels and converging tracks. If AI could take on some of these checks, it could reduce controllers’ workload.
• Fatigue and alertness monitoring: Using biodata, AI could check the stress or fatigue levels of air traffic controllers and adapt rosters, sleeping plans, or breaks during a shift. This would rely on controllers agreeing to allow a company to monitor and analyze their biodata, such as heart rate and blood pressure.

Of course, there are limitations to AI’s capabilities, including:

• Privacy concerns: From data collection and sharing to surveillance and consent, AI privacy concerns are a significant factor for the ATM industry. Businesses must put the right measures in place to combat these issues as much as possible.
• Psychological safety: New technology promises exciting new opportunities but also comes with potential risks to psychological safety in ATM. Over-reliance on technology increased complexity and ambiguity, and even reduced human interaction all impact the well-being of air traffic controllers.
• Just culture and ultimate responsibility: The ATM industry must foster a just culture to maintain high safety and efficiency standards. Prioritizing accountability, fostering communication, and empowering individuals to learn and improve continuously will always be a core part of our culture that AI is not able to replace. Similarly, a human will always have ultimate responsibility, particularly if issues arise.
• Lack of trust: A study by The Conversation found that less than half of us trust AI at work. A study by UBS found that 61% of people would not fly knowing that their plane is not operated by a pilot. There’s still a long way to go before AI is universally trusted (by both its end users and those who deploy it).

Can AI take humans out of the equation in ATM?

With the positives and pitfalls of using AI to support HF, it’s clear there is a lot to consider. ANSPs can benefit from AI use in their daily lives. But how far should we allow AI to actually do a controller’s job?

Simply put, AI can’t replace humans' role in air traffic management. It learns by analyzing data patterns, then using that to make decisions or predictions. But there’s nothing predictable about our skies. For example, AI can’t tell you how to handle the rapid and intense wind changes that come with a thunderstorm. It can’t strategize the next steps if a fast-moving military aircraft (or a slow-moving student pilot) flies off course into controlled airspace. Of course, there will be patterns sometimes — but there might also be wild variations, such as different fuel levels, pilot skill, loading, etc.

When asked if AI could replace humans, we found that even ChatGPT agreed that it’s unlikely that AI will be able to fully replace air traffic controllers in the near future.

We are confident that AI use will continue to grow and evolve. And indeed, we may reach a point where humans are only involved in the design and control of AI systems.

But for now and the foreseeable future, humans play an irreplaceable role in ATM. AI can’t replace ATCOs. But it can help them do their complex job safely and effectively. There many ways that AI can do this:

• Automating routine tasks
• Dealing with repetitive tasks
• Offering suggestions

But leave the human brain—which works brilliantly well when it is not fatigued or stressed—to deal with the new and the unexpected. Let’s continue to design and build an environment that caters to humans' diverse needs and prioritizes safety over maximum efficiency. As ever, balance is the key to getting it right.

(Image credit: We used HubSpot's AI tool to generate a featured image for this article as an experiment. It seemed appropriate!)