Loading summary
Transcript101 lines
- [00:01]
A
You're listening to Is Business Broken, a podcast from the Mehrotra Institute for Business, Markets and Society at Boston University Questrom School of Business.
- [00:11]
B
I'm Kurt Nickish.
- [00:16]
A
For years we've been promised a world
- [00:18]
B
of self driving cars and you can
- [00:20]
A
now take a robo taxi in a growing number of U.S. cities, including San Francisco, Los Angeles and Phoenix. But are Bostonians ready to hop in one?
- [00:30]
C
No. I would not ride it if it came to Boston because driving in Boston is already difficult as it is.
- [00:37]
A
Waymo last month announced plans to test its newest fully autonomous cabs in the city's narrow alleyways, roundabouts and snowy weather. So our producer CCU asked folks around Boston if this is something they're ready for.
- [00:51]
D
I think I'd wait a little bit just to get a gauge of whether there's any accidents in like the first few months, I think.
- [00:59]
C
And I've actually taken one and they're really cool. It was kind of surreal because I sat in the back and the car was built like a normal car with the steering wheel and everything but no one was in them. And it functioned like a normal car. It was able to change lanes and turn and I felt pretty safe in it.
- [01:19]
E
It feels a little bit unsafe personally,
- [01:22]
F
but you have to put full trust
- [01:23]
E
in the company and whatnot. Yeah, I would take them if it was cheaper than Uber, Lyft or any
- [01:27]
B
other like platforms I wouldn' because I
- [01:30]
C
feel really scared that I'd crash here. People are really unsafe driving here and
- [01:35]
B
I think you need like an actual
- [01:36]
C
driver to drive a car.
- [01:43]
A
Some people we interviewed say they do trust the self driving car, but not everyone is ready to leave the driving to an artificially intelligent cab. That tension is not new. It happens every time a breakthrough technology
- [01:57]
B
moves from the lab into everyday life.
- [02:00]
A
So how do we cross that barrier? What makes people trust or reject a technology that asks them to give up control? And this isn't just a psychological question, it's an economic one. Why invest billions of dollars in expensive new technology if people won't actually use it? Carey Morwitz joins us today to help us understand why. He's a professor of marketing at BU Questrom and an expert on the psychology of technology and decision making. We're also joined by Bashak Erzer, who provides a perspective from industry. She's an entrepreneur and board director specializing in physical AI businesses. She's also the former chief product officer at self driving carmaker Motional. Now part of Hyundai. Motional has tested robo taxis in places
- [02:49]
B
like Las Vegas and Boston.
- [02:52]
A
Carrie and Bashak, thanks for coming on the show.
- [02:55]
E
Thanks for having me.
- [02:56]
G
Thank you.
- [02:57]
B
KERRY Bashak, Makers of these autonomous vehicles
- [03:01]
A
say that they're statistically safer than human drivers. But it's clear that people are afraid
- [03:06]
B
to hand over the wheel. You know, just by listening to those voices we just heard, what jumped out at you as you listened to them.
- [03:13]
E
When we look at the psychology of these kinds of things, too, there is a psychology of adopting a new technology. It's unique in the sense that we're giving up control as a species over many kinds of things and as a person, right? So for people who think about driving as the same way that they might buying a car, as the same way that they might buy a washing machine as a commodity or an appliance, where it's a thing that gets me from A to B that may be not as threatening, and you're just thinking about, is this safer, is this cheaper, is this more convenient for me?
- [03:50]
B
We heard some of that in that tape, right? We heard some, like, reasons that they would have pain points to give it a try. It was also interesting to me that the one person who had written in them didn't seem that concerned about him.
- [04:03]
D
Right.
- [04:03]
E
And as we get more familiar, that will be something that people become less hesitant around.
- [04:08]
A
Vishak, what did you hear in there?
- [04:10]
G
We have seen in thousands of rides that we have given is the first few moments is sort of the white knuckle. People are just really nervous. And then as you watch them, they transition into just this relaxed mode and they go about their business. They look out the window or to their phone. And that's thanks to giving them a lot of visual cues as to what the robot car is seeing. What people want is predictability, right? So if there's a jerky moment or if there's an unexpected slowdown, the person, the passenger, wants to know why that's happening. You know, we've done a lot of putting a screen in front of of the rider to let them see exactly how the robot car is seeing the world. So we spent years teaching the robot car to decipher between is it a toddler chasing a ball on the side or is it an adult waiting for bus? You know, so there's a lot of that teaching, and that needs to translate into this human machine interaction.
- [05:14]
B
Well, Carrie, you co authored a study on consumer acceptance of automated vehicles two years ago, and I think it really applies here. The research basically says that we as people are hypocrites. Like, we're okay with other people using Them, but not ourselves as much. Why does that gap exist?
- [05:31]
E
Okay, so we would want AVs if we thought that they were going to be safer and cheaper and more reliable. Waymo released a huge study where they looked at 127 million miles of driving in LA, San Francisco, Phoenix and Austin. And they compared Waymo cars to National Highway Transportation Safety Administration data. And they basically find that there's like 90% fewer crashes without a serious injury or death in the Waymo car than a comparable human driver. And there's 81% fewer crashes without any injury. And it's safer for pedestrians, it's safer for cyclists and motorcyclists in general. So the data is pretty clear that these cars can be much safer than human drivers. But when we ask in our surveys, we look at nationally representative samples of people, we ask them to pick what level of automation they would like for themselves. And there are certainly technologies that everyone wants, like anti lock brakes, when you get into more automated kinds of features. People think that would be great for other drivers, but not for themselves. And it's because they trust themselves and think they're safer drivers than other people. But underlying all of that, the data seem to suggest that a big part of these kinds of hangups that we have about autonomous technologies is that when we see them, if we are comparing ourselves to that technology and it's something that we care about, that kind of comparison can be threatening. And so I'm going to be likely to take a biased evaluation of myself or the robot.
- [07:10]
B
The tech takes away a piece of my identity potentially.
- [07:13]
A
Well, Bashak, let's look at the tech.
- [07:15]
B
I think we kind of get that like just showing people a spreadsheet of data, or data from Waymo, for instance, showing that self driving cars are safer isn't going to be the automatic way to convince people. When you were emotional and testing robo taxis in the streets of Boston, what kind of safety benchmarks did you apply to yourselves and how did you try to reduce the risk but also communicate that to people who were using the robo taxis?
- [07:48]
G
Yeah. So the challenge is how the vehicle interacts with other road users. Right. When you remove the driver from the vehicle, what you're removing is common sense, because you can teach the robot car all of the rules of the traffic,
- [08:06]
B
everything, and they see the road better than people can.
- [08:09]
G
Yes, but what you're removing is common sense. So just a simple four way stop. And regardless of who arrives there first, we usually communicate with each other just with eye contact or a nod, or just with a Hand gesture. That's what you're removing from the vehicle.
- [08:28]
B
They look like they're in a hurry. I'm just gonna let them go ahead.
- [08:30]
G
Exactly.
- [08:31]
B
Even though I was technically there first.
- [08:33]
G
Exactly. That's why I think we come back to we need to make sure that the robot car communicates with the rest of the world, all of the other road users, whether it's pedestrians or motorcycles or other human driven vehicles, in a very clear and predictable way. What we do is make the robot car act very deliberate, slower than a human driven vehicle, but always predictable. And that's, I think, to your question. That's how we tried to get around the issue of communication and making it safe, how we communicate the issues. Originally it was a lot about accidents. Any accident data, obviously those are very far and few between and not statistically valuable enough to draw conclusions. So when we test, of course we test and see a lot of any jerkiness or any feedback from the vehicle, we test those. But I think the real valuable information that tech companies can offer is not accidents, but near misses. There are a lot of near misses. And so in the hands of an urban planner, then they can potentially design an intersection slightly differently or take that into consideration in the future. And that's how you also develop trust with cities, with this open, transparent communication between the tech companies and the cities. And we're not talking about cars driving anywhere independently by themselves. We're talking about geo fenced areas and fixed routes for the foreseeable future. So there are some really key applications that lend themselves to that sort of technology. So last mile delivery, for example, is a great one. Public transportation, eventually, when this is proven safe, could be a great one. Campus transportation and even robotaxi, because it's all fixed routes and you're going from point A to point B, why not anywhere?
- [10:35]
B
Is that because it's hard to do from a regulatory perspective or it's just, it's hard to do from a safety perspective?
- [10:41]
F
Both.
- [10:42]
G
Because you go and study an area, it's more complex. Yes. You map the area and then you expand and you gradually increase your geofenced area where that robot car can drive itself.
- [10:54]
E
There's another really interesting issue around that kind of mixture, which is that much of human behavior is illegal. If you're another driver, you see people breaking the law all the time.
- [11:06]
D
Right.
- [11:07]
E
And so how do you navigate building a system that's designed to imitate human driving and integrate with human driving, but also follow the rules when everyone on the Mass pike is driving above the speed limit?
- [11:25]
D
Right.
- [11:26]
E
And so there's a friction there between designing a system that is legal and designing a system that can interact with the sort of mild violations of the law that many people engage in every day.
- [11:41]
B
Yeah, that's really interesting because I imagine that if you're at one of these companies, the last thing you want to see bashok is like a video of your car, people saying, oh look, they're speeding or look, they just ran a red light or just ran a stop sign, then it looks unsafe even though they see people do it every day.
- [11:58]
C
True.
- [11:59]
G
If anything, people could get frustrated in the future that the autonomous cars are a little slow.
- [12:05]
B
Right.
- [12:05]
G
It goes back to how we build trust with individuals with this boring way of driving.
- [12:11]
B
So it follows the speed limit on the Mass pike. It yields at all the right places in these chaotic Boston intersections.
- [12:18]
G
Yes, absolutely. If the AV driving feels boring, that's when we won. That's when adoption starts.
- [12:27]
B
I feel like we can see a future where, you know, there are more self driving cars on the streets, where there are fewer accidents and some benefits to society and to individuals through their availability. But this conversation is making clear that it's not like it's not a switch. This is like a, it's a complex process. And so I'm wondering if the necessary trust and familiarity that's going to be built comes through regulation, lived experience, cultural normalization, what are the kind of the key stumbling blocks you see to getting to that future?
- [13:02]
G
I think it's a combination of all three of those. It's exposure, it's building trust with individuals, building trust with the cities and expansion, expanding slowly with use cases that really make sense. My child goes to a Cambridge public school and takes public schools bus. When our beloved driver is away for four weeks to visit his family overseas, we're stuck. We just don't have enough drivers that we can pull in immediately to take that job. So that's one example that an autonomous public bus can solve. But I think that also leads us to the question of is there a point where eventually even people who want to take those jobs cannot find those jobs because a lot of them are automated. So that is a risk that we need to think about. But I liken that to our parents went to the bank to get cash. There were a lot more tellers. We in our generation, we used ATMs, our kids are just using cash apps. So those tellers lost their jobs, but they moved on to other jobs, hopefully. So for autonomous driving again, my future dream is not that all of the cars become autonomous. I think people will always keep their cars, fun cars on the side. But if we have a lot more public service vehicles and some of those jobs are lost, then you gain it through remote monitoring or, you know, some other design and instrumentation of the vehicles and there are new jobs created.
- [14:42]
B
And Carrie, maybe for you, you know what's kind of the key experience or messaging that's going to drive that forward?
- [14:48]
E
People have invested a lot of money in their cars. There is a technology that is there and is costly to replace for individuals. I 100% agree with Prashog in the sense that this is going to be more rapidly a business proposition where you're getting in a self driving car than a taxi or you are in an autonomous public vehicle. This is asking people to make a really big behavioral change about something they do every day. And so that is going to be probably a much slower process. But that doesn't mean that we'll still not adopt it. It's just going to take a while.
- [15:38]
A
Up to this point, we've been talking about why adopting self driving cars is as much a psychological challeng as it is a technological one. And that's also evident in the debate that's taking place right now right here in Massachusetts. As Waymo expands its efforts to test a new generation of autonomous vehicles on Boston streets, the company says it's working with state lawmakers to create a clear legal path for fully driverless cars. Now, it's worth pausing on that safety
- [16:08]
B
data for a moment.
- [16:09]
A
The figure Kerry mentioned earlier that compared to human drivers, Waymo cars were involved in about 90% fewer serious crashes. Well, that data is based on millions of miles driven in places like Phoenix, San Francisco, Los Angeles, and those cities have generally wider roads and milder weather than Boston. So we should acknowledge that those data were collected under favorable conditions, which may not represent all driving scenarios. And that uncertainty is part of what policymakers, drivers and safety advocates are now wrestling with in Massachusetts. And as we move forward, this isn't just a conversation about software and sensors and the size of streets. It's also about who's behind the wheel. What is the impact of self driving cars not just on riders, but on drivers? What rules should govern autonomous vehicles in Massachusetts in the future? To explore those questions, I continue the conversation with Gary Morwege, along with Mark Shieldrop, spokesman for AAA Northeast, and Mike Vardabedian of the International association of Machinists District 15, which represents APP based drivers.
- [17:23]
B
We're talking about Massachusetts here, but we could be talking about almost any state because AVs are not regulated fully by Congress. It's a state by state issue. About half already allow autonomous vehicles. And in Massachusetts right now, in front of lawmakers, there are sort of two competing proposals. One legislative bill wants to build a regulatory highway for fully driverless cars. Another would require a human operator to be physically present. Mike, of those two bills your union supports the one that requires a human operator to be physically present at all times. Critics say that human requirement is like Luddite move to slow down progress. What's your view?
- [18:06]
H
So there's a couple of reasons why a human operator still sitting in the front seat would be necessary from our point of view. First of all, safety. You know, I'm hearing just a few minutes ago about how the data shows that they're safer. I will tell you that there is more to safety than just the record number of accidents you have. When the WAYMU gets confused, its first thing is to just stop right in the middle of traffic. You've got emergency vehicles maybe that need to get through gridlock in the city, power outages that shut down all of the driverless cars.
- [18:41]
B
That famous incident in San Francisco recently.
- [18:43]
H
Yes, exactly. So having a live person in that driver's seat and not somebody over in the Philippines or India controlling these vehicles that may not be able to control them with a power outage is necessary to prevent gridlock, safety and so on. But then also talk about the labor aspect of it again, you're not replacing people with robots that don't need to earn a living. These drivers need to support families, need to support their household.
- [19:11]
B
We know that when cars get stumped, they call a human assistant. And Waymo's chief safety officer recently said that many of their assistants are based in the Philippines. Mark, when your members hear that autonomous vehicles are coming and realize that this driverless car might be guided by somebody, you know, 8,000 miles away, does that help the trust gap? Does this widen it? Like, what's your approach to that?
- [19:33]
F
Well, those types of incidents certainly don't help the trust factor at all. We've been taking the temperature of the American driving public about autonomous vehicles for a number of years now, and the vast majority are still uncomfortable with them. Last year, the results of our annual AV survey indicated that 13% of the survey respondents said they would trust riding in a self driving vehicle. That's up from 9% the year before. But 6 in 10 still say they're too afraid to ride in one. So there's certainly a disconnect there between what Waymo's self reported Data shows and how people actually feel. Our research has shown that generally when a technology arrives in a vehicle, people are very hesitant and uncomfortable with them at first. But there's a big concern that over time time, once they actually use them regularly and become comfortable with them, they disengage from the task of driving. We have some real significant concerns about the potential for self driving technology to come to a consumer vehicle. We've taken issue with things like Tesla calling it Autopilot, which gives the suggestion that the car's driving itself and you don't have to pay attention. We've all seen the videos of folks napping in the driver's seat of their Teslas on the Mass Pike. That's not how the technology is supposed to be used, but unfortunately the systems and the marketing suggest that's what you can do. I know that's different from these robo taxis that we're talking about, but the general public isn't seeing much differentiation between that. To them, self driving means you don't have to pay attention to your driving, even though these systems currently in consumer vehicles require the driver to stay just as focused on the driving task, whether the systems turn on or off.
- [21:19]
B
Yeah, I can see a big education gap there that has to be closed. Kerry, what have you been hearing about that?
- [21:26]
D
So I think there's sort of two issues that Mark raised. One is that for a lot of these technologies, whether it's in a car or in the doctor's office, people are comfortable with having some automation, but they want some human oversight on top of that to feel level of trust that they would with a fully human, human centric kind of system. Right. So if there's a driver in the car and they're picking me up, I trust that person to get me where I want to go. If there's a self driving car, people still want to have some kind of human oversight in case something goes wrong. And the idea of the human oversight being somewhere where like a power failure or other kinds of issues could arise would disrupt that is like pretty threatening to people. Right? Like that's a trust issue. The second issue that Mark raised is in the, in the psychology literature and in the behavioral economics and the business literature, we call this cognitive offloading. And so a lot of autonomous systems and AI are designed to replace what humans do not only in terms of like their motor actions, but also their cognitions, their thoughts.
- [22:31]
B
So it's, it's more than just leave the driving to us, it's leave your thinking to us.
- [22:35]
D
Right. And so a Question is like, as these systems kick in, so there are techniques like trail braking that most people don't do because we have anti lock brakes. Many people don't know how to drive out of a skid when they get into one in the winter because we have these kinds of technologies that help changes in terms of how the brakes are being applied to the cars.
- [22:56]
H
Right.
- [22:56]
D
That's not to say that there's not a gain overall in terms of safety, but the question is when these systems go out, what happens? So if I'm conducting a symphony in my car while I'm driving or reading a book and then the system goes out or the sensor has some salt over it, like am I going to be able to react because I am either distracted or I've not learned how to break in snow.
- [23:21]
G
Right.
- [23:22]
D
If the person who's overseeing the autonomous vehicle that's doing this is thousands of miles away, are they going to have time to react to that or if there's an issue?
- [23:32]
B
Gotcha.
- [23:33]
H
Yeah.
- [23:33]
F
And to your point, there's oftentimes people find ways to use systems kind of beyond their limits and inappropriate ways and that reduces safety. So we have to kind of change how we teach people how to use the technology as the technology changes. And oftentimes the technology changes a lot faster than the teaching that we're doing.
- [23:55]
E
One other thing to add on the
- [23:57]
D
psychology of that too is that. But what's the standard of comparison that we want to think about? Right. So is it that we're looking at these failures will happen? People also make human errors, right. How do we balance the kinds of mistakes that machines will make with the kinds of mistakes that will happen by a person being distracted or a person maybe being impaired while they're driving or not having the skill set to deal with a circumstance. And so, so we have to think about like as a society, like what are the kinds of trade offs that we want to make?
- [24:31]
B
Mike, you're in the middle of a historic push to unionize app based drivers in Massachusetts. We heard earlier in this episode an industry leader who said driving jobs won't die. They'll evolve into other technical monitoring roles, jobs for people to maintain these vehicles, take care of them. Another economic argument is that Massachusetts is home to a robotics industry and that by joining the other states that are already doing this, it kind of opens the door to more jobs, more startups, more related businesses around this growing industry and that it's going to create new opportunities that we can't even predict.
- [25:06]
A
Now.
- [25:06]
B
What do you say to the Fact that this will create new jobs that
- [25:10]
A
may not be one to one to
- [25:12]
B
the middle class jobs that you're talking about.
- [25:14]
H
I would argue that the people that are driving right now with their vehicles off the road, which they have a significant investment in, they would lose their jobs, not be able to pay for those vehicles. First of all, that would cause an economic problem right there. But those vehicles potentially being off the road and being replaced by waymu. I don't know that it's going to be really a net positive when it comes to mechanics. You're just basically replacing vehicles that are currently driving out there with robots. I would also say that the remote, the people monitoring, we've seen a lot of, okay, so those jobs are going overseas. I would say the emergence of AI getting more and more integrated into our system. There'll be more monitoring by AI on these vehicles and then probably need less people. And I would also say that people monitoring are probably going to make less skill wise unless there's a lot to remote control and so on. Well, if we're going to be replaced by AI and robots in a lot of our jobs and again, it's not just affecting blue collar workers, affecting white collar workers as well, what are we going to do to replace these jobs? And if we don't have replacement of jobs, what are we going to do for replacement of income and how people live and how they pay for things? And it's really just a transfer of wealth if we're going to be honest.
- [26:40]
B
Okay, let me ask about the future then. If you imagined Boston 10 years from now, Mike, what gives you hope about how this transition's gonna be handled?
- [26:51]
H
I do encourage the legislators community to really think this through and do an in depth study. Really think about the impacts of safety to the public, the impacts of even different types of people that would use these vehicles. I know that, for example, we've been in contact with the disabled community. It might be good for some of the disabled, but what about the people have mobility issues? How are they without a human to a system? How are they going to get into these cars and how are those going to help and could they figure out a solution? But then we're back to the society aspect of it and the economy. So there needs to be an in depth study, think about the impacts of it and figure out whether one, it's good for Boston or two, if it is good for Boston, what kind of regulations would be necessary to ease the impact of economy and so on.
- [27:45]
F
Mark, you know we have a lot of work to do. And I think technology can certainly play a role, but really it boils down to our own perceptions about safety and what we find to be acceptable behavior on the roadway, whether it's a robot or a human driving the vehicle. Ultimately it's humans who decide ultimately what happens on the roadway.
- [28:07]
B
Yeah, Mike, Mark, Carrie, thanks so much for talking about this.
- [28:11]
F
Thanks for having us.
- [28:12]
H
Thank you for having us. Thank you.
- [28:19]
A
That's my conversation with Bishak Erzer, Mark Shieldrop, Mike Vardebinian, and Carrie Morwege. Carrie's forthcoming book is How We Think About Our Stuff and Ourselves As We Own Less and Use More. In the next two episodes, we'll explore whether AI is a complement to human work or a substitute that eventually replaces it. Please join us for those, and in the meantime, we would love it if you would please rate and follow Is Business Broken? Wherever you get your podcasts, that's where you can also check out any episodes
- [28:55]
B
you may have missed.
- [28:57]
A
Thanks for listening to Is Business Broken?
- [29:00]
B
I'm Kurt Nickish.