Crash Statistics: Human vs. AI Drivers

The number of autonomous vehicles on the road increases every year, but at what point is a vehicle considered autonomous? Most new cars have some sort of cruise control, lane assist, and braking assist, but these features alone don’t make a car autonomous. A truly autonomous car is able to handle at least most aspects of driving without a human passenger intervening. So how do AI (artificial intelligence) drivers compare to humans? And are they really going to make the roads a safer place?

Autonomous Drivers in Theory

According to the NHTSA (National Highway Traffic Safety Administration) more than 90% of auto accidents in the US are caused by driver error in some way. It stands to reason, then, that by eliminating human drivers 90% of car crashes would be eliminated as well, right? Unfortunately, while it would be great to eliminate 90% of car crashes, that simply isn’t a realistic outcome. Despite how flawed we are as drivers, we have the ability to react to a whole host of dynamic situations on the road. AI simply don’t.

It would be easy to look at the crash statistics and say that AI are inherently better drivers because they are involved in fewer crashes, but that’s only half the story. What about all those times that crashes are avoided because a human was behind the wheel? For example, a human might know that a bridge is out due to a sign, but will an AI?

Comparing Crashes

Car Accident

Here at Zingani Collision we know that humans get involved in many different kinds of crashes. Sometimes they get rear ended, other times sideswiped, and occasionally T-boned among others. However, according to the California Department of Motor Vehicles, AI drivers overwhelmingly get in rear end accidents. And no, the AI aren’t rear ending other drivers, other drivers are rear ending the AI. But why is this the case? Does this prove that humans really are unsafe drivers? Let’s not jump to any conclusions just yet.

A Different kind of Driving

Driver

It’s thought that a major reason for the large percentage of autonomous vehicle crashes that involve being rear ended is due to the way AI tend to drive. AI drive in a much more “cautious” way than humans do. If they aren’t sure what an object is, they’ll typically brake for it just in case. This causes a lot of stop-and-go driving which humans aren’t used to. A human driver isn’t expecting the car in front of them to suddenly brake for a fire hydrant (which it thinks is a small child), and so they end up rear ending the autonomous car.

Does this mean AI are actually the problem? It’s hard to say. As AI become better at driving, they’ll make fewer of these “cautious” mistakes. At the same time, as humans gain more experience driving with AI around them, they’ll become more aware of their inhuman tendencies.

Limited Data Samples

The large percentage of AI accidents that are the result of being rear ended aside, it would seem that AI do truly get into less accidents than humans. However, the data on this is skewed. Humans drive trillions of miles each year while AI only drive millions. That’s a big difference, and it means there’s a lot less data available for AI drivers. In fact, the sample size for autonomous vehicle miles driven is so small that it can’t realistically be compared to the miles driven by humans.

snowy road

We also need to take into consideration the environments in which AI are driving compared to humans. Generally, AI are driving in places like California and Arizona where the roads are relatively flat, dry, and free of obstacles. Humans are great at driving on flat, dry, and obstacle-free roads too though. If AI were driving in more variable environments and in more weather conditions we would likely see an increase in crashes among them. Rain, snow, and Ice are major contributing factors to car crashes, and these are conditions AI drivers have largely avoided thus far. How would they fair on the snow-covered, pothole-riddled roads of Pennsylvania in the Winter?

Conclusion

Although we humans aren’t always the greatest drivers in the world, our experience and ability to adapt to a wide variety of situations help us avoid a great number of auto accidents every year. While AI may be more cautious drivers than we are most of the time, they can’t react to unforeseen circumstances in the same way we can. However, there will always be crashes that even the most experienced driver is unable to avoid. Brakes sometimes fail, trees sometimes fall, and deer will always be deer.

Zingani Collision

If you enjoy driving, you don’t have to worry about being replaced by AI any time soon. But even if you are, the bad news is that unavoidable crashes will still occur. However, the good news is that your local collision repair shop will still be here to help with the same professional care it always has. Whether you rear ended an autonomous vehicle, got into a multi-car accident at an intersection, or just barely didn’t fit into that last parking spot, Zingani Collision is here to help. With over 40 years of auto repair experience, we’ve seen it all and will work to get your car back on the road both safely and quickly. Contact us today to receive your free estimate!