How insurance is driving the future of autonomous cars

speaker-0: Hello and welcome to the Digital Lighthouse, where we get inspiration from tech leaders to help us navigate the exciting and ever evolving world of digital transformation. I'm Zoe Cunningham. We believe that meaningful conversations can illuminate the path forward, helping us harness the power of technology for innovation, scalability and sustainability. In this episode, I'm delighted to introduce David Williams, who's a senior executive and non-executive director with over 30 years experience in insurance.

He's currently the chair at the Fire Protection Association and a non-executive director at claims. David's previous roles include working at AXA Insurance UK, where he led AXA's work on connected and autonomous vehicles, including work with five government-backed consortia. He was also the chair of the Association of British Insurers Motor Committee. In this episode, we'll be exploring the future and the present of driverless vehicles.

and how we can utilise insurance tech to improve our daily lives. David, welcome to the Digital Lighthouse. Thank you. So can you tell us maybe a bit about the journey?

You you've had this kind of long and varied journey to get to where you are today. What are the kind of significant, you know, milestones and roles that you've had on that journey? speaker-1: So like most people, I stumbled into insurance and found I enjoyed it and loved the people and therefore stayed in it. And I've moved around the country and I've done lots of different roles whilst mainly working for one company.

The whole connected and autonomous vehicle piece comes, suppose, from, you know, I've always had an interest in cars. I was on the board of Thatcham, which is the motor vehicle research institute. But then we got to about 2015 and we were just doing one of these sessions where you think what are the future risks to the business? know, not pricing risk, but what's the risk to the business?

speaker-0: As in catastrophic risks that will wipe out the business, those kind of speaker-1: Or deliver a completely different market that we wouldn't be ready for. one of the things I said was, well, what if driverless cars come? You know, make a lot of money from, well, we don't make a lot of money. We get a lot of premium from motor insurance.

So what if people aren't driving cars anymore? And we asked ourselves the question, what do we know about driverless cars? In reality, was we knew nothing. You know, we'd maybe seen them in a movie or read about them and, you know, got excited about them.

And this was 2015 and the government were just starting to ⁓ fund consortia whereby they would try to get cross-industry groups together to develop different areas of technology and one of those areas was connected and autonomous vehicles. So we had this great idea. What we'd do is we'd see if we could join one of these consortia. We'd be then hanging around lots of clever science people.

that did know lots about driverless cars and we'd naturally be able to learn. So we thought that was good idea. We thought we'd fail miserably, but what we, I suppose, didn't understand is that if you work for a very large insurer, your brand has a value, which people like to be involved in development work, but also insurance is fundamental when you're talking about motor vehicles. It's still the biggest cause of deaths of males under the age of 25 globally and therefore people wanted us to get involved.

speaker-0: Because, I mean, in many ways, you must have expertise within the insurance industry of what can go wrong with a vehicle, right? speaker-1: Yes, but going back to Thatcher and the Motor Vehicle Research Institute, it was great if there was a new vehicle coming onto the market, we'd crash it into some things and work out how much it was going to cost to repair. insurance relies on data. We look at past patterns.

When you're talking about driverless cars, automated vehicles, it's a whole new world. Nobody's got any data or nobody had any data on that. So one of the things that we thought we'd do is again by hanging around with these you know, really intelligent people and getting involved, we'd be able to build some data. And before you know it, over sort of three, four years, we were involved in five government funded consortia, all looking at connected and autonomous vehicles from different aspects.

And I was leading it because my company wanted me to, and also it was really interesting. so I then ended up ⁓ chairing the Association of British Insurers. autonomous driving group and the ABI motor committee. And I found myself just being asked to get involved in lots of things from really interesting things like going around and seeing tests, seeing cars crash into each other or avoid things like that, to really quite dull stuff like discussing, you know, the minutiae of regulations in British standard testing and stuff like that.

So that's where I ended up really where I was in terms of I was seen as the sort go-to person within the insurance industry to speak to on driverless cars. speaker-0: which is probably not somewhere you could have possibly imagined when you started your career. speaker-1: Not at all. As I said, my dad was very into motor vehicles and I think he felt badly let down when I joined an insurance company.

He would rather I'd picked up a spanner or a spray gun. I do remember actually the only time he took interest in my career was when I got involved with Thatcham initially. But I think the reality is when you're looking at a job, there are going to be some bits that are more sort of processed and a bit boring, let's say. But what I think you need to do is you need to look for the exciting and interesting bits within that job.

therefore if you can balance that within a role, it's much easier to commit to it and do your better job overall. speaker-0: Yeah, definitely. And I think tech is a great industry for that. There's so many exciting.

mean, and no one has a job, like you say, that's all the exciting bits, but there are a lot of exciting bits out there. Absolutely. So maybe could you set the scene for us with what kinds of automated driving are already on our roads? Because actually there's a whole continuum, isn't there?

There is. Fully autonomous to, I mean, I guess even looking back 10, 20 years ago, there were aspects of it within our vehicles. speaker-1: Yes, so there are these things that refer to as the SAE levels of autonomy and they run, you can argue from zero through to five and one would be sort of cruise control which has been around for a long time and level two is when you get into sort of adaptive cruise control where you you can be behind a vehicle and it'll automatically slow down when the vehicle in front slows down and things like that.

People think that Teslas are autonomous They're not. They're great technology, but generally they're level two. So level two is very much their driver assistance. In fact, they've recently been banned in some countries from using the expression autopilot because it's not autopilot, it's driver assistance.

And I get a bit obsessed about that because it's really important for people to understand that up to high level three, probably level four, The car might be doing a lot of the work, but you need to remain engaged. It's not until you get to level four and level five, which are the true autonomous levels, that you can just completely relax and do more than taking your hands off the wheel. So at this point in time, we're just about dipping into level three with... I suppose I'd call it sort of adaptive cruise control on steroids where there's a little bit of potential lane changing, but it's not true autonomy.

And certainly in the UK, you are still required to be available to step in if the machine decides it's, yeah, I can't cope and I want you to drive. speaker-0: And hence you are very clearly accountable as the human in the machine. so what kind of approval process has there been to get to this cruise control that's out there already? speaker-1: Yeah, it goes through two different phases.

there's vehicle type approval, which applies to mass produced vehicles. So, know, cars generally, you can sort of get around it if you're prototyping a very small number. And that's where a lot of the test vehicles come from. But you have something called UN ECE, which is ⁓ a UN committee that considers specifications and capability.

and signs of vehicle off effectively as being considered safe generally. speaker-0: And so that's what like a couple of years or months or what's the speaker-1: yeah, nothing proceeds quickly though, so yes, probably a couple of years. But even then, once the vehicle has been type approved, it then needs to get local permission. Now, a lot of the time, it's just like UNEC, you're setting the rules with regard to colours of indicators, reversing lights, things like that.

But when you get into automated driving, then local ⁓ countries, individual countries need to set their own regulations to allow them to be on the road. And that's the case in terms of So legislation was required to allow a couple of vehicles to have that sort high-end level two, base level three. So they'd been approved and were available. So there was a Mercedes S-Class, unsurprisingly, I suppose, and that was available in Germany probably for about a year before it was allowed on UK rates.

speaker-0: Yes, all right. OK, OK. So we're getting closer. Yeah.

But it sounds like we're not actually there in terms of what you would consider autonomous, you know, cars on the road. speaker-1: So I don't think we'll ever see, or maybe not in my lifetime, ⁓ when you talk about autonomous you probably talk about level five, you talk about a car that can drive absolutely anywhere. speaker-0: So what I'm talking about is I actually haven't passed my driving test and I was hoping that I would kind of be redeemed in that, ⁓ look, I didn't actually need it.

I can now just go and sit in a car and the car will, you know, like getting on a train, the car will take me wherever I want to go and I can switch off from. speaker-1: You will get that in the UK, I believe, this year. Now, it might still be in pilot stage and you might have to put up with the strange experience of a safety driver sitting in the front and not actually doing anything. so Waymo, who are known globally, owned by Google, they are trialling in London from September, I think it is.

OK. And Wave, who are a UK firm who took a very different approach, ⁓ They're on the streets of London now out there testing. you can't, I don't think you can actually get in one of their vehicles yet, but what they need to do as well as all the, you're buying the keyboard ⁓ machinations, they need to be out there driving around gathering data. So that's what they're out there now.

speaker-0: Yeah, so it's not just about kind of theoretically proving that you're car can- speaker-1: This is the fine-tuning to prove and then those vehicles will be effectively licensed and so it will be in a taxi environment rather than owning one yourself but they will be available in London. I'm confident this year in limited numbers and then probably more so in 2027. speaker-0: That's very exciting. But you kind of mentioned in 2015 when you started to get involved in this area that the government was kind of starting to set up these consortia and ⁓ invest in this as a direction for the country.

How good would you say the UK government has been at supporting these kind of tech innovations or not? speaker-1: I think they've been excellent. I genuinely believe they've been excellent. In terms of numbers, I think they talk about a figure of 600 million invested so far, but that's probably joint government and industry funding.

But they have really been trying to accelerate it. They see the benefits of the UK being at the forefront. What tends to happen a lot in the UK, particularly in technology, is we come up with these great ideas and then Yeah, the firms get poached by Americans or, you know. speaker-0: Either the individuals go or they buy the whole thing.

speaker-1: So I think the fact they've been prepared to commit some investment has been great, but also in terms of whilst everybody seems to think that the US is so much further ahead than the UK, I would dispute that. think if you, I'm coming from an insurance perspective, but you've got 50 states with completely different legislative frameworks. So you will find that a particular firm is doing a lot with automated driving. in one city and one city only.

Whereas I think what the UK have tried to do and we've now had the Automated Vehicle Act which came into law I think it May 2024 but even before that we had what was originally the modern transport bill and then the Automated and Electric Vehicle Act and things like that. We've tried to set out as clearly as we can what the legal framework is going to be to encourage firms to rather move away from the UK, but to come to the UK. And certainly if you read what Waymo have been doing lots overseas are saying, I think they really appreciate what the government and legal structures are trying to do in terms of set things out clearly.

speaker-0: Right, because it's one of the preconditions for investment, If you don't know what the conditions and criteria are going to be, you don't know whether your investment is going to pay off or not. So you kind of have to have that to start with. And does that mean that we will see things rolling out across the whole country rather than London only? speaker-1: I think it varies.

Certainly one of the original benefits that we talked about was bringing mobility to isolated and elderly people. know, little villages where they haven't even got a bus route, very elderly people, that sort of thing. speaker-0: And it's got worse, hasn't it? are fewer bus routes now than there used to be, so hence more isolated.

speaker-1: So the idea of being able to summon an autonomous vehicle sounded like a great idea and I think that's still on the cards. The problem is you do have to spend a huge amount of money. So WAVE, for instance, the UK firm, they've just raised £1.1 billion to cover development costs and therefore naturally you're going to focus where the market is.

And that's why it's not that London's easier. London's not easier. So I think it's a case of you've got the size of the market and the opportunity there combined with if you can get a vehicle to drive around some of the roads in London, then my word, it can drive anywhere. Do see what I mean?

So it will eventually spread out. inevitably, just like when Uber were launching their driven taxi service, initially it was just in the very large markets. speaker-0: But hopefully, at least the legislation is the same. speaker-1: Yeah, the legislation will be completely standard, speaker-0: Yep, just a question of rolling it out.

So you've seen 30 years of technology changes across the insurance sector in general. And what would you say have been the most significant, you know, high level changes to technology that have affected insurance? So I suppose either things you're insuring now that you weren't before or ways of doing business that that really are quite revolutionary. speaker-1: I do go back to a time when everything was on paper and really some the green screens that would just be a financial transaction and that would be all.

think what we've seen now is, and I think it's been good generally, is there's much more complexity. So genuinely I remember when I started in insurance there were two rates for buildings anywhere in the UK. Just two rates. And now we get sort of motor insurance where you'll have 280 different rating factors.

that's... speaker-0: the complexity of the models. speaker-1: and understanding the risk more, because clearly there's more than two different types of houses in two different types of neighborhoods and locations and individuals in the UK. I think it's, technology has enabled us, know, machine learning was massive in terms of being able to liberate actuarial time to focus on other things, but in terms of also availability of data.

So if you think of the number of publicly available databases which are really useful and then on top of that there are a large number of firms that make very good money by selling additional data and services. So that's the biggest transformation I think in terms of that complexity which hopefully the customer never sees. And it really is a case of rather than being just dumped in a couple of very big buckets we are now trying to understand the risks and needs presented by an individual.

So you're virtually getting down to a pool of one, as we would say, which is interesting when the whole principle of insurance is meant to be levelling things out and sharing across a broader pool. But we do understand much more what individual people want and also the risks that they run, which enable us to both price and also tailor the product to their needs much better. speaker-0: Yeah, I mean, and that's absolutely fascinating because I imagine at the same time, the user experience, you're getting a more complex, like you say, product or service from the insurance broker.

But at the same time, the response times must be quicker, right? Actually, you've not got people going off and having to do loads of sums and research your situation. speaker-1: Yeah, I mean, so I hate to always go back to motor insurance, but a typical insurance company will do sort of 20 million quotes a day. You know, if you're doing 20 million quotes a day, that shows you how rapidly.

And when we talk about external databases, access time is all important. Because if it takes two, three seconds to send a message, get the data back, well, it's too long. Because people are expecting pretty much an instant quotation. and with all these things coming together.

So it's not perfect, but I think I would be amazed if I'd sort of woken up after a 30 year sleep from how we were doing things back in the day to now just absolutely be amazed how much we've progressed. speaker-0: Brilliant. And then driverless cars this year, it can't get any better. Indeed.

Can we just finish with a fun question? So there's a well-known problem called the trolley problem, where traditionally it's about a train on two tracks. And if you're the driver and you will kill a varying number of people, depending on which way you go, what's your moral obligation to change tracks or not change tracks? And it brings up questions of is action different from inaction?

maybe could you formulate it in terms of driverless cars and then tell us the answer. speaker-1: Well, yeah, so, and people like to take this example of, you've got a young mother with her children, their lives ahead of them, pushing the pram across the street, and the only way to avoid hitting them is to veer off the road and there's some elderly pensioners on a bench or in a bus queue. So what decision does the car make? speaker-0: And how does it do that?

speaker-1: And it's a great question in terms of, because everybody always asks it, but the easy question is there is no way the car is ever going to make an assessment as to the value of individual life. Yes. It also, to a degree, suppose, shows an underestimation of how really good a lidar detector, for instance, is in terms of you'd have to be going at one hell of a speed to be able to get your pram and kids out in front of the road without the car noticing it. But the dull answer is that the cars will not be looking to assess the value of life.

What they will be looking to do is avoid accidents at all costs effectively and to comply with the law. And therefore what they will do is they will not be looking to come off the road and crash into a bus queue. They will be staying on the road. They will be automatically slamming on those brakes and hoping that they are able to stop.

it's a case of, suppose, rather than making a value decision, which in the trolley problem, you have to either choose, do you flick to send the train down a different track or not? There is no such decision being made. The vehicle will do its best to avoid the accident and comply with the law. And that means staying on the road as intended.

But I genuinely, these sensors, they can see around corners. They can see 100 meters ahead. Statistically, it's already established that... But also in terms of safety, drivers, cars, versus manually driven.

90 % of road accidents currently are caused by human error. They are much safer. And therefore it's an interesting question. I think there'd be all sorts of legal problems if they started trying to assess the value of life.

speaker-0: who's on the pavement and expanding. speaker-1: there would be outcry. Mercedes made a stupid comment at one point that they were asked the trolley question. said, well, what they would do is they would concentrate on protecting their drivers, their owners, which again, you can understand, but even that got a lot of criticism.

So it's about being as safe as you can and complying with the law. speaker-0: And actually, like you say, the number one rule of the law is stop. Is stay where you are and stop. And actually, it's almost a human understanding to think, ⁓ you know, you could imagine swerving if there was a dog or a child.

could imagine that happening. And what would that mean? Whereas actually, all the car is going to do is say something is in my way. Yeah.

I'm not going to crash into that. And that actually doesn't matter if it's a child or a dog or a tree. It's still going to stop. So that's simple.

speaker-1: Exactly. Very much. speaker-0: Well, thank you so much, David, for coming onto the Digital Lighthouse and helping us shine like brothers. This Digital Lighthouse episode was edited by Steve Folland and produced by Patrick Anderson.

speaker-1: Thank you. speaker-0: The theme music was written and recorded by Ben Balo. A huge thanks to our sponsor, Softwire, for their continuing support from the inception of the show in 2019 to the present day. If you love the podcast, please let us know with a rating and review on your platform of choice.

We're always looking for feedback to ensure we're making the best show possible. And if you'd to take part, please drop us a line at thedigitallighthouseatsoftwire.com. You've been listening to The Digital Lighthouse with me, Zoe Cunningham.

Thank you for sharing your time with us and stay safe on this wild technological ride that we're all on.