The robotaxi is recording me sitting in the backseat, and I am recording it. Someone in the neighboring car is recording us both.
It’s an unusually hot day in San Francisco, and I am in a self-driving car named Charcuterie, operated by Cruise. Next to me is William Riggs, a professor at the University of San Francisco who studies self-driving cars. The front seats are both empty, and the wheel silently shifts as the car maneuvers itself along a thoroughfare next to Golden Gate Park.
When I notice the stranger filming, we are stopped at a red light. Riggs rolls down his window to chat. A pleasant robotic voice chimes in and warns him to keep his hands and arms inside the vehicle.
“It’s weird!” the woman in the car says, assessing our futuristic setup from behind her phone.
“It’s totally normal and forgettable!” Riggs replies.
This is summer 2023 in San Francisco, where hundreds of self-driving taxis are crawling the roads. Two companies, Cruise (a subsidiary of General Motors) and Waymo (owned by Google’s parent company, Alphabet), have been operating their cars in the city for years. Earlier this month, California gave the green light for the two companies to expand their presence, charging for rides and operating 24/7. There have been lots of hiccups since then. One car got stuck in wet cement; several stalled during the city’s Outside Lands music festival, causing traffic jams; and, most serious, a Cruise vehicle collided with a fire truck. (After that incident, Cruise agreed to cut its operating fleet in half and has said that it’s investigating what went wrong.)
Robotaxis aren’t perfect. But neither are humans. Some 46,000 people perished on America’s roads last year, a toll that far exceeds that of most other developed countries. Behind guns, cars are the second-biggest killer of American kids. In theory, an autonomous-driving future could help make our roads safer: Computers don’t drive drunk, or get distracted by their phone, or speed. These cars are full of promise, even if the present is far more complicated. The question is how much risk—and just plain old annoyance—we’re willing to put up with on the journey to that utopian vision.
Somewhere has to serve as the proving ground for these cars. Cruise also operates in Austin, Texas, and Phoenix, Arizona, while Waymo is also in Phoenix and soon expanding to Los Angeles; both companies have tested their cars in many more cities. Still, “no other city has been at the tip of the spear like San Francisco has,” Missy Cummings, a professor at George Mason University and a longtime advocate for autonomous-vehicle safety, told me. A representative for Cruise told me the company picked San Francisco in part because it is so hard to drive in: “San Francisco exposed us to a lot of chaos—the exact kind of chaos that our technology needs.”
The backlash from certain residents has been blunt. Some have taken to putting cones on the cars’ hood in protest, which triggers their safety system and prevents them from moving. Viral videos across social media show people laughing at the cars’ incompetence when one stalls out. Driving can be unpredictable, and unusual situations that these cars haven’t experienced before can pose problems. San Francisco Fire Department Chief Jeanine Nicholson told me that having to work around the vehicles in emergency situations has been “frustrating.” She relayed stories of taxis blocking the path of fire trucks and ambulances on the scene of emergencies, and first responders having to “babysit” the vehicles so they don’t make a mistake. “I will continue to sound the alarm, no pun intended, for public safety,” she said.
Safety experts worry that the recent incidents are a harbinger of what’s to come. Phil Koopman, an engineering professor at Carnegie Mellon University, told me he thought the cars were “actually doing amazingly well … But there’s a difference between amazingly well and safe,” he added. Cruise and Waymo have defended their safety record, arguing that their vehicles have driven millions of miles without any fatalities or life-threatening injuries; spokespeople for both companies pointed me to data suggesting that their vehicles are safer than human drivers. Koopman, for his part, compared the rollout of robotaxis to running a marathon: The vehicles have had a solid first mile, but that doesn’t mean they will do well the rest of the race.
Still, sitting in these taxis, it’s easy to imagine a future in which they work well. In addition to my ride with Riggs in Charcuterie, I also took a loop of San Francisco in a Waymo car. Unlike Cruise, Waymo does not name its vehicles. It does, however, allow you to broadcast your initials in the color of your choice on the spinning sensor atop the vehicle, which looks like a little hat. Mine arrived with a glowing CN in aquamarine. Both trips felt surprisingly normal, the cars handling tricky streets, pedestrians, and even—in the case of Charcuterie—a fire truck, relatively easily.
As driverless cars hit the streets, human drivers continue to kill people. This month, a 4-year-old in San Francisco was killed by an oncoming vehicle while crossing the street with her father. As we drove around in Charcuterie, Riggs told me that robotaxis were personal for him—he’d recently lost a friend who’d been hit by a human-driven car while riding a scooter. The fire-truck collision, during which the Cruise passenger was sent to the hospital with minor injuries, Riggs argued, was “a one-in-a-million situation.”
The future of driverless cars could mean safer roads and fewer unnecessary deaths. Alain Kornhauser, an engineering professor at Princeton, told me that the car has been so successful because it is a “DIY, Home Depot solution” to transportation—you pay upfront, and then you can drive yourself anywhere. But what about those who can’t afford a car? Or are too young to drive? Or have a disability? “What about those folks? Don’t they deserve mobility too?” he said. Cruise and Waymo cost roughly the same price as ride-hailing apps such as Uber, but eventually autonomous vehicles, he argued, could be cheaper than taxis. Experts I talked with saw them as expanding beyond just private taxi service and into self-driving shuttle operations that could fill in the cracks in the nation’s public-transportation system.
Whether that future ever does come is hardly guaranteed, and the path toward it likely includes even more deaths. In 2018, a self-driving car operated by Uber killed a woman in Arizona. Tesla’s autopilot feature has logged more than a dozen fatalities. Those for and against the expansion of robotaxis in San Francisco both argue that lives are at stake. This conversation is effectively about consent and risk. “It’s wrong to expose residents of the city to increased risk of harm because maybe someday they will get benefits,” Koopman said, “but we don’t actually know when that day will be.”
The potential of self-driving cars is easy to envision. The applications seem incredibly promising. They are good now, and they need to be tested somewhere to get even better. How much risk and inconvenience should a city like San Francisco be willing to take on to get there? The disconnect between the present and future of these cars can feel disorienting; no one can blame people in the city for being skeptical of technology companies that promise big things while creating real harm in the meantime. The same contradiction afflicts so much else in Silicon Valley. Perhaps chatbots and AI will transform the world, but for now that feels far away, and the road forward is rocky.
After Riggs and I successfully completed our drive, we sat under a tree to continue our discussion. A few minutes later, a Cruise vehicle drove by. I wondered if it was Charcuterie, heading off to pick up its next customer. It wasn’t. That car was named Winter.
San Francisco is stuck in a weird space between what driverless cars are now and what they could become, and soon enough even more of the country might be too. Amazon and other companies are also testing driverless cars, and Sandy Karp, a Waymo spokesperson, said the company is working toward building a product that “can be applied to any city, on any type of vehicle, and support a range of use cases from ride hailing and long-haul trucking to local delivery and eventually personal car ownership.” Kornhauser, who lives in New Jersey, said he’s eager for the day that he himself can test the future of transportation. “There’s a whole country out there,” he said. “Come to Jersey.”
