Speaking of Psychology - Self-Driving Cars (SOP57)

Episode Date: May 16, 2018

Several technology and automotive companies are already testing highly automated vehicles on public roads, and many automobiles can be driven with the assistance of semi-automated systems. Through the development of these systems, significant public attention has been placed on the promise of removing drivers from the vehicle; however, more limited focus has been drawn to the role of people in automated vehicle systems. David Friedman discusses how automation inside and outside vehicles may shape the future of self-driving cars. APA is currently seeking proposals for APA 2020, click here to learn more https://convention.apa.org/proposals Learn more about your ad choices. Visit megaphone.fm/adchoices

Hello and welcome to Speaking of Psychology, a podcast produced by the American Psychological Association. I'm your host, Kim Mills, and I'm speaking to you from APA's Technology, Mind, and Society Conference in Washington. This is an interdisciplinary meeting aimed at assessing current efforts to understand and shape the interactions of human beings and technology for identifying priorities for future work, and for promoting exchange of ideas and collaboration among participants. David Friedman is the director of cars and product policy and analysis for consumers union, the policy arm of consumer reports that focuses on product testing and consumer safety. Mr. Friedman serves as a thought leader across consumer reports and leads consumers' unions' public policy and analytical efforts

to ensure automobiles and other products are safe for consumers. to improve consumer options for energy efficiency and cleaner energy, and to guide other important initiatives. An engineer, he previously served as both deputy and acting administrator of the National Highway Traffic Safety Administration and was Deputy Assistant Secretary for the Office of Energy Efficiency and Renewable Energy at the U.S. Department of Energy. Mr. Friedman also worked for 12 years at the Union of Concerned Scientists

as senior engineer, research director, and deputy director of their Clean Vehicles Program. Thanks for joining us today. Thank you. So I'd like to start by talking about self-driving cars. And my first question is whether Consumers Union has a position on the need for and feasibility of self-driving cars at this point. Well, we're really excited about the potential for cars that can drive themselves. I mean, just think of a world where it doesn't matter whether you, have a disability, what your age is, or what your driving capability is, you can have mobility

and access all over your city or town. Plus, if you combine those self-driving cars with electric cars and car sharing, we could dramatically lower cost for consumers while increasing mobility and lowering emissions. So the potential is amazing, but that word potential is important. The reality is the technology of self-driving cars in many ways is that, best equivalent to a novice driver or a teen driver. There's a lot of work that still needs to be done. So are we ready psychologically for self-driving cars? I think that is a fantastic and really critical question because when I look at self-driving cars, one of the things that I think is important is to make sure we have, we design automation for humans instead of trying to design humans for

automation because we will fail at that latter task. Humans are. Humans. centered automation is going to be the key to success here, and that means understanding more about how people are going to interact with this technology. What are the pros and cons of self-driving cars? I mean, I think it cuts both ways. Well, when I think of self-driving cars, I kind of think about we have two possible futures with this technology. In one future, we have a utopia where we have better mobility. We effectively eliminate traffic crashes, we cut costs, and have a great convenience feature for consumers so they can get wherever they want to go whenever they want to go. That's the utopia

situation. The dystopian world is where we have self-driving cars that aren't safe, put people at risk, that increase the amount we travel and increase the emissions we create and create a world where we're all basically locked into our cars instead of spending time with our family and friends. So you mentioned the possibility of fewer accidents, and of course there's been a lot of publicity around the accident that just happened in Arizona where the Uber hit a cyclist. What went wrong there?

I mean, why didn't the car know that the cyclist was there? Well, you know, there's several investigations going on, so I always want to be careful not to prejudge. We have two federal agencies that are going to dive in and try to understand the details. What I've observed from the video associated with it is a couple of things. One, it looks like the system wasn't capable of stopping in time in approaching a pedestrian in the dark. That's shocking. I mean, when we think of self-driving cars, we think of a car that's going to be safer than a human driver.

So that's one thing. It looks like the technology wasn't up to task. The second thing is they had a human in the, in the driver's seat who was supposed to be there to take over in case of an emergency. But the human was distracted. And what that tells me is something I think we all already know. Humans aren't very good at watching paint dry. I mean, that's the job of these folks is to try to pay close attention to everything that's

going on around them in a car that's supposed to be able to drive itself. That is a very difficult task. These cars should have driver monitoring systems to make. sure that safety driver is actually paying attention. I would say the biggest thing that went wrong here though is fundamental human error in the design, testing, and rush to market of these self-driving cars. And yet we've been talking about self-driving cars for something like 50 years, right? So how are we rushing to market? So if you think about automation in cars, we've We've actually been increasingly automating cars for over 100 years actually.

I mean, originally you had to walk up to a car and literally crank the engine to get it started. And then we created starter motors. So from starter motors to self-driving cars, we've had basically an evolution of automation in our cars. But what's different now is that the computing power and sensors available to car companies and tech companies is light years ahead of where we're going to companies. we've been. And they're now trying to figure out how to apply that to cars. So why does that mean we're rushing things now? Well, the problem is there's so much money to be made from a world of

self-driving cars that companies appear to be rushing to test these vehicles on our roads and sell these vehicles on our roads before they have demonstrated and proven that they're actually safe to be on our roads. So what do you think ultimately is the role of the human being inside a self-driving car? Well, I think really the big question is both what is the role of the human inside the self-driving car, but also what is the role of the human being in the design, manufacture, and approach and governance of self-driving cars? I mean, hopefully in the long term, we'll get to a world where cars are truly smart enough

and capable of driving themselves. But we're only going to get there if we recognize that self-driving cars do not eliminate human error. All they do is shift the propensity and the likelihood of human error from the driver where it is today, to the people designing the vehicles, to the managers approving those designs, and to the corporate executives who are establishing the culture of that company. Are you going to make the error of putting profit over safety, or are you going to make the choice to put safety over profits when it comes to self-driving cars? And that's so many people try to say self-driving cars will eliminate human error. That's not the case.

Unless our robot overlords take over, humans will always be in loop. So here's a conundrum, a self-driving car that has to make a decision. Let's say that there are pedestrians in a crosswalk and the car can either hit the pedestrians or it can pull into oncoming traffic. So it has to decide effectively between whether you kill or maim pedestrians or kill or maim the people inside the car. How does it choose? Well, I mean, this is the classic trolley car problem that has been an ethical question

for as long probably as people have been thinking about this. And let me answer it two ways. The first answer is we have no idea. Why? Because there are no standards or requirements or policies for ethics when it comes to self-driving cars. And in fact, at one point, the agency involved with overseeing auto safety, the National Highway Traffic Safety Administration, put out guidelines saying that car companies should be transparent about how they're going to approach these issues. Well, in the most recent administration, they eliminated ethics from the guidelines. So I'm very concerned that we don't know and we're not going to know for a long time how the car is going to make that decision. What should happen is the cars should be designed to never get in that situation in the first place. It's kind of like the Captain Kirk Kobayashi-Maru approach, you know, when given a difficult test, change the rules of the test. And that's what self-driving cars need to do, is we need to design them so that they're safe enough that they can avoid those tough situations. And then if they are in those tough situations, yes, ultimately there needs

to be some public agreement as to what moral and ethical standards are we going to apply to the design of those vehicles. And that gets back to my earlier point. Humans are always going to be in the loop when it comes to self-driving cars. So what do you see as the role of psychologists in the development of self-driving cars? To me, there's so many possible critical places. that I can see bringing psychology into the development of self-driving cars. I think first of all, it is understanding human error and helping people figure out where does human error come from and how you can minimize it. Because, as I said, yes, maybe you're ultimately, you'll get the human driver out of the loop. You'll still have human error throughout

the design and decision-making process. So how do you, how can psychologists help understand and corporate culture and designer culture and software coder culture such that you can influence that culture, help us change that culture to ensure that the proper steps are in place so that safety comes first. I think another critical role is realizing that there's also humans are going to be in the loop when it comes to pedestrians and cyclists. Even if tomorrow, every new car was a self-driving car, you'll still have people on the roads who are walking or biking. And even if every new car on the road was a self-driving car next year, you still have a legacy fleet of human-driven cars that are going to be on the road for

10 to 15 years. So how do you deal with that mixed-use situation is really critical. Another key role is how do you do human-centered automation? How do you ensure that a car that has no driver can talk to a pedestrian or other cars? How do people think? How do people perceive that technology? And what kind of communication from that technology around its intents can help save lives? So the car can't just scream, hey, stupid, get out of the way? Exactly.

Or just think of if you're a pedestrian walking across the road, rocking across a crosswalk, what do you try to do? I try to lock eyes with the driver of those cars so that, A, I know they see me and I humanize myself as someone who shouldn't be run into. Exactly. And if they don't look at me, then I'm nervous and maybe I won't cross. Well, what if there's no human in there? How do I know that car is paying attention to me? So this whole concept of human-centered automation, both the design of the automation for the people inside the car, and the design of the automation for the people outside of the car, I think is a rich and critical area for the world of psychology to get better engaged.

Last but not least, it's understanding how humans deal with risk. Think about it this way. How many fatalities in the United States do we tolerate when it comes to commercial airliners? The answer is zero. How many fatalities do we tolerate every single year when it comes to our cars? A lot. Around 40,000. In fact, in 2016, it was 37,461 lives lost.

It was a record year, a 14% increase over two previous years. And yet we tolerate that. Why? Is it because we tolerate our own error? We can see ourselves in the drivers who are driving drunk, who aren't wearing seatbelts, who are making all these critical errors, but we can't see ourselves in the pilot or the company responsible for the plane,

or is it a question of control? If I'm in control, I know I'm safe. I mean, I'm a safe driver. All of you are unsafe, but I am a safe driver. That is a very classic attitude, and it comes with some risks. But if someone else is in control, how does that alter our perception?

These are critical things we need to better understand, not so that we can manage people around the safety of these vehicles, but so that we can manage the safety of these vehicles around people. So some of what you just said raises in my mind the question of legal liability. So who is liable? Is it the car and the engineering, or is it the person who's sitting there ostensibly doing little to nothing? Well, and this brings up a couple of things.

First of all, I think in a world where the car truly drives itself and there's no human driver other than setting the destination, I think it should be pretty clear. It's the company who designed and operates the vehicle. But the world we're in right now, there are none of these true self-driving cars for sale today. The only cars we have are cars that can take over some of the driving functions, but not all of them. And in 2016, we had a case where someone was driving a Tesla vehicle with its autopilot system. They ignored the warnings over and over that they needed to be having their hands on the wheel. They ended up plowing into the side of an 18-wheeler because the car couldn't see it. I mean, imagine that.

The car couldn't see an 18-wheeler going broadside in front of it and so plowed right into it. I would say there's double fault there. And in fact, the National Transportation Safety Board, that's exactly what they said. They said, one, the driver bore fault because they ignored the warnings, but also the company bore fault because it was allowing this car to operate in a situation. It couldn't handle. It allowed the car to keep cruising down the road, even though the driver was ignoring it. it allowed the car to operate on a road where you could have an 18-wheeler crossing in front of it.

So until cars can truly drive themselves, both the human and the humans behind the car are going to be responsible when these cars hurt and kill people. Any other aspects of self-driving cars that we haven't touched on that you want to talk to people about? Well, I would say one thing that's, I think, critical for folks to think about is what is the role of government within all of this? And the challenge we have right now is a lot of the effort is actually to get government out of the way. So you have Congress is working on a bill that would allow hundreds of thousands, if not millions, of self-driving cars to be sold or commercially used on our roads that could be. exempt from critical safety standards. You have an administration that is making clear that all guidance on self-driving cars is voluntary.

This laissez-faire attitude, I fear, is going to end up delaying the technology. The excuse that folks give is it's about accelerating the technology. But as we start to see fatalities from self-driving cars, I think that poses a huge risk for consumer backlash. And if consumers think that this technology is unsafe, it's going to delay the technology by years or even decades, which means it's going to delay the potential to save lives by years or even decades. So that's one of my biggest concerns right now is there are no protections in place. And we need, in the current situation we have where government isn't stepping up,

We need companies to step up and put safety first, to share data making clear how safe their vehicles are, to not allow their vehicles to be operated outside of situations where they're safe, and to stop working to undermine existing consumer protections in Washington. Well, thank you very much, David Friedman, for joining us today. It's been very interesting. Thank you. I appreciate it. Speaking of Psychology is part of the APA podcast network, which includes other great podcasts such as APA Journal,

dialogue about the latest and most exciting psychological research and progress notes, which discusses the practice of psychology. You can find all our APA podcasts on iTunes, Stitcher, or wherever you get your podcasts. You can also go to our website, www.com speakingof psychology.org to listen to more episodes and see more resources on the topics we discuss. I'm Kim Mills with the American Psychological Association, and this is Speaking of Psychology.