Since March, when an autonomous vehicle killed a pedestrian in Arizona, forecasts for AVs have been decidedly less optimistic. But autonomous vehicle promoters are undeterred. AI entrepreneur Andrew Ng contends that self-driving cars will be safe for pedestrians when walkers and cyclists conform to their limitations. “What we tell people is, ‘Please be lawful and please be considerate,’” he told Bloomberg.
Has Mr Ng ever walked for as much as an hour in a city? If so, he should realize that consideration of pedestrians' needs—and motorists' compliance with the few laws that protect pedestrians—are so deficient that any pedestrian who values their time (as drivers do) must improvise. And in fact, such improvisation can even make pedestrians' journeys safer.
To be fair, Mr. Ng's mistake is a common one. From a driver's point of view, pedestrians' behavior may appear erratic, lawless, and even suicidal. The solution, then, is to train pedestrians to do better, or to restrict them. In actuality, most pedestrians are much smarter than the dumb systems that are intended to control them—far smarter than signals, and even smarter than self-driving cars. A pedestrian who is on the right side of the street and wants to turn left at the next intersection may cross early, at mid-block. What may appear to some as selfish and dangerous rule-breaking may actually be safer and less disruptive to vehicular traffic. In one study of pedestrians aged 65 or older, for example, researchers found that the risk of a pedestrian-motor vehicle collision was 2.1-fold greater at sites with marked crosswalks, particularly those with no traffic signal or stop sign.
In the 1970s, research teams led by William H. Whyte filmed pedestrians on busy sidewalks as they walked around New York City. Walkers filtered past each other with extraordinary efficiency, coming within inches of each other but almost never touching. Such performance requires human intelligence. No one would propose putting pedestrians on autonomous Segways as a way to keep them from colliding with each other. Either traffic would slow almost to a stop, or collisions would increase.
Autonomous vehicles are frequently touted as safer and more efficient alternatives to conventional cars. But if safety and efficiency are indeed primary values, then cities should not deter walking by making it harder, but invite more walking by making it easier. This would entail, among many other things, urging drivers to be more lawful and considerate about pedestrians.
Indeed, the success of self-driving cars depends upon a rise in walking as a practical means of getting around. AVs cannot deliver on their own promises of safety and efficiency if they deter walking. Safety matters because we care about human health. Sedentary living is already inducing health conditions such as heart disease, obesity, and diabetes; public health can only worsen if an autonomous future compels people to ride in cars for every mobility need. And self-driving cars will not be more efficient if we negate their per-mile efficiency benefits by increasing the total miles each person spends in the car.
Smart traffic signals can increase streets' vehicle capacity by shepherding cars safely through intersections without compelling them to stop. But we don't yet know how they'd work for cyclists and pedestrians, those who make the most efficient use of street space, use the least energy, and cause the least danger to others. Either they will have to be equipped with devices that incorporate them into signal systems, or smart signal systems will have to get much better at detecting and tracking them. The social and technical complications of either alternative are substantial.
In the meantime, we have access to innumerable low-tech possibilities. Traffic calming—design features that slow vehicles down—can make select streets much safer for everyone. Planners in the Netherlands, for example, apply humans' smartness, instead of trying to suppress it, by designating certain streets "bicycle streets"; though drivers can still use them as "guests," they must defer to cyclists. By conventional U.S. standards, this method is considered dangerous because it depends too much on human judgment. But the traffic safety record in the Netherlands should compel us to reconsider. In 2013, there were 3.4 road traffic deaths per 100,000 people in the Netherlands; the figure for the U.S. was 10.6. Extravagant promises about the driverless future too often distract us from implementing effective, inexpensive, low-tech improvements today.
To succeed on their own terms, AV developers will have to do much better by pedestrians. Bloomberg reports that AV developers are looking into "distinctive sounds—much like the beeping noise large vehicles make when reversing—to help ensure safe interactions between humans and autonomous vehicles." This technique, in the form of the klaxon or car horn, is well over a century old. Honking was then attacked as a public health menace. Today, such noises can only make the walking environment less inviting, relative to the quiet, climate-controlled interior of a vehicle. For pedestrians who can't afford this alternative, walking will be less pleasant than ever.
Too often we hear extravagant promises for self-driving cars, or warnings that "the AV future is coming; we have to get ready." But the saw does not use the carpenter; the carpenter uses the saw. AVs are a tool. We humans have to decide if and how we want to use them. Despite the public relations, AVs will not, on their own, deliver safety or efficiency. We have to put them to work for the purposes of our choosing.
WIRED Opinion publishes pieces written by outside contributors and represents a wide range of viewpoints. Read more opinions here.