Don’t let robot cars go out of control on California roads

A prototype of a Google self-driving car is on display at the state Capitol in 2015.
A prototype of a Google self-driving car is on display at the state Capitol in 2015. Sacramento Bee file

Self-driving cars cannot yet drive themselves. This is the unmistakable conclusion of new data from the state Department of Motor Vehicles.


The DMV reports show that robot cars cannot go more than 5,600 miles on California roads, even in the best-case-scenario, without the test driver having to take control. In most cases, the vehicles cannot travel more than a few hundred miles without needing human intervention due to quicker-than-average yellow lights, rapid changes in traffic, sudden lane blockages, cars parked incorrectly or GPS signal and other software failure.

Jamie Court new

Automakers testing on California roads must report annually when human drivers have to take the wheel. Remarkably, this is the only public window into the state of robot car technology nationally.

As self-driving cars are deployed beyond the testing phase, the state must mandate more disclosure of collisions and other incidents. A hearing in the Assembly Transportation Committee on Tuesday offers an opportunity for such a discussion. Unfortunately, Silicon Valley engineers dominating the show don’t like to share their data.

From a historical perspective, the development of robot cars is miraculous. The problem is carmakers are rushing the cars to the roads in two years, and the Trump administration and Congress seem all too willing to go along.

For example, federal law and legislation does not create any meaningful regulations and gives manufacturers full authority to decide whether their vehicles are prepared to roam the streets without human supervision. California could see its human testing standards preempted by such federal legislation.

The industry insists that self-driving cars will kill less people than vehicles with human drivers. That may well be true. However, without enough information we cannot be certain of the ethics and programming of vehicles that will do the killing in the future.

When a child wanders in the middle of the road, for example, does the oncoming robot car swerve to avoid her, putting its occupants at risk? Or does it sacrifice the child to avoid liability to the car owners?

We must know human values still control the roads, not the cold calculus of corporate robots. If California doesn’t require more information about collisions involving the technology, we just won’t know who really has the wheel.

Jamie Court is president of Consumer Watchdog, a nonprofit advocacy group. He can be contacted at