Future Tense

Lack of Focus

Safety drivers are supposed to be our bridge to a self-driving world. They might not have the attention spans for it.

An Uber automated vehicle taking a test-drive on the 31st Street Bridge in Pittsburgh.
An Uber automated vehicle taking a test-drive on the 31st Street Bridge in Pittsburgh. Jeff Swensen/Washington Post via Getty Images

The transition from human drivers to fully autonomous vehicles is going to take some time. While driverless vehicles are in advanced testing stages on our roadways, they may not hit streets in earnest until closer to 2025, according to some experts. In the interim, new cars will increasingly rely on semi-autonomous autopilot driving modes, and we’ll see greater numbers of autonomous vehicles with operators sitting behind the wheel. These standby drivers are important: They’re responsible for stepping in if an autonomous car isn’t reacting to a situation, documenting any anomalies in its behavior, and otherwise ensuring a successful ride. While having a human there just in case sounds comforting, it might not be as safe as it sounds.

Research conducted by Missy Cummings, a Duke University professor who studies interaction between humans and unmanned vehicles, found that humans have trouble staying focused when expected to monitor an autonomous system. Cummings studied 27 subjects in a four-hour-long driving simulation and found that, on average, their vigilance decreased after just under 21 minutes—a physiological phenomenon known as the vigilance decrement. (In other studies, it’s been documented as occurring between 10 and 30 minutes.)  After this period, driver performance worsened: That is, drivers were more likely to deviate from their lane, use their mobile phone, and cross road edge lines. Self-driving car operators typically work in eight-hour shifts, well beyond that 21-minute good attention span. This observation could foretell problems with semi-autonomous cars and trucks.

“The bottom line is that humans are terrible babysitters of automation, which is why we are in such a dangerous period of time with good, but not great, autonomous cars,” Cummings told Slate in an email.

Last week, we saw the first pedestrian death in a self-driving car accident. A 49-year-old woman died from injuries sustained when an Uber vehicle operating in autonomous mode struck her as she crossed a Tempe, Arizona, street at night. Police investigating the incident immediately called it unavoidable, but forensic engineers believe that a human at the helm may have been more likely to spot her and avoid the fatal crash.

We’ve already seen issues with attention span and semi-autonomous operation in another field—flying. But for pilots, it’s not as big of an issue as drivers on crowded roadways with numerous types of potential hazards. “In aviation, we never figured out how to get even the most disciplined pilots to stare at a computer that, most of the time, seemed to be doing just fine,” Steve Casner wrote for Slate in 2016. “We get away with this and stay safe the vast majority of the time because airplanes are small and predictable and the sky is spacious and friendly.”

Airplane pilots have three main concerns in the air—mountains, adverse weather, and other airplanes. These things are easy for computer systems to monitor well in advance of danger, giving pilots plenty of wiggle room to appropriately take action. Car operators, as the unfortunate case in Tempe shows, have only split seconds to make life-or-death decisions and take actions—and that’s if they’re fully alert and engaged with the task at hand.

There is a silver lining, however: If we know that human attention is likely to decrease after a certain time period, then we have a chance to develop autonomous systems that can counter that issue. In fact, Cummings’ team at Duke was successfully able to monitor the vigilance decrement in their research. Self-driving and semi-autonomous cars could eventually use a similar approach to detect when their emergency operator isn’t fully focused on the task at hand, and do something to draw their attention and reduce boredom.

Self-driving vehicle–makers are working to ensure their operating and environment-sensing technologies are robust, but they’re not doing as good of a job considering the human aspect of the equation, both inside and outside the vehicle. Before autonomous systems can fully take charge, we need to ensure humans relaxing behind the wheel are just as vigilant as they would be without autopiloting systems in place. Otherwise, we risk more lives as we push a nascent technology onto the road.