Back in August, the autonomous vehicle industry was riding high. Regulators in California had just allowed driverless taxis in San Francisco to compete 24/7 with Ubers and Lyfts, an AV milestone.
Fast-forward three months, and the California Department of Motor Vehicles has suspended the robotaxi company Cruise, owned by General Motors, from operating anywhere in the state. Federal regulators have also opened a probe into multiple incidents involving Cruise cars.
Andrew Hawkins, the transportation editor for The Verge, has reported on the long-awaited autonomous vehicle revolution for years. In an interview with Marketplace’s Matt Levin, he explained the trust issues and other potholes in Cruise’s path, starting with a grisly accident in San Francisco.
The following is an edited transcript of their conversation.
Andrew Hawkins: There was a hit-and-run involving a pedestrian. The hit-and-run vehicle driven by a human hit this pedestrian, she was flung into the path of a Cruise driverless vehicle there, and the Cruise vehicle inadvertently ran her over. But then instead of stopping where it was, it proceeded to execute what is called a pull-over maneuver. It was trying to pull over to the side of the road and the woman was trapped under the vehicle at the time, and she was drug around 20 feet to the curb, where she sustained further injuries. She’s still in pretty bad shape in the hospital.
According to the California DMV, Cruise sent a video to regulators, but they left out about seven seconds of the video showing the Cruise vehicle dragging the woman to the side of the street. Cruise denies this, but it did factor into the DMV’s decision to pull the permit.
Matt Levin: In the piece that you wrote for The Verge last week, you said these robotaxis have serious trust issues. Can you elaborate on what exactly you mean by that?
Hawkins: I think it has a lot to do with the way that these companies are communicating with the cities and with the regulators. There are huge gaps in what the companies are trying to portray as some of the advantages of the technology versus what these city officials are saying. I think what it all sort of comes down to is there’s a lack of national standards and federal regulation around the deployment of autonomous technology. That’s because Congress has not passed any legislation sketching out what it wants from these companies. There is a voluntary system for disclosure that has existed since the [Barack] Obama administration, but there aren’t any mandates in terms of what the companies actually have to disclose. So, there’s just a lot of gaps in what the companies are actually required to say to the government. And that leads, I think, to a breakdown of trust, and without trust, no one’s going to use this technology and these companies are going to fail ultimately, especially if people perceive them to be dishonest or incompatible with how city life is expected to unfold.
Levin: What types of data would be valuable for regulators to have that AV companies are not currently disclosing?
Hawkins: The companies claim that a lot of this data is proprietary, that it’s a trade secret and if it was out in the public, it would undermine their business in terms of how they are competing with other companies. In a place like California, that carries a lot of weight. But that said, there’s just a lot we don’t know about how this technology operates. It’s basically a black box. And so, we just don’t know much about their vision systems, their perception systems, the way that [artificial intelligence] is being used. All that we get from them is marketing materials essentially, a lot of boasting about how great their technology is and how it never gets tired or distracted the way that human drivers do. But we lack a lot of details about how the technology works. There’s just this, this big gap between what they, what the companies say and what would actually be useful to regulators in terms of the disclosures and the transparency that they might receive.
Levin: That argument that autonomous vehicles aren’t perfect, but they are kind of inherently safer than human drivers, that does seem to resonate, at least with me. Over 40,000 Americans die in car crashes every year. Does that argument hold water with you?
Hawkins: Yeah, it’s a statistic that is held up very often by the companies themselves when they are trying to make the case for robot-driven vehicles, that a terrible number of people die in traffic crashes every year and that a certain percentage of those deaths is the result of human error. There are a lot of experts that say there’s some false equivalency going on here because when we talk about human-driven vehicles, we’re talking billions and billions of miles every year. But when we’re talking robot-driven vehicles, they’ve driven far fewer miles than human-driven cars. So it’s not an apples-to-apples comparison. We still need to give autonomous vehicles more time to operate in the public in order to make a stronger safety case about how much safer they potentially could be over human-driven vehicles.
That said, I think the marketing pitch from the companies that their cars never get distracted, they never get drunk, they never get drowsy is a very strong case, and I think it resonates with a lot of people, especially people who have physical disabilities and have difficulty with mobility and getting around. But it still comes down to this trust [issue]. I think the companies tend to only publicly report numbers that regulators force them to report or that they think make them look especially good. And that’s true for all car companies, not just Cruise. I think they need to get a little bit more serious about what they are actually disclosing and how useful the numbers they are disclosing really are, so that we can get a better, more well-rounded picture of how this technology actually operates.
Levin: Cruise obviously isn’t the only robotaxi company out there. There’s Google-owned Waymo, which is already operating in four cities. How are the issues that Cruise is confronting affecting the entire robotaxi industry?
Hawkins: I’ve heard it said many times before that none of these companies are operating in a vacuum. And I think most people out there today aren’t differentiating Cruise autonomous vehicles from Waymo autonomous vehicles, although I think the companies would say that there’s a lot of difference between how they operate. But basically, a rising tide lifts all boats, and the same goes for whatever the opposite of that phrase is. And, like it or not, all of these companies are sort of in the same boat together.
Levin: I want you to play Nostradamus for us for a second. Five years from now, when I stumble out of a bar at 1:30 in the morning and I open up whatever rideshare app, is the car that arrives probably going to be a robot?
Hawkins: First of all, I would encourage you to take public transportation if you’re sober enough to make that choice. So often public transportation gets lost in this conversation. But yes, assuming that you’re opening an app and summoning a car to you, five years from now, I think there’s maybe a 50% chance that it’s an autonomous vehicle. I wouldn’t call it an overwhelming chance that it is, and I think it will depend a lot on where you live and what state you’re in. If you live in a state like Arizona or Texas that’s trying to encourage more of these companies to come and test their vehicles on their roads, it’s very likely that it’ll be an autonomous vehicle. But maybe if you’re in a place that’s a little bit more skeptical about the safety case or whether or not this technology is actually ready, maybe it’s going to be less likely. I don’t see anything shifting on the federal standpoint that would make it so that Congress suddenly wakes up and decides to start regulating self-driving cars. So, I think that there’s a growing chance that five years from now it’ll be an autonomous vehicle that comes and picks you up and shepherds you home and then hopefully you take some Tylenol and get a good night’s sleep. But at the same time, I think a lot of that will depend on where you live and how those states are responding to the increasing use of these vehicles but also the safety questions that are still swirling around them.
Last week, Cruise said it would recall 950 of its driverless vehicles. Andrew Hawkins reports that it’s not a physical recall but instead, an “over-the-air software update.” The recall is in response to the Oct. 2 hit-and-run Hawkins mentioned earlier. It will program those vehicles to remain stationary rather than pull over to the curb during certain crash scenarios.
When Cruise announced the recall, the company acknowledged that in certain situations, the pullover maneuver “is not the desired post-collision response.”
I also recommend that you check out a recent New York Times story on how GM’s aggressive approach to AVs may have pushed Cruise to deprioritize safety. It also has telling details about just how often humans are intervening remotely to assist robotaxis. It’s more often than you think.
Maybe they should call them semiautonomous vehicles.
The future of this podcast starts with you.
Every day, the “Marketplace Tech” team demystifies the digital economy with stories that explore more than just Big Tech. We’re committed to covering topics that matter to you and the world around us, diving deep into how technology intersects with climate change, inequity, and disinformation.
As part of a nonprofit newsroom, we’re counting on listeners like you to keep this public service paywall-free and available to all.