The safety case for self-driving cars is getting hard to argue with

For years the honest answer to whether self-driving cars are safe was that we did not have enough data. IEEE Spectrum argues that this has changed. Waymo has now logged 20 million paid rides across 220 million miles, enough to compare against human drivers on the same roads. On that record its cars were involved in 92 percent fewer crashes causing serious injury or death, 96 percent fewer injury crashes at intersections, and 83 percent fewer crashes severe enough to deploy an airbag. A separate July study from the Insurance Institute for Highway Safety found Waymo taxis had 68 percent fewer crashes overall and an 81 percent lower injury-crash rate per mile across San Francisco, Phoenix, Los Angeles, and Austin.

The gains are not limited to full autonomy. Ordinary driver-assist features already move the needle: automatic emergency braking cuts pedestrian crashes by 27 percent, and automated braking reduces rear-end crashes by half and the injuries from them by 56 percent. Regulators are starting to respond to the evidence. The NHTSA granted Amazon's Zoox the first exemption from certain vehicle safety standards for a purpose-built robotaxi with no steering wheel, and announced a $5 million partnership to write the country's first performance standards for autonomous vehicles.

Why it matters

If you work in transport policy, insurance, or automotive engineering, the debate is shifting from whether these systems are safe to how fast to let them scale. Road crashes kill more than a million people a year worldwide, and the data now suggests that slowing deployment carries its own body count.