What does the future of driverless cars look like?
DAVID FOLKENFLIK, HOST:
And finally today, in the near future, residents of Seattle may well be startled as they notice a fleet of new cars cruising down the city streets. These are to be cars without drivers behind the wheel.
(SOUNDBITE OF ARCHIVED RECORDING)
UNIDENTIFIED NEWS ANCHOR: (Laughter) It looks like a - maybe like a bus from the front and a toaster on the side.
FOLKENFLIK: That's a local news anchor commenting on Amazon's autonomous car unit, Zoox, which will soon start test driving downtown there. Zoox engineers say the city's wet, windy and congested streets are the ideal testing ground for what are being called robotaxis. Their aim is to one day deploy a fleet of driverless taxis. Advocates for these cars say they'll solve a ton of problems, like reducing collisions and carbon emissions. Critics are casting doubts about how safe these vehicles are and questioned the motives propelling the embrace of this technology by tech giants.
Someone who has studied and written about this issue is Peter Norton. He's the author of "Autonorama: The Illusory Promise Of High-Tech Driving," and he teaches history in the Department of Engineering and Society at the University of Virginia.
Peter Norton, thanks so much for being with us today.
PETER NORTON: It's great to be here.
FOLKENFLIK: Before we jump into the debate around these cars, Professor Norton, can you explain just what a driverless car is and how it works?
NORTON: Well, it's a car where the steering, the braking and the other basic driving operations are performed by robotics, which is a combination of software, computers and hardware. And it's guided by sensors instead of human vision to detect what's around it and navigate through space that way.
FOLKENFLIK: So what's the broader appeal here?
NORTON: Well, you know, the promises - and I underline that this is a promise and not a reality - is that this will be primarily safer, that it will be less expensive for the operators because they won't have to eventually pay a human driver to be in the vehicle. Some companies also make claims about congestion relief, which really don't appear to be standing up at all. But that's the appeal.
FOLKENFLIK: Take a moment. Let's drill down. Companies such as Zoox claim these driverless cars will improve traffic and mobility in urban areas like Seattle. You just said it doesn't really hold up. Why?
NORTON: Well, I mean, you have to take it a piece at a time. Take safety, for example. One of the things that Zoox is claiming is that their vehicles will be safer. And the usual explanation is that these vehicles will not be susceptible to all the human feelings that human drivers have, like distraction, fatigue, impatience and so on, which is all true. But what they tend not to point out is that the vehicles are also susceptible to deficiencies that robotic drivers have that humans don't, discriminating between, say, a pedestrian waiting at the curb to cross the street and a pedestrian who's just, you know, looking in a store window. That's a hard thing for a robotic system to distinguish but something that human beings are very good at distinguishing.
FOLKENFLIK: Based on your research, how would you characterize the environmental impacts of these self-driving cars versus, you know, prototypical human-driven counterparts?
NORTON: Basically, all of the visions for autonomous vehicle futures have them being battery electric vehicles. And so there's an obvious advantage to electric vehicles over combustion engine vehicles. They're definitely cleaner. It depends a lot on the power grid that they're charged from. But there's also the problem of batteries. The world is going to be really straining to get the lithium, the cobalt, the nickel that it needs for these batteries.
And there's some very high human and environmental costs that come with that as well, most notoriously in the Democratic Republic of the Congo, where 70-some percent of the cobalt comes from for the lithium ion batteries. And the labor conditions there are quite appalling.
FOLKENFLIK: When this conversation was first suggested by one of my colleagues, what came to mind first was a string of past incidents in which driverless cars caused real-life harm to real people, to pedestrians in particular. There's one in 2018 where one of Uber's autonomous cars struck and killed a woman in Arizona. How much improvement have we seen on the technology side over the past few years?
NORTON: So the answer to that is mixed. Machine learning is helping these cars perform better all the time. And so in that sense, there is steady improvement. The thing that hasn't changed at all is that the vehicles are going to have a bias that favors the experience of the vehicle's occupants, you know? They need to have people who are willing to pay money to get in one of these vehicles. And to be willing to pay money to get in a vehicle, you want to be sure that the vehicle gets you somewhere fairly soon and isn't constantly breaking on the small chance that it's detected something that might be a human being or something with a human being in or on it. In Seattle, of course, there are going to be pedestrians. And if they want to have it being a paying business proposition, the vehicles will have to take chances or be so slow and break so often that no one will be willing to pay to ride in them.
FOLKENFLIK: Let's flash forward, say, 50 years from now or more. Assuming we're living in a society recognizable to the way we live now, people still need to get from A to B or A to B to C. What does that look like? How do we get there?
NORTON: Well, I mean, what I do look forward to is a future where we have choices we don't have now, where - for example, if you live in a typical American suburb, you probably can't walk to a grocery store. And that's probably because your suburb is zoned for single-family residences only. But if we just relaxed our zoning rules, then it would be possible for people to actually walk for practical purposes, which was, of course, the primary mode of transportation for most people throughout history.
For very low cost, we can make cityscapes that are bikeable, and we can learn from other countries where bikes are part of a public transit network, where you can take the bike to the train station, leave your bike there. And when you get to your destination train station, there's a public transport bike that you can pick up with the same card you used to ride the train. These are things we can do now with the technology we have now. We don't have to have mind-blowing machine learning to make that kind of thing work. We don't have to ask if it's possible because we already know that it's possible.
FOLKENFLIK: That's Peter Norton. He's the author of "Autonorama: The Illusory Promise Of High-Tech Driving" and a professor at the University of Virginia. Peter Norton, thanks for joining us.
NORTON: It's been my pleasure, David. Thanks very much.
(SOUNDBITE OF SPACE'S "MAGIC FLY")
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.