Car insurance is one of the oldest and best understood forms of insurance. More than a century’s worth of data can assign risk based on the make, model and value of a vehicle; the age and gender of drivers; the location where the car is garaged, and hundreds of other factors.

But traditional car insurance is founded on one core assumption — that there is a human in control of the vehicle. Whether the driver is a middle-aged woman in a $300,000 Ferrari, or an 18-year old man in a used $5,000 Camry, the insurance industry has enough data to assess the likelihood of that car and driver being involved in an incident. The insurers can assign a premium and determine what payout they may have to make if the vehicle is damaged, or the driver causes damage to another vehicle or property.

But what if the assumption of a person being in control is invalid? Carmakers around the world are developing technology to take the driving away from humans. These aren’t cars in the way we are used to. They are really robots that can carry people from one place to another, whose actions will be governed by software. While today’s cars offer many new safety and convenience features, such as cruise control, intelligent braking that slows or stops a car before a driver has the chance to react, and lane assistance to keep the car moving in a predictable line, they still allow the driver to take responsibility in case of an incident.

But we are fast approaching a time when cars won’t have pedals, gears or steering wheels. In fact, there won’t be any drivers — only passengers who tell the car where they want to go and then sit back and ignore what’s going on outside. This means driver error will be removed as a cause of incidents.

Today’s car insurance industry assigns responsibility proportionately to a driver when a vehicle is involved in an accident. But what if that driver is software? Let’s suppose two transportation robots collide on a street. How can an insurer assign blame? What if the software actually acted in exactly the way it was intended and the collision saved lives, avoided injuring passengers, and minimised property damage?

Could an insurer justifiably charge an excess or deductible to the owner of the vehicle if it behaved exactly as intended?

The other issue for insurers is that most predictions expect the number of car insurance claims to reduce significantly. Robotic cars won’t operate in a vacuum. Eventually, traffic management systems and cars will work together to route cars in the most efficient and safest way. That means when a car takes off on a journey, it will communicate with other cars on its route and adjust its journey to avoid potential hazards. For example, your car may take a less direct but safer route to avoid the risk of slipping — something a human driver today is unlikely to know how to avoid.

Even theft becomes less of an issue. What if cars can actively avoid being stolen by “running away”?

Car ownership is also expected to change as people rely more often on shared vehicles. So insurers will, potentially, have fewer personal customers and have to deal more often with companies.

If drivers don’t need to be insured because there won’t be any drivers, what happens to car insurance?

It evolves.

Insurers will turn to the providers of the ecosystem. Car manufacturers will be assigned responsibility based on the algorithms and software they create to drive our new robotic vehicles. A report by RAND  suggests that manufacturers’ liability is likely to increase and personal liability is likely to decrease. In the United States, some states, such as Michigan, are already passing laws that stipulate carmakers assume liability when driverless systems are at fault. The government of the United Kingdom has been considering similar laws.

If a vehicle and a human share driving responsibility, the insurance issues could become more complicated. The good news for insurers is that they are experienced at doing complex calculations of liability. The flip side is that the number of factors they may need to consider is different, and they lack historical data to build models.

Insurance is, at its heart, a data-centric business. New ecosystems that support robotic transport can collect a lot of data about how cars operate when humans aren’t part of the driving equation. And this will help insurers build new models as they shift their business from focussing on individual drivers to car manufacturers. Although it’s taken the car insurance industry decades to build its models, a sample of 2,000 fully autonomous vehicles can produce as much data as all humanity did in 2015, according to Dr. Chang Huang, co-founder and vice president of Horizon Robotics, who spoke on the topic of robotic cars during a keynote address at CES Asia in 2019.

It’s clear that insurers will need to model various scenarios and the probabilities of incidents resulting in a claim. Hence, the play of big data and analytics will be interlinked to calculate pricing, risk assessment and simulations based on the data.

Car insurers will need to adapt their data models and business practices as robotic transportation become more common. But they will have access to ample data, and the days when drivers are no longer required are still many years away. By engaging with the creators of new ecosystems and looking at new data models, car insurers can prepare for the changes automated driving will bring.