Air traffic controllers need help. We don’t yet know what caused the recent Japan Haneda airport runway collision earlier this month, but there is evidence, both abroad and in the U.S., that air traffic control is stretched too thin. And that could have disastrous consequences on planes’ and passengers’ safety.

Last month, the New York Times reported that air traffic controllers have become “an exhausted and demoralized workforce that is increasingly prone to making dangerous mistakes.” While there are more flights, planes, and passengers than ever before, the number of U.S. air traffic controllers has trended downward, from about 14,750 in 2013 to around 10,700 in 2023. We’re flying more, too. Air traffic controllers moved 825,322,062 passengers in 2013, compared to 877,473,273 passengers in 2023 within 5.3 million square miles of domestic airspace and 24.1 million square miles of international oceanic airspace. But the problems go beyond a staffing shortage: The Times piece details issues with mold, faulty equipment, broken elevators, and bug-infested and deteriorating buildings, making air traffic controllers’ work even more challenging. To date, the U.S. airspace has hung on in relative safety, but it’s obvious the infrastructure is badly in need of an update.

That push for more effective air traffic is made all the more urgent given that, thanks to Silicon Valley, our dependence on the service is only going to increase. The tech elite’s plans for flying cars and robo-taxis, alongside more advanced drone fleets, portend a future sky that is clogged with individual craft (many autonomous or unmanned) all going to different destinations. When there are multiple flying objects from multiple vendors with multiple controls, the skies will be even more crowded, with no easy way to direct traffic. This will impact both international and regional airports, too.

The default solution in business these days is to try to employ artificial intelligence and other computer technology (rather than human labor) to solve complex human problems, but AI cannot yet solve air traffic control. Privatized (and potentially unreliable) software isn’t market-ready yet and there are a lot of regulatory issues about trusting (and turning over) mostly public infrastructure to private technology companies, too. And AI shouldn’t control air traffic. The risks are too great. But that’s not to say we have to eschew automation entirely. Even low-level automation such as working elevators, and coffee machines that are maintained regularly, would probably go a long way to help.

An anonymous airline pilot (with a background in economics) told me that, “ATC staff hasn’t had it very good since the controller strike under Reagan . . . The rise in near collisions of late has been mostly attributed to controller staffing, stress, and fatigue. Many quit because they’re tired after years of mandatory overtime.”

Air traffic controllers are mostly Federal Government employees and the infrastructure problems in the Control Towers are also those of a slow-rolling parsimonious bureaucracy, subject to varying agendas to limit government spending—without considering consequences or context. When applied to people with desk jobs, this cultural practice might make fiscal sense, but air traffic controllers are high-stakes thoroughbreds, who are responsible for multiple lives—like airline pilots and doctors—and each of those professionals require support (and down time) to function properly in their jobs. To its credit, the FAA did announce this week a panel to study air traffic controller’s fatigue, but it is unknown how long that will take, or if it intends to include the infrastructure complaints.

Adaptation is what we do to accommodate unfavorable circumstances, by adjusting our needs to work around obstacles. Each adjustment adds stress, which compounds. Many air traffic controllers are having to adapt to more than what they need to do their jobs. This added stress increases cognitive load, which fragments attention—the core competency of air traffic controllers. The issue is the air traffic along with the conditions to which air traffic controllers must adapt to in order to be able to control it.

Modest innovation in the form of small physical improvements could probably help air traffic controllers in the short-term, and thus, air traffic control considerably—and not cost as much, nor be as risky as trying to automate air traffic control right now..

Small changes, relatively speaking, that are funded immediately would help, and be better than losing even a single aircraft with passengers to a collision. These improvements would help attract new workers, too. Broken elevators could be fixed. Insects could be eradicated. Coffee machines could be automated and functional. Regular maintenance could be done to ensure that basic necessities stay in good repair in the Towers. Mold could be eradicated. Proper temperature with cooling and heating could be installed as well. Monitors that give air traffic controllers a consistent view as they go down the hall or to the break room “just in case” could be installed. Counselors could visit weekly or biweekly.

These basic working condition improvements will reduce the number of extra adaptations that the air traffic controllers are having to increasingly make on a daily basis. Start there. Treat the Control Tower like the critical part of aviation success that it is—and maintain it as one would an aircraft—because it is that important. Same with air traffic controllers, who are just as valuable to the aviation ecosystem as our pilots.

Innovation doesn’t always require “disruption” or new computer technology. Sometimes the most radical improvement is to mend a legacy system. Repairing air traffic controllers’ infrastructure is a small investment, a comparatively faster and easier fix. It’s one we owe them—for their well-being, and ours.