On Wednesday morning, during Google’s annual meeting, a shareholder named John M. Simpson stood up to question the company’s top executives about its self-driving car program. They were not friendly questions.
Simpson, 67, works for a nonprofit called Consumer Watchdog, where he directs its Privacy Project. In recent years, he has focused largely on Google, which, he told me, he hopes to prod into “being more respectful of people’s privacy when they do business.” Owning Google stock allows him to ask questions at the annual meeting.
In the run-up to this week’s meeting, Simpson issued a string of press releases critical of Google’s self-driving vehicles. He feared that Google would collect data from car owners, stripping away even more of people’s privacy. He noted that Google’s “autonomous cars,” as they’re called, have been involved in 11 accidents (two recent fender-benders brings it up to 13). He listed what he said were the technology’s flaws: for instance, that it can’t make out hand signals from a driver in another car.
Finally, Simpson noted that Google — and Google alone — envisions cars that have no steering wheels or brakes, cars where everyone is a passenger. Simpson views this as Google’s hubris, pointing out that other car companies view self-driving technology as a complement, not replacement, for the driver. “We think there always needs to be the ability of a human to take over if need be,” he told me. Having looked into it more closely, I’ve come to the opposite conclusion.
Google’s effort to build a self-driving car is part of the division called Google X, led by a scientist with the too perfect name of Astro Teller. The goal of Google X is to attempt “moonshots” — efforts that require a radical solution that, if they succeed, would solve a huge problem (while making a nice return for Google, of course). The big problem self-driving cars could help solve, said Teller in a recent speech, is the “1.2 million people who die every year in car accidents.”
During the six years Google has been working on self-driving technology, its cars have been taught to understand how to traverse the roads. With their combination of robotics, sensors and computing power, they know how to stop at a stop sign, look for oncoming pedestrians, change lanes, get on the freeway and anticipate all the various problems that drivers face.
Using retrofitted Lexuses, Google has driven a million miles autonomously. More recently, it has built several dozen small cars without steering wheels and brakes and is ready to test them in the streets of Mountain View, Calif. (though the State of California is insisting that Google add a steering wheel and brakes to the cars it sends out for this experiment).
It’s true that Google is alone in envisioning a world of completely driverless cars, while other car companies see self-driving technology as merely an extra feature that can be turned off. Google’s conclusion is not the result of hubris, however. Unlike its new cars, the retrofitted Lexuses also allow for human driving.
Alain Kornhauser, a self-driving car expert at Princeton University, pointed out to me that when the auto companies install autonomous features to aid drivers, it won’t be the humans who escape accidents by taking over from the technology — which is what Simpson assumes. Rather, the technology will step in to override human error.
Google notes that in every accident its cars have been involved in, all of them minor, the self-driving cars have never been at fault — except on the one occasion when a Google driver took the controls. And all the “flaws” Simpson notes are things that Google has either solved or is in the process of solving.
At the annual meeting, Simpson asked Google if it would pledge not to use any customer data it gathers from driverless cars for marketing purposes. David Drummond, the company’s general counsel, ducked the question, saying it was too early to make any such pledge. Simpson also asked Google to release the accident reports. In truth, Google has released plenty of information about the accidents, and on Friday began issuing monthly reports that include descriptions of accidents.
Simpson and other consumer advocates are right to press Google — and all the big tech companies — on privacy issues. The profligate use of our data has become a big concern for many Americans. But on the question of whether Google should be promoting completely autonomous cars, he couldn’t be more wrong. The sooner they are a reality, the safer we’ll all be.