Canadian Underwriter
Feature

Who’s at Fault?


June 15, 2018   by Adam Campbell, Senior Associate, Human Factors, -30- Forensic Engineering


Print this page

Who is to blame when a human operator could have performed better than their self-driving car?

The widely reported fatal collision in Arizona involving a pedestrian and a late model SUV outfitted with autonomous driving technology has raised questions about the technology’s safety and readiness for today’s roads.

While there are both technical challenges and ethical questions stemming from this unfortunate event, one of the most pressing among them will be in determining the extent of liability to attribute to the autonomous system, regardless of whether there are allegations of its failure.

Determining the cause of a collision is crucial to understanding the liability landscape.

Close-up of robotic hands on steering wheel while driving autonomous car. 3D illustration.

When humans are behind the wheel, there are known limits and ranges of behaviour to be expected of normal, attentive drivers. The consequences of departures from normal behaviours can then be studied directly through computer simulation of alternative scenarios.

However, when an autonomous vehicle is involved in a collision, there is – at least, so far – no clear understanding about the standard to which the technology should be held. If there is no benchmark for the expected capabilities of an autonomous vehicle, how can the extent of liability be properly attributed? These and related questions should be at the forefront of the legal and insurance industries as they begin to navigate this uncharted territory.

The intent of autonomous vehicles is essentially remove the human element from the system. It may therefore be a prudent test of liability in autonomous vehicle collisions to determine what a human operator of the vehicle would have done had they been engaged in the task of driving instead of self-driving technology. In effect, normal human behaviour may serve as a logical benchmark for the minimal capabilities expected of autonomous technologies. Simply put, if self-driving cars cannot behave as well as a human driver in a given situation, there is limited justification to fully remove humans from the equation.

Read the full article in the Digital Edition of the June 2018 Canadian Underwriter.

Click here to subscribe to Canadian Underwriter, available free to qualified industry professionals.