Keep the Driver in Driverless Cars

Let’s take a few cues from the aviation industry.

As a law professor who studies mobility, I spend many waking hours thinking about fully automated vehicles, those cars that drive themselves without any need for a human operator. As a true believer in this technology, I think the widespread deployment of these vehicles will get more people home safely, give commuters their time back, and reduce the amount of carbon dioxide in our environment. However, lately, I worry whether this will ever happen. A recent Pew study showed that only 26% of the U.S. public believes that self-driving cars are a good thing for society, which is an all time low. To the public, the deployment of automated vehicle technology seems rushed, inconvenient, and dangerous. The more I hear from people outside of the mobility industry, the more it feels like the industry is focused on creating a solution in search of a problem.  

Yet a problem exists. Nearly 40,000 people will die on the road this year in the U.S. alone, with human error as a causal factor in the majority of those crashes. We will spend hundreds of billions of dollars repairing the damage to people and property caused by vehicular collisions. People will waste weeks of their lives in unnecessary traffic. Knowing this, I wonder why there is a disconnect between society’s desire to address this problem and its receptiveness to the solution of automated driving technology. Developing the technology is hard, but why does getting people to believe in the promise of the technology seem just as hard? In pondering this question, I turned to a mobility industry that incorporates a high degree of automation, yet still enjoys the public trust: aviation, the safest mode of transportation in existence.

Automated aviation technology, referred to within the aviation industry as autoflight, has been in commercial use for decades. Pilots welcome a certain level of automation in the cockpit because, when properly used, it reduces their workload and eliminates the mundane and repetitive tasks involved in flying. On any given commercial flight, the pilots are managing the autoflight system more than they are manually flying the plane. This is not to say that the pilots are disengaged. The autoflight system requires constant monitoring and adjustment, and if it encounters something that it cannot handle, the pilot must take over. For these reasons, pilots are trained to maintain situational awareness even while the autoflight system is in command of the plane. Along with other advances, this increased use of automation has made commercial aviation safer than it has ever been. And the public understands this; Americans rate flying as the safest mode of transportation. 

Years ago, an Airbus executive confirmed that the technology already exists to allow for safe commercial flight without a pilot at all. Indeed, private air taxi companies are currently testing pilotless passenger operations and the U.S. military has been utilizing pilotless aircraft for years. Yet today, most commercial flights employ at least two pilots and sometimes even four pilots for exceptionally long flights. While this may change in the long term future, no major airline has seriously sought to deploy pilotless commercial flights, and U.S. government regulators have made it clear that the concept is a nonstarter.

Such restraint feels antithetical to the culture of the tech scene today where startups and legacy automakers alike are racing not only to develop automated driving technology, but to remove the human driver from the vehicle. After speaking with a variety of pilots and administrators, my belief is that the aviation industry is more responsive to its reliance on public trust. While today’s commercial aviation enjoys the highest safety ranking from the public, 81% of Americans would not feel comfortable in a pilotless plane, even if the airfare were significantly cheaper. The aviation industry understands that it cannot take the chance of losing the public confidence it enjoys. Accordingly, airlines have chosen to develop and deploy a high level of automation to make flying safer, but have not focused on eliminating the pilots.

The automated vehicle industry similarly needs the public’s trust. The industry obviously will need consumers to purchase and ride in the vehicles once they are widely available. Additionally, the widespread deployment of fully automated vehicles is going to require billions in public investment for the necessary infrastructure. The federal government will need to focus its attention on automated vehicle regulations while state and local governments will have to create new laws to allow for automated vehicle operation and avoid passing legislation against automated driving, which we have already begun to see around the country.

Despite needing public support, the automated vehicle industry does not have it. Currently, Americans feel the same way about vehicles without drivers as they do about planes without pilots. According to a recent survey, 70% of Americans would not feel comfortable in a self-driving car. Nearly the same amount would not even feel comfortable sharing the road with such vehicles. And 90% believe self-driving cars should always have a human driver in the car in case of emergencies. Yet the automated vehicle industry is adamant not only about removing human drivers from vehicles in the future, but today as it develops and tests the new technology on public streets. The leaders in automated driving are not heeding the sentiment from the public the way that the aviation industry leaders have, and, in my opinion, this is a mistake.

In many ways, the aviation industry provides us with a valuable roadmap to deploy automated technology in a safe, accepted, and successful manner. We should follow aviation’s example and think twice about removing the human driver from even the most highly automated vehicles, at least until we get public buy-in. We need to introduce the technology in an intentional and familiar manner to give society time to develop a sufficient amount of comfort and confidence in it. 

And keeping the human drivers would not be for show, especially now in the testing phase. Just as pilots routinely need to take control from the autoflight system, human drivers would be a valuable failsafe. Yes, pilots are better trained than your average driver. However, whether in planes or in cars, automation can struggle in situations that are unexpected, rare, or complex. In some of those failures, a human driver, despite our flaws, would be better than no driver at all. After all, even a newly licensed human driver knows not to block ambulances or stop in the middle of an intersection, things we have witnessed today’s automated vehicles do.

I look forward to a day when cars can drive themselves in every scenario or condition, without the need for human intervention. Beyond safety, fully automated vehicles have a number of expected benefits for society. I am not advocating that we abandon our quest for fully automated driving. Instead, let’s abandon the quest to be the first to get to a car without a driver in it. In trying to reduce tragedy caused by human error, let the automation focus on removing the error, not the humans.