Autonomous vehicles deliver on the promise of safe driving

Co-workers from the East Coast and Midwest found the novelty to be both intimidating and exhilarating. We took lots of videos, and the technology has moved well beyond the news clips of cars getting stuck in roundabouts. 

Here is what I consistently observed: Waymo never exceeded the speed limit, not once. It came to a full and complete stop at every stop sign and red light. It maintained deliberate, consistent distance from other cars, cyclists, scooters, pedestrians and whatever else San Francisco throws into a lane on a given day. When something strange happened — unexpected pedestrian movement or a vehicle backing up — it slowed down or stopped entirely instead of forcing its way through.

The other thing I noticed is that because there is a real concentration of Waymos in San Francisco, the human drivers around them had no choice but to drive more conservatively.

I watched this happen again and again. A Waymo at the front of a lane sets the tempo of traffic. Waymos on both sides mean there will be no cutting off of anyone. No one rolls 10 miles over the speed limit, no one creeps through stop signs, there’s no aggressive inching into crosswalks. It was really interesting to witness what happens when the car leading the pack simply refuses to participate in the informal social contract of “speed of traffic.” 

Something else stood out: I never once felt motion sick riding in a Waymo. This almost never happens to me as a passenger. Waymo prioritizes smoothness over speed, extending braking distance and accelerating gradually. There are no hard starts, no last-second stops, and no swerves born from impatience. The ride feels calm (the music doesn’t hurt, either).

Humans, on the other hand, drive while tired, distracted, rushed, hungry and emotionally flooded. We make decisions based on partial information and social guesswork, convincing ourselves it’s safe to bend the rules just this time. Most of the time it works … until it doesn’t, as we’ve seen with the ongoing string of DUIs, vehicle-involved deaths, and cars jumping curbs into buildings and people.

Algorithmically driven vehicles don’t negotiate with mood or urgency, and they have access to more data than any human driver ever will in that moment in time. There are no blind spots, they brake earlier, keep wider margins, accelerate more slowly, and are willing to stop entirely when conditions become ambiguous. On the screens inside the car, you can see a 360-degree rendering of the world, with every object classified, tracked and updated in real time.

The data broadly supports what my experience suggests. Imagine a world where we see 92% fewer pedestrian injuries, 82% fewer cyclist injuries, 96% fewer injury-involved intersection crashes, and 82% fewer motorcyclist injuries on the road. Ten-year San Mateo County crash data shows over 26,347 crashes with 8.5% of those crashes resulting in a fatality or severe injury. Vehicles struck 2,139 pedestrians. How different the numbers could be.

Of course, nothing is perfect. Last week, a child was struck by a Waymo traveling about 6 mph in Southern California. The vehicle had been moving around 17 mph before braking hard when the child entered its data view. The child walked away without major injuries. The incident deserves scrutiny, but the counterfactual matters: A human driver almost certainly would have been moving faster — closer to the speed limit, or above it — and very likely would not have reacted as quickly.

The insurance industry is starting to price based on data. Lemonade now offers discounted rates of up to 50% when drivers use Tesla’s Full Self-Driving system, based on telemetry distinguishing assisted driving from fully human-controlled miles. Anyone who knows me knows I have strong feelings about Tesla’s hardware-at-software-speed approach, but insurers care about behavior and outcomes data and the autonomous data is compelling.

As a society, we’ve tried signs, tickets, slogans, speed bumps, DUI checkpoints and social-media shame. And as I wrote last week, since 2020 the United States leads (with velocity) in vehicle-involved deaths among the world’s top high-income countries. Autonomous vehicles may have to be the structural correction to a problem we’ve blamed on individual behavior for far too long. It’s long overdue, and it’s arriving whether we’re ready or not.