Instant Genius - The future of self-driving cars, with Jack Stilgoe

Episode Date: September 29, 2022

Jack Stilgoe, a professor of science and technology policy, talks us through the future of self-driving cars, upcoming changes to the law surrounding them, and how they could change society. Hosted on Acast. See acast.com/privacy for more information. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Ambition comes in all shapes and sizes. At First Citizens Bank, we roll with your goals because we're built for what you're building. Fit for your ambition for Citizens Bank. You said this place was steps from the water. We just haven't found the steps yet. How much did we save? Enough.

Enough to get lost. Or you could book a stay with Hilton. Welcome to your ocean front room. Just steps from the water. The Hilton sale is on now. Book on Hilton.com or the Hilton app and save up to 20% to get the stay you expected. When you want savings, not surprises.

It matters where you stay. Hilton, for the stay. Study and play. Come together on a Windows 11 PC. And for a limited time, college students get the best of both worlds. Get the unreal college deal,

everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft off 365 premium and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller. Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m.m.m.m. This podcast is sponsored by name, audio, and focal. Streaming has made music more accessible than ever, but true listening is about more

than ease. It's about quality. British audio experts name audio, alongside French acoustic specialist focal, combine handcrafted tradition with cutting-edge innovation and high-end materials, delivering digital precision with analogue warmth. So you can experience exceptional sound at home. Music just as the artist intended. Visit name audio.com to learn more. From BBC Science Focus magazine, this is Instant Genius, a bite-sized masterclass podcast for. I'm Alex Hughes, staff writer at BBC Science. Science Focus magazine. This week, I'm joined by Professor Jack Stilgole. He's an expert in science and technology policy. He explains the future of self-driving cars, the laws and ethical guidelines

needed, and how we are likely to interact with them when they do hit the road. So this feels like a very obvious question to start with, but could you explain what a self-driving or autonomous vehicle actually is and how they work? So, well, I can have a go. I mean, the sort of, the simple answer is that a self-driving vehicle is a vehicle where the human is no longer doing the driving or taking responsibility for the driving, and that that is being done by a computer. In terms of how they work, I mean, different companies have different approaches,

But I guess at the most superficial level, they are robots that move through the world by sensing what's around them, by working out what the things around them are going to do next. They sort of often do prediction of the movement of cars and other road users. And then they plan safe paths through that world and then act on it. So the sort of easy explanation is that you have a computer doing what a human does, but rather than eyes and ears, you're replacing them with cameras and other sensors, and rather than a human brain, you've got an artificial intelligence. But the more you look into it, the more complicated it becomes, and the more you realize that actually computers and humans don't and won't ever drive in exactly the same way.

So in the same way that a human that's been driving for quite a long time and they learn the ways of the road, I assume that's a similar sort of way that a self-driving car would work, it builds on its knowledge. Yes, with the added advantage, the self-driving car developers would tell you, the added advantage that when one of their cars learns something, they can all learn it. So rather than having one human being learn things, either by being taught them or learning them from experience, you can have whole fleets of cars learning together, which might go some way towards explaining why we've seen

pretty rapid progress in the development of artificial intelligence for driving cars over the last five, ten years or so. So what is the need for self-driving cars? Is there something that could drastically improve our society, or is this maybe more on the side of a gimmick? So if you talk to the people that are developing them, and I'm a social scientist that does that for a living, I've been speaking to a lot of the people developing the technology who are excited about it, as well as the people who have concerns about it, they would say the sort of major justification is one of safety, which is that even though driving is getting safer over time as technology improves and roads improve and. you know, we enforce rules better. Driving is still one of the riskiest things that we do.

Being on the road is one of the riskiest things that most of us do. And if you could take away some of the human fallibility and replace it with a reliable computer, then you should have massive improvements in road safety. So at the moment, about pretty much every year, a million people, die around the world as a result of crashes on the road, which is, you know, at one level, just a public health disaster. And so the AI enthusiasts would say, well, you know, we could solve that problem. We could take away human error and replace it with a computer.

In reality, it's not quite that straightforward because a lot of those deaths are happening in places where we might not see self-driving vehicles for a very long time, maybe even ever. But that's the sort of strong justification. The people developing the tech might also say that there are benefits that would come from efficiency, right, that maybe if you have computers driving vehicles, they would be able to come into city centres and then not have to park in those city centres and then go back out. again, which might create nicer spaces for us to live in. But those sorts of benefits are a bit more speculative. And you mentioned it a little bit there, but is it likely that this automation will

change our view of how cars can be used? I mean, for example, will we become a society of backseat drivers maybe? And, you know, people don't own cars as much as they do now. If a transition to self-driving vehicles is anything like as profound as the, you know, the technological optimists would say, then yes, as with every other transition in how we get around, it will lead to radical changes in our lives, in our lifestyles, maybe in our livelihoods as well, you know, who makes a living from driving and all the rest of it. Now, in terms of predicting what direction that's going to go in, that's pretty difficult because it depends on the assumptions that you make in your model. You can imagine a future in which,

you know, we have radical cuts in carbon dioxide emissions, for example, as we have a transition to electric self-driving vehicles. But equally, you know, if you have self-driving internal combustion engine vehicles and people are just getting into them falling asleep, then maybe that means that people move further out of city centres. Maybe they don't mind a two-hour commute to work, so you might see a sort of sprawling of cities eventually. And one thing that we know about past technologies for getting around is that they start off, you know, changing just how we interact racked on the road, but also, you know, within decades can reshape entire landscapes so that, you know, cities where we live, our surroundings are affected by our mobility choices.

The UK government has now stated expectations to see self-driving cars about needing drivers by 2025. Do you think that's a realistic time frame? We need to ask, well, sort of in what conditions on what sorts of roads for which people, are they likely to be just constrained to motorways, for example, which are already a really neat and tidy environment where lots of road users are banned from being there. They're already quite a safe environment. So we might see a self-driving vehicle being able to cope with a motorway scenario in quite a sort of predictable way by 2025. We might see those vehicles owned by people that buy high-end vehicles for whom this is just another driving option.

That's rather a long way from the, you know, we'll have brand new ways to get about in our computer-driven cars going wherever. wherever we want to go. So, you know, there are some places in the world where you can already get in a vehicle that doesn't have a driver, which is an extraordinary thing. But there are also reasons why that technology is not just immediately available to everybody, right? They tend to be stuck in particular places. And in a similar sort of viewpoint, would there need to be a change in the laws of the road for this expectation. Obviously, you're saying that these cars will only be appearing in a very small number in certain areas, but do we need to readjust how we view the legality of the roads? Yes, I mean, absolutely, in a few ways. And this is

exactly what government has been doing the last couple of years, working with the legal profession and thinking through questions of engineering and ethics. The first and I guess the most important task legally is to work out questions of liability and responsibility. So being really clear on who is in control of a vehicle so that if and when things happen, we know whether a driver is liable or a company operating. or manufacturing a self-driving vehicle is liable. And for the history of the motor car, we have located liability with the driver.

And there's a big change that happens if you allow a driver to, for example, check their emails on their phone or even maybe in a few years' time to go to sleep behind the wheel, in order for that to be a legal thing to do, and in order for people to be able to trust it, it's just important,

then you need to have laws that hold companies liable. But there's also a more subtle thing, right, which is, so it's not quite laws, it's the more general category of the rules of the road, which we are taught at school, right, the green cross code or however you, you phrase it, right? The question of who has precedence in what sorts of environments, who should take responsibility for guaranteeing the safety of vulnerable road users.

In a world where self-driving vehicles become common, those sorts of things will need real thought because a self-driving vehicle will not just interact with other road users in the same way as a human driver would. And so you need to reconfigure the rules of the road. And as we've seen with recent changes to the highway code in Britain, those things are not without controversy. You change the rules of the road and suddenly you're prioritising one group of people over another group of people. And if we were to change these rules, I mean, as we said before, the self-driving cars are learning and as a group and better understanding the roads, is there any concern that these self-driving cars would learn the maybe slightly more chaotic driving of people and then

try and have to work around that and try and understand the human error side of things? Yeah, so if you're developing a self-driving car and you get stuck in this model where you're just trying to mimic a human being, then you risk developing a type of, technology that yes, can learn to drive and do something, do something magical, but maybe doesn't realize the benefits of having computer-driven mobility. So, for example, you might learn all of the transgressions that we normally live with as we drive around. And this has already become an issue with Tesla in the US, whose limited self-driving technology, had a mode that meant that you could, that you didn't have to stop at a stop sign in the US, and a lot of their junctions

are controlled by stop signs rather than traffic lights or roundabouts. And Tesla's response was, oh, well, everybody does it, right? Everybody just rolls gently through a stop sign rather than stops automatically. But understandably, there was some outrage from others that that company was programming in intentional rule breaking. Now, we actually, as part of our driverless futures project at University College London, we did a survey of members of the public asking people if they thought that self-driving vehicles should be able to break the rules of the road in certain circumstances.

And there was a clear majority that suggested that no self-driving vehicles had to behave them. even though people recognize that human drivers don't always behave themselves, to give that sort of sense of predictability. But it's one of those things where the rules are sort of already, or some companies are attempting to rewrite the rules of the road in order to suit them. And unless we think about them and debate them, then we risk slipping into a new of rules of the road that it turns out suit the self-driving car developers and maybe aren't so good for the rest of us? We mentioned it a little bit there, but are there certain standards that

would need to be met for self-driving cars? I mean, for example, there was a survey recently that found that a lot of people would want to see self-driving cars labelled, you know, so people could tell that a certain car didn't have a driver in it? Is there standards like that that we would need to meet to make people feel more comfortable on the road? So I would say certainly yes. And the labelling question was one of the questions that we put in our survey, which was something that we thought was just particularly interesting because, you know, in some cases you can see a self-driving vehicle and think,

well, that thing looks new and it's weird. And therefore I know that it's something different. And I might choose to change my behavior around those things. but in some cases we might be completely unaware that a car is driving itself. And the question is, you know, should we know? And what might our rights as other road users be to know that a car is being driven by a computer rather than a human? And does it matter? It's one of those examples of an area where there aren't actually any standards at the moment.

So most companies are just doing their own thing, and some are behaving more responsibly than others. But everybody knows that standards will have to emerge, and there may even be, you know, competition between different standards and different countries may take different approaches. There's a lot of activity going on within the industry about how to standardise the behavior of self-driving vehicles and the rules that, control what they can and can't do and how they do and don't do and don't look. I would like as a social scientist that process to be a more open one so that we have a democratic discussion about what those rules should be rather than the industry just writing its own rules and marking its own homework.

Do you mention a little bit there about the differences in different countries, but do you think is the UK in a similar pace to other countries in terms for adoption? of autonomous cars or are we a little bit ahead, but a little bit behind? So British policymakers would say, you know, this is an area where Britain can lead. I think given, you know, we lost our manufacturing capacity for motor cars a few decades ago. And we have not been quite so bullish as certainly some places in the US and increasingly China, actually, which is seeing the rapid acceleration deployment of these vehicles in cities, we definitely don't see as much activity happening in Britain. And where activity does happen,

it tends to be of a different sort, which is actually a much closer collaboration between self-driving vehicle developers and public transport, right? So we're likely to see more of those sorts of experiments. But where I think Britain can play a really important role, which historically it has done with new technologies is in this process of standard setting, is helping to write the rules that shape the responsible development of the technology. And I think Britain does need to play an important role,

even if it's not our car companies or our tech startups who are the ones that are leading the way. Not to take this in a very morbid direction. But what happens if a self-driving car was to hit someone or crash into another car? I mean, who is the who is at fault in that situation? Or is that one of the many things that is still up in the air? Two things. I mean, we already have some experience of people being killed in technologies that were driving themselves

or where the users thought the technologies were driving themselves. So we have some examples. And those rules do still need to be written, which is what the British government regards as a priority, is clarifying exactly those questions of liability. The most telling example that we've seen so far was in 2018, when a prototype Uber, self-driving Volvo, ran into and instantly killed a woman called Elaine Hertzberg in Arizona.

Uber were testing their technologies on the streets of Tempe, which is part of the sort of Phoenix area, wide open streets, very driver-friendly, not the sort of place that you want to be a pedestrian. And what that case revealed, the finger-pointing, the blame game that happened after it suggests something quite, well, it is quite unpleasant about how these discussions might play out if we're not careful, which is, you know, initially the pedestrian got

blamed, then people wondered whether it was the manufacturer of the car, people wondered whether it was Uber's fault. The eventual person that seems to have shouldered most of the blame was the so-called safety driver, the person behind the wheel of the self-driving vehicle who was supposed to be taking responsibility and watching what was going on. That's how a lot of self-driving vehicle companies have been testing their technologies, right, by locating responsibility on the shoulders of this safety driver. But there's, you know, the question that that raises is how we train the safety driver, what we expect of them, whether we think that their job is an easy one or a hard one,

and whether we learn from the history of the use of things like autopilots in airplanes that tell us that if you put a human on the loop, as they call it, that human can quickly become very, very complacent. So even if they know that the technological system does not work, they will often behave as though it does because, you know, an AI works absolutely fine right up until the moment that it doesn't. The other crucial thing about what happens after a crash is knowing and learning from that. And there are some real concerns there about how we know what an artificial intelligence is doing and why. So it's a question of transparency and explainability, the computer scientists would call it.

And what we expect in terms of the companies operating these vehicles sharing their data with regulators, crash investigators and other companies so that the process of learning can be a collective one and that the whole system can get safer over time rather than companies hoarding their data, claiming that it's a trade secret, which might mean that we see a lot of unnecessary deaths. We've spoken a lot about the legal side of things and the ethical side of things, but another topic I think is quite interesting is about self-driving cars and their improvement within cities. Do you think it's the, I guess, the best way to improve transport within a cityscape, or is there an argument that improvements in public transport are a better route to go down?

The one thing I would really caution against is saying that self-driving cars will do this for cities, right? Are the one good thing or one bad thing? I think, you know, the technology is sufficiently immature that it could be shaped in some different ways. It could be made to be really good for cities, it could be if it's allowed to develop thoughtlessly, it could be really bad for cities. So, for example, if you're a city like London, where I live, your concern, if you're a transport planner, is trying to get cars out of the city centre because congestion in the city centre is an enormous problem. So if you think, if that's your priority and you're worried about single occupancy vehicles, people driving into London, and it's just them in the car.

Then the only thing that's worse than that is a zero occupancy vehicle. The idea of an empty car using up road space, shuttling around the middle of London, is not something that you want. So a city has to think about self-driving vehicles, I would argue, in terms of they have to start with what they need from their transport system, whether that's prioritising you know, decarbonisation or prioritising efficiency, getting people onto shared transport to reduce congestion, and that depends on, you know, the shape of your road network and what your other available options are. So if your London, for example, you might have particular

holes in your transport network, particular transport deserts where, you know that residents struggle to commute, for example, you might think, well, yeah, maybe a self-driving, vehicle system is going to really work for us in that well-defined environment. But that's a very different thing from just saying, great, let's just open up our streets to self-driving vehicles and hope for the best, because without careful thought, that experiment could end really badly. I don't want to make you have to predict something else, but let's say skip on to 2035 or even later, Do you think the roads are going to actually look any different? Or is this very much going to be an incredibly slow implication

that will take years and years for it to be noticeable? So I'm cautious about predicting that things are going to be very, very different because, as I said earlier, it depends on where. So for some people, if you're in San Francisco or Phoenix at the moment, you can't deny how extraordinary it is being able to get in a car without a driver. Now, that's genuinely new, and we've seen the scaling up of that technology in those situations pretty quickly, and that's pretty impressive. It may well be, for example, that the insurance industry gets involved and says,

actually, this technology is so much safer that we are going to, make it cheaper for you to use a self-driving vehicle than to drive your own vehicle and the insurance premiums on human driving go up and maybe change happens really quickly. As changed happened really quickly, you know, in the early part of the 20th century with the arrival of the motor car. But the arrival of the motor car also required cities and town planners to change their rules to get pedestrians out of the way. And that wasn't without controversy either. So it depends. We might see a slow rollout of the technology in which the beneficiaries of the technology actually are rather small. So there might be people

in rich cities who already have access to quite a lot of ways of getting around anyway. They might be rich consumers of cars for whom this is just another gimmick. So it really depends. My worry would be if the technology is developed in that sort of thoughtless way in which it either, you know, it basically follows the money. Then other people might look at the technology and say, well, this isn't doing anything for me. And therefore, you know, I'm not sure I want these things on my road, especially if they are. are running over the odd person or if they are clogging up the streets, then the question of who really benefits from self-driving vehicles will be a really important part of the social

contract. And then it might be that they become another novelty that's sort of consigned to the dustbin of history. Now, I think actually we'll see the development of self-driving technology in some boring but profound ways. So here in London we already have driverless tube trains, we already have driverless light rail, the Docklands Light Railway. And we might see some improvements to that sort of technology that actually, you know, it delivers things for people as part of a public transport system, but it's rather different from the extraordinary novelty of seeing a computer do the driving on an open road in an uncertain, unpredictable environment. Thank you for listening to this episode of Instant Genius.

That was Jack Stillgo, examining the future of self-driving cars and upcoming laws and regulations announced by the UK government. The Instant Genius podcast is brought to you by the team behind BBC Science Focus magazine, which you can find on sale now in Supermarkets. and newsagents, as well as on your preferred app store. Alternatively, you can come and find us online, sciencefocus.com. This podcast is sponsored by Name, Audio and Focal.

The texture and emotional depth of music can be lost through digital sources or poor signal. Name Audio believes you can have digital precision with analog warmth. Alongside French acoustic specialist focal, Name creates high-end audio systems combining innovation with craftsmanship so you can listen to music, just as the artist intended. Discover more at nameaio.com. In California, staying compliant means watching the state laws and the city rules at the same time. And no wonder it feels overwhelming.

Meal breaks, rest breaks, wage rules, constant updates, it's a lot. And that's why Southern California businesses rely on Guardian HR. They're local in L.A. and they understand this community, and they help you stay compliant to avoid costly missteps. You get accurate payroll software and a real HR expert who keeps you ahead of issues. Get your Southern California business protected at GuardianHR.com.