In 1869, 42-year-old Mary Ward made industrial history in a gruesome way. Thrown from her cousins’ experimental steam-powered automobile in Ireland, she became the world’s first fatal car-crash victim.
This May, in Florida, 40-year-old Joshua Brown joined her, becoming the world’s first self-driving car fatality. Brown’s death has caused people to question whether self-driving cars are safe.
Answer: They will be — as long as we understand our own human limitations.
Brown, a former Navy SEAL, was an early adopter. He bought a $70,000 Tesla Model S and posted enthusiastic videos about it.
In one video, a truck nearly sideswipes “Tessy,” as he calls the car, as he drives down a highway. Brown says that he “actually wasn’t watching in that direction,” but Tessy easily avoids the truck, neatly swerving onto the shoulder.
A month later, Tessy messed up. Tesla reports that “neither autopilot nor the driver noticed the white side of [a] tractor-trailer against a brightly lit sky, so the brake was not applied” as the truck driver turned in front of Brown’s car.
Brown, or the car, drove into the truck, and Brown died.
The New York Times said that the death puts “the belief that computers can operate a vehicle more safely than human drivers . . . in question.”
Not exactly. The tragedy instead reminds us of what we already know: Computers don’t eliminate human error.
Brown himself made one critical mistake: using the car in a way it wasn’t ready to be used. Tesla tells its customers that drivers must “remain engaged and aware when Autosteer is enabled. Drivers must keep their hands on the steering wheel.”
That’s because the technology is still experimental. Tesla’s self-driving technology is good at maintaining speed relative to other cars, and parking.
But as Brown noted in other videos, it’s not good — yet — at doing certain things. “We’re filming this just so you can see scenarios where the car does not do well,” he says, as the car veers over a double-yellow line.
But it isn’t entirely, or even mostly, Brown’s fault that he apparently stopped paying attention.
Humans have a hard time paying close attention when nothing interesting is happening. That’s why you’re more likely to crash on a sunny day on an empty stretch of highway than on a rainy day in the big city. When nothing demands your attention, you zone out.
It may be Tesla, then, that was unwise, in asking people to do the impossible: Keep watch when the car is telling them they don’t really have to.
As driverless-car technology progresses, it will probably do better than humans in most circumstances. The problem is that all circumstances are not most circumstances.
Google cars have piloted themselves more than 1.5 million miles (with human drivers aboard). Its technology has caused just one minor crash.
But Americans travel 3.06 trillion miles a year. Google is nowhere close to having absorbed every possible real-life experience.
As technology helps us avoid the most obvious crashes, the crashes that do happen will get weirder — making headlines.
That means that even if driverless cars become twice as safe as human-driven cars — as federal regulators hope — you’re more likely to hear about the exceptions.
We don’t think much about the 38,300 Americans who died last year in auto crashes because we expect people to do dumb things that lead to their or others’ deaths.
Last week, for example, the death of 84-year-old Gerald Walpin as he crossed an Upper East Side street with the light didn’t make much news — because these tragedies happen regularly. Almost every American remembers a relative or friend, or friend of a friend, who died in a car wreck.
If driverless technology someday were to cut our annual auto deaths in half, saving 19,000 lives a year, we’d hear constantly about when the technology didn’t work.
We’d want to be back in control, even if that control makes us less safe.
It’s just how like most people are still more comfortable behind the wheel rather than strapped into an airplane seat. Even though they know by now that flying is safer than driving, they like having the illusion of being in charge of their own fate.
More people will die in self-driving cars — just as tens of millions of people, worldwide, have died in automobile deaths.
Still, carmakers and regulators gradually get better at making people safe — as we learn from the deaths of the people we couldn’t protect.
Nicole Gelinas is a contributing editor to the Manhattan Institute’s City Journal.