The myth of perfect safety in a world transported by autonomous cars has endured until today. A pedestrian in Tempe, Arizona, was killed by an Uber autonomous car with a driver at the wheel on Sunday night. The company suspended testing throughout the U.S. and Canada as a result.
Perfect safety in complex systems is impossible, yet autonomous car developers have insisted that no one could be hurt by a computer-controlled car as a matter of faith. Visions of perfectly ordered autonomous vehicles avoiding humans without fail set safety expectations far too high, and the industry, not just Uber, will pay for dealing in superlatives instead of reality.
Autonomous cars will improve their safety performance, just as overall driving safety has continuously increased since it became a regulatory priority the 1960s. The U.S. currently enjoys a traffic death rate of 1.25 per 100 million vehicle miles driven. Waymo, Google’s autonomous car group, had counted four million miles driven by November 2017. Despite this lack of comparable miles driven, absurd claims, such as “Google’s Self-Driving Cars Are Ridiculously Safe” have become common
Reality bites hard. Uber CEO Dara Kosrowshahi tweeted his regrets to the victim and their family. He needs to visit the site of the accident, get a complete review of the circumstances, and reset safety expectations accordingly. This may require the company fallback to emphasizing human drivers over self-driving cars for the time being.
Autonomous car safety is in its infancy. Now that a real-world death has proven the technology needs to advance substantially to best human driving safety, ridesharing and delivery services will need to reconsider the scope of testing, and the timelines promised for the general introduction of the technology.
This is not an-anti technology view, it’s a realistic product safety criticism that will deeply affect people’s views of the companies hoping to deliver autonomous vehicles over the next couple years. Stop marketing autonomous vehicles until they can beat human drivers consistently, but keep developing and testing in safe situations.
As autonomous cars become available, we think many consumers will depend on local influencers who experience the rides, share their stories, and package mobility for the consumer’s comfort level. If a person is afraid of being in a self-driving car, there is still an opportunity to provide products or services delivered by autonomous vehicles. The solution will be person-to-person marketing that finds those first uses for smart vehicles that consumers will accept.