Legal Update: Between a rock and hard place

It is generally accepted that driverless vehicles will arrive on our roads in the not too distant future and, when they do, the road network will gradually become a safer place than at present. However, the term “driverless” is currently being applied both to the specific, short-duration assisted driving modes delivered by Advanced Driver Assistance Systems, such as lane-assist and self-parking, and to vehicles which in future will be fully autonomous all of the time.

Technology aside, the distinction is important for insurers and lawyers to think about. Incidents will inevitably occur and questions of liability arise according to the circumstances and the ability of the driver or the technology to respond. It won’t always be a question of driver error or technical malfunction. In some situations, the outcome may have to be pre-programmed into the controlling algorithms. What should underwriters be thinking about?

In the case of ADAS-equipped vehicles, it is likely the driver will continue to be liable unless technical malfunction is established as the cause, liability then passing to the manufacturer. Such vehicles will be in one of three modes (manual, automated and transition) and all drivers will be required by law to be able to regain control from the autonomous mode when required. Implicitly, the driver will be required to supervise the automated behaviour at all times, even if they are not actually driving.

Fully autonomous vehicles will, however, be outside the “driver’s” control, except perhaps for onboard programming. Supervision will either not be required or be unable to prevent an incident in the extreme cases where the vehicle itself is unable to avoid it. How the vehicle then responds will be determined by its pre-programmed algorithms and its in-built ethical solutions.

On this point, researchers from the Toulouse School of Economics, University of Oregon and Massachusetts Institute of Technology published a paper entitled Autonomous Vehicles Need Experimental Ethics in October. They commented on the need for “moral algorithms” to be devised for situations of unavoidable harm –
the assumption being that some situations will require autonomous vehicles to choose the lesser of two evils. To minimise harm might require the vehicle to sacrifice its occupants, not just choose between running over a child or hitting 10 people at the bus queue, which would give potential buyers pause for thought.

If injury or loss occurs as a result of the ethically approved decision of an algorithm, where harm to someone or something was unavoidable, who then is at fault? The owner, the manufacturer, the algorithm licensing authority, or a third party?



As with any assessment of liability, a root-cause analysis is required: how did the vehicle come to be in that unenviable position in the first place? Given the technology involved, it is likely to have been an extreme set of circumstances. Possibilities might include: a precipitating mechanical problem with the vehicle itself; a loss of control by another (adjacent) automated vehicle; two pedestrians simultaneously running in front of the car but in opposite directions; two or more vehicles converging simultaneously from various directions.

In each case, the liability position does not appear to be greatly different from now. The fact that the automated vehicle has had to make a choice between outcome A and outcome B is one of its functions, triggered by a preceding event to which liability can be attached.

Whatever the circumstances are, it is likely there will be far more detailed technical data available to pinpoint guilt or innocence of the other parties at the scene of an incident. Not only from vehicles: most pedestrians carry mobile phones, many of which already track speed and motion.

So, while the predetermined ethical response of a fully automated vehicle will be a factor in the final outcome of an incident, it does not seem right to suggest that it should at the same time determine liability. As an owner, however, this would be small consolation if your car decided that injury to you and your passenger represented the least damaging outcome in the circumstances.

Nick Rogers
Partner and head of motor, BLM