Do you trust mathematics? I ask you this because the accident statistics are telling a story that perhaps we are not yet ready to accept. After approximately 80 million kilometers traveled, the driverless cars of Waymo are proving to be dramatically safer than us humans at the wheel. I'm not talking about small marginal improvements or statistics manipulated to make everything look rosier. The numbers are there, raw and merciless: a reduction in83% in crashes serious enough to trigger an airbag. This means that for every crash that causes an airbag to deploy on a Waymo, there would have been nearly six with human drivers in the same conditions. The crash statistics are becoming increasingly difficult to ignore, especially when we consider that these vehicles are now operating in complex urban environments like major Chinese cities, or (in Waymo’s case) US cities like San Francisco and Phoenix.
The numbers don't lie
Using human crash data, Waymo estimated that Human drivers on the same roads were reportedly involved in 78 crashes serious enough to deploy an airbag. Self-driving cars, on the other hand, have been involved in only 13 accidents with airbag deployment (obviously all duly described by newspapers and the media, one by one, fueling a distorted narrative). That's an 83% reduction - slightly less than the 84% estimated last September on the first 34 million kilometers, but at the time it was said that "it was too early to evaluate". And now?
In the same roughly 80 million kilometers, Waymo estimates that human drivers were involved in 190 accidents serious enough to cause injury. Self-driving cars, on the other hand, have been involved in only 36 accidents with injuries between San Francisco and Phoenix. An 81% reduction in crashes resulting in injury.
Crash statistics show a significant improvement from last September, when Waymo estimated its cars had 73 percent fewer injury crashes.
Bottom line: Not only are self-driving cars already safer, they continue to improve as experience accumulates.
Accident statistics, confirmed also by third parties
The above analysis counts all accidents, regardless of whether Waymo's technology was responsible or not. But things look even better if we focus on accidents where Waymo's technology alone Waymo has been held accountable. To assess this aspect, the company under the Alphabet umbrella has co-signed a study in December with the insurance giant Swiss ReThe study focused on accidents that led to insurance claims against Waymo. This data seems especially credible because third parties, not Waymo, decide when an accident is serious enough to file a claim. And insurance adjusters, not Waymo, decide whether to hold Waymo liable for an accident.
Waymo had completed 40 million driverless kilometers by July 2024. And by the end of November 2024, Waymo had faced only two potential personal injury claims. Furthermore, both claims are pending, meaning they could still be resolved in Waymo’s favor.
What these numbers mean
The accident statistics don’t lie: self-driving cars are proving to be significantly safer than their human counterparts. That’s lives saved, injuries avoided, families that won’t have to deal with the trauma of a car crash.
How many of us are ready to accept this reality? I still see a lot of resistance on the topic. We are used to thinking of driving as a quintessential human skill, something that requires judgment, experience, and that particular form of emotional intelligence that allows us to predict the actions of other drivers. Yet, accident statistics suggest that AI, with its uninterrupted attention, its inability to get distracted, tired, or drugged, is already outpacing our capabilities on the road.
It's a thought that requires some psychological adjustment. Are you ready?
