Logic, Not Emotions, Should Guide Autonomous Vehicle Deployment

On October 2, 2023, a driver in San Francisco hit a jaywalking pedestrian, knocking that person directly in front of a driverless taxi, which then dragged her for 20 feet. While it’s tempting to draw quick conclusions about autonomous vehicles’ safety from such high-profile accidents, a comprehensive analysis of data reveals that these vehicles typically have fewer accidents compared to those driven by humans. Policymakers should respond logically rather than emotionally to such isolated accidents; otherwise, they risk impeding the progress of a technology that has the potential to enhance safety standards and improve urban transportation significantly.

Autonomous vehicles (AVs) are emerging as a safer alternative to human-driven vehicles. In the United States, human drivers have an accident rate of 65 crashes per million miles driven. In contrast, Waymo and Cruise, the two companies authorized for robotaxi services in San Francisco, report substantially lower incident rates of 20 and 23 crashes per million miles, respectively. In addition, a study comparing insurance claims of Waymo’s autonomous vehicles with human drivers found that Waymo generated significantly fewer insurance claims. In over 3.8 million miles driven in rider-only mode, Waymo’s autonomous system incurred zero bodily injury claims compared to the human driver baseline of 1.11 claims per million miles. Moreover, property damage claims were reduced by 76 percent, from 3.26 claims per million miles for human baseline to 0.78 by Waymo cars.

However, the California DMV suspended Cruise’s permit to operate in San Francisco shortly after the incident with the pedestrian, citing an “unreasonable risk to public safety.” While regulators should respond to new evidence, the DMV appears to have made this decision primarily because of the high-profile nature of the incident rather than a change in the overall safety record of AVs. This incident in particular appears to be an overreaction because a human-driven vehicle caused the accident not the robotaxi.

The media also shapes public perception. Sensationalized coverage of AV incidents, like the Cruise event, often overshadows the broader safety advantages. News coverage of the accident included misleading titles, such as “Driverless Car Traps Woman in San Francisco.” Editors use these headlines to garner significant attention and evoke an emotional reaction. Meanwhile, the numerous accidents that AV technology potentially prevent go largely unrecognized. This imbalance in reporting creates a skewed narrative, emphasizing the few failures over the many successes. This heightened visibility of AV incidents poses challenges for both policymakers and companies. It influences public opinion and drives regulatory responses. Companies are forced to react so that they can save their public image, as seen with Cruise’s subsequent halt of its operations nationwide, stating a safety reassessment. This is a huge setback to the public’s access to safer AV technology. Similarly, an accident in 2018 prompted Uber to halt its AV testing program, and the company never reinstated it.

To mitigate these issues, policymakers should establish and regularly review safety benchmarks for AVs. Clear performance expectations would provide AV companies with a stable regulatory framework, allowing them to focus on service improvements without fear of overreactions to isolated events. Policymakers should also communicate this information to the public, but in context, such as by showing accident rate data for AVs next to accident rate data for traditional vehicles. Better information will help the public and media understand the value of AVs.

Policymakers should also create clear guidelines for reporting incidents. Much of the recent dispute around Cruise has to do with whether the company properly reported information about the accident to regulators. The California DMV and California Public Utilities Commission, has accused the company of not properly disclosing evidence about the accident, which it has denied. Establishing these guidelines will help prevent future disputes and ensure regulators get access to timely information about incidents.

Policymakers should aim for continuous improvements in transportation safety. Policy decisions should be data-driven, comparing the safety records and benefits of different technologies and policies. In this light, AVs stand out as they offer the potential to reduce accidents and improve urban mobility. Policy decisions need to encourage this potential, rather than being swayed by emotional reactions to isolated events. The primary goal is to enhance road safety by using data and technology in a balanced and well-informed way.