Actuaries and Autonomous Vehicles

In 2017 I spent 6 months on a Fulbright scholarship, in the desert heat of Phoenix and Tempe at Arizona State University. This was also the location where Uber and Google’s Waymo decided to test their artificial intelligence powered driverless cars.

Sunny weather and flat, pot-hole free roads provided an ideal testing ground for this grandiose experiment. In addition, Arizona’s light touch regulatory approach meant that the state ushered in a tech boom where the cities of Phoenix and Tempe were flooded with autonomous vehicles

Watching and driving alongside these “computers on wheels” I was fascinated by how they interacted with their surroundings, seemingly just like any other vehicle. Aside from the large sensor on the roof they didn’t act or look much different from other vehicles navigating the city. My 3 kids, sitting in the back, would often joke that the robot car was coming for us. Yet they braked and accelerated smoothly, stopped at red lights and didn’t even honk at any of my bad driving behaviours.

Robots driving cars! If you had told me 10 years ago that I would be living and driving in a city alongside robotic cars I would have said it was pure science fiction. But it seems we are now indeed living in a different world. One that is leveraging technology, especially AI, to change our lives at an exponential rate. It feels like Moore’s law has run amok.

Just think, for a second, how driverless vehicles might change the world – reduced car ownership, car design radically changing towards comfort and luxury, more senior drivers, reduced taxi/bus/truck/delivery drivers, hospital and emergency services unburdened and huge improvements to rush hour traffic – to name just a few.

Of course, the diffusion of new innovation is in reality never linear and straightforward. For example, Uber’s self-driving cars have recently come under huge scrutiny following a tragic accident where a pedestrian was killed, whilst crossing a road, in Tempe.

Tragic as this was, I still believe that autonomous vehicles are much safer than human-controlled vehicles. Sure, the underlying AI neural nets may not be 100% robust to cope with all situations, but let’s not forget that 100 people die per day in the US in an auto accident with human drivers behind the wheel. At the time of writing, Waymo’s autonomous fleet has self-driven more than 5 million miles without a fatality. In my opinion, the researchers that are predicting vehicle autonomy will create a 90% reduction in accidents by 2050 are probably not far off the mark.

So what does this mean for actuaries? 

Naturally, the auto insurance world is going to get turned upside down as these vehicles become mainstream. Considering it’s estimated that more than 90% of road accidents today are a result of human error, personal car insurance will be redefined as risk shifts from vehicle users to vehicle manufacturers and software/hardware suppliers.

Attribution of liability will become a much more grey area as shown by AIG’s survey respondents when asked who would be “most liable” in crash scenarios involving driverless cars:

Source: AIG (The Future of Mobility and Shifting Risk)

As the inevitable driverless world takes over, many traditional auto-related risks will no longer be as prevalent. Risks such as those caused by reckless or distracted driving, speeding, running stop signs/red lights, unsafe lane changes, tailgating and road rage will be replaced by new emerging risks such as malfunctioning software and cybersecurity.

The migration and ensuing calculation of risk will be particularly challenging during the “chaotic middle” transition period where vehicle owners and the AI software share responsibility for the vehicle’s operation and any resulting liability.

Given the number of deaths attributed to vehicle accidents at present (over 1.2 million people die in road crashes each year, on average 3,300 deaths a day) mortality curves may also eventually be significantly affected. For example, the spike in mortality rates that we currently see around the late teens/early 20s (the “accident hump”) may start to flatten out.

Risk, and hence insurance, is the playground of actuaries so surely they will be at the forefront of untangling and understanding these uncertainties as the autonomous vehicle landscape unfolds.

Mark Farrell FIA PhD

I'm a UK qualified Actuary and currently Programme Director of Actuarial Science at Queen's University Belfast. Prior to academic life, I spent 10 years working as a consulting actuary.

Click Here to Leave a Comment Below

Leave a Reply: