What if Every Car Were Self-Driving? An 80% Drop in Accidents

Ninety four percent of car crashes are caused by human error. That’s not a statistic from an advocacy group — it’s from the National Highway Traffic Safety Administration, the same agency that regulates vehicle safety in the United States.

Humans looking at phones. Humans running red lights. Humans misjudging a turn. Humans driving tired. The weakest link in the safety chain is, and always has been, the person behind the wheel.

The Numbers Don’t Lie

A Tesla operating on Full Self-Driving gets into a fender bender roughly once every 5 million miles. The average human driver in the United States gets into an accident roughly once every 500,000 miles.

That’s a 10x improvement before we even account for what happens when every car on the road is self-driving and communicating with every other car in real time.

According to NHTSA estimates, a fully autonomous fleet — where every vehicle is self-driving, networked, and communicating — would reduce total accidents by 80 to 90 percent.

The Irony

The news runs every minor fender bender involving a car with a sensor suite on the roof as a headline. FSD grazes a curb? National news. A Waymo taps a parked car? Front page.

Meanwhile, the same outlets run the overnight highway fatality count — 100+ deaths per day in the US alone — as a ticker at the bottom of the screen. A million people die worldwide in car crashes every year. The vast majority are preventable. The vast majority are caused by humans.

The Actual Question

The question isn’t whether self-driving cars are perfect. They aren’t. The question is whether they need to be. Because the alternative — 94% of crashes caused by human error — is a bar so low that an imperfect autonomous system clears it by a wide margin on day one.

A technology that eliminates 80-90% of traffic deaths isn’t something that needs to be perfect before it’s worth deploying. It’s something that should have been deployed yesterday.

— TechnoWonderboy