There is one question everyone is asking about the Autopilot features available on Tesla cars: is it as safe? Tesla seems to think so. The company is proudly displaying the crash statistics of its Autopilot system and touting that its system is safer than a human driver. How true is this? What do these crash statistics really tell us about Tesla Autopilot?
While Tesla’s Autopilot crash statistics look favorable, these numbers exist in isolation. When you start considering the context, the figures are not nearly as impressive. Tesla’s Autopilot is still safe, but whether it is safer than a human driver is still quite questionable.
The hard numbers that Tesla has published give the impression of surety, but statistics can be dangerous without proper context. This article will explore both the crash statistics of Tesla’s Autopilot and what those numbers truly mean for drivers. Read on to find out the larger story of Tesla’s Autopilot crash statistics.
Tesla Autopilot Crash Statistics
Since 2018 Tesla has been tracking crashes that occur when their Autopilot is engaged. Here are the statistics from Tesla’s website.
|Crashes (Autopilot)||Crashes (other Tesla safety)||Crashes (no Tesla safety)||National crash average|
|2018 3rd Quarter||1 per 3.34 million miles||1 per 1.92 million miles||1 per 2.02 million miles||1 per 492,000 miles|
|2018 4th Quarter||1 per 2.91 million miles||1 per 1.58 million miles||1 per 1.25 million miles||1 per 436,000 miles|
|2019 1st Quarter||1 per 2.87 million miles||1 per 1.76 million miles||1 per 1.26 million miles||1 per 436,000 miles|
|2019 2nd Quarter||1 per 3.27 million miles||1 per 2.19 million miles||1 per 1.41 million miles||1 per 498,000 miles|
|2019 3rd Quarter||1 per 4.34 million miles||1 per 2.70 million miles||1 per 1.82 million miles||1 per 498,000 miles|
|2019 4th Quarter||1 per 2.87 million miles||1 per 1.76 million miles||1 per 1.26 million miles||1 per 436,000 miles|
|2020 1st Quarter||1 per 4.68 million miles||1 per 1.99 million miles||1 per 1.42 million miles||1 per 479,000 miles|
|2020 2nd Quarter||1 per 4.53 million miles||1 per 2.27 million miles||1 per 1.56 million miles||1 per 479,000 miles|
|2020 3rd Quarter||1 per 4.59 million miles||1 per 2.42 million miles||1 per 1.79 million miles||1 per 479,000 miles|
|2020 4th Quarter||1 per 3.45 million miles||1 per 2.05 million miles||1 per 1.27 million miles||1 per 484,000 miles|
|2021 1st Quarter||1 per 4.19 million miles||1 per 2.05 million miles||1 per 978,000 miles||1 per 484,000 miles|
These numbers look pretty good, and they have led Elon Musk to declare that Tesla Autopilot is ten times safer than a human driver. It is quite clear that Tesla Autopilot is responsible for fewer accidents than the average, but is it really ten times safer? Continue reading for a breakdown of these numbers.
What the Statistics Do Not Say
There are several reasons that taking statistics at face value is a risky business. Numbers do not lie, but they also do not give the entire picture.
There are some things you need to keep in mind to truly understand what Tesla’s Autopilot crash statistics mean.
Amount of Data
Tesla has access to all the crash data from their own cars, so that should mean that their sample size is as good as it gets right? Getting accurate statistical samples is challenging. Tesla Autopilot has only been around for a few years. The amount of data Tesla has on crash statistics involving their Autopilot system pales in comparison to the data involved in determining average crash statistics for the entire nation.
You can see this clearly in the amount of variance in Tesla’s crash statistics. The number of miles per one crash with Autopilot varies by as much as 1.81 million miles while the average crash statistics of the nation as a whole during that period only varies by 62,000 miles. That is a 38 percent difference from Tesla versus a 12 percent difference from the national average.
What this means is that Tesla’s data is far from steady. The varying numbers indicate that Tesla does not yet have enough data to reach a hard number when it comes to how much safer Tesla Autopilot is.
As for Elon Musk’s claim that Autopilot is ten times safer, well, it depends on which quarter you choose to look at and that means that number is far from concrete.
What Is Tesla Competing Against?
The other vital thing to consider with statistics is the context. Tesla’s numbers look impressive, but what are those numbers not saying?
No matter how you look at it, Tesla cars on Autopilot have a far smaller crash rate than the national average. However, is that really because of Autopilot? Instead of looking at why Tesla’s number is so high, look instead at why the national average for miles per car crash is so much lower.
- Tesla Autopilot has only been around since 2015, which means that any cars with it enabled are only a few years old.
- The average age of cars driven in the United States today is close to 12 years old
- One in four cars (25 percent) of cars on the road are 16 years or older.
Car crashes on a national average are higher because most people are not driving brand new cars with precision brakes and the latest safety features. You can easily see this discrepancy by comparing the crash statistics of people driving a Tesla without Autopilot to the national crash average. The miles without a crash is significantly higher in a Tesla even without Autopilot.
Yes, Tesla has much better crash numbers than the national average, but they may be more because most cars on the road are not in the best condition rather than because Tesla’s Autopilot is so much safer.
When Do People Use Autopilot?
Another thing that Tesla’s statistics do not consider is when people use Autopilot. Autopilot does not make a Tesla a self-driving car. It allows your car to steer, accelerate, and brake within its lane. Like cruise control, Autopilot is beneficial on long stretches of road but not for driving around town.
This means that Autopilot is most frequently engaged on the highway, but this is not where the majority of crashes happen. Car crashes more frequently occur in these areas:
- Rural roads
- Urban areas
- Close to home (in neighborhoods)
- Parking lots
Thus, places where people are more likely to have a crash are also where they are not as likely to have Autopilot engaged. Autopilot’s grand crash statistics are at least partially affected by this reality.
This may also affect why the likelihood of a crash in Tesla’s own cars without Autopilot increases. In areas where they are more likely to crash Tesla drivers likely have their Autopilot off. Therefore, the increasing crashes may not be because of the lack of Autopilot but rather because of where the Tesla is being driven.
Autopilot Does Not Replace the Human Driver
It is tempting to read into these statistics as accurate depictions of how safe Autopilot is versus how safe human drivers are. However, that is not quite what is happening here.
Autopilot does not replace the human driver. It is a feature that can be engaged by the human driver. When the Autopilot is engaged, the human driver is still present. They can turn off Autopilot and take over manual control if traffic worsens or the driving becomes more challenging in any way.
Hence, it is not Autopilot versus human drivers but rather human drivers with Autopilot versus human drivers without Autopilot. In many ways then Autopilot is an additional safety feature rather than a totally alternate mode. The human driver is still there.
This means that, in terms of crashes, you cannot give total credit to Autopilot for anything.
- If a car has a wreck while Autopilot is engaged the driver may still be partially at fault.
- In the same way, though, you cannot assume that all avoided crashes are due to Autopilot.
- The human driver knowing when and when not to use Autopilot has a lot to do with the program’s success.
There is also the factor of other drivers. Sometimes you can do absolutely nothing wrong and still be involved in a crash. If another car runs a red light and crashes into the side of a Tesla that has Autopilot engaged, the Autopilot is likely not to blame for that accident.
What Does This Mean?
Tesla says that their Autopilot is much safer than a human driver, but the fact is that the numbers they are comparing to make this claim are not on equal terms.
The national average of car crashes is affected by the numerous older vehicles on the road and encompasses all types of crashes. Tesla’s Autopilot figures are coming exclusively from newer cars and are likely not including areas where accidents are more likely to occur. It is not exactly an apples-to-apples comparison.
This is not to say that Tesla’s numbers are a fraud. It is simply pointing out that these statistics are not the product of an isolated experiment.
You know that car accidents can be caused by a variety of things. The statistics which Tesla is presenting are coming from real-world situations, and that means there are multiple factors at play for each crash. It is impossible to truly isolate Autopilot as the reason all of those crashes occurred or as the reason crashes did not occur.
Is Tesla Autopilot Safe?
If the numbers do not tell the whole story, how safe is Tesla Autopilot truly?
While Tesla’s statistics do need to be taken with a grain of salt, that does not mean that Tesla’s Autopilot is unsafe. Tesla’s statistics show their Autopilot as being unquestionably far better than human drivers. That is doubtful, but there is no evidence to indicate that Autopilot is worse than a human driver.
Tesla’s numbers may not be as impressive as they first appear, but less impressive does not mean bad. Autopilot clearly does not cause an increase in crashes.
This is especially evident when comparing the crash statistics of Teslas with Autopilot engaged to Teslas without Autopilot engaged. The crash likelihood increases and, though other factors may contribute to that, it seems likely that Autopilot has something to do with it.
Can Tesla Autopilot Be Dangerous?
As previously stated, Tesla Autopilot is far from making a car self-driving.
- Tesla Autopilot is probably closer to cruise control than it is to fully autonomous self-driving.
- Tesla Autopilot can be a great tool, but placing too much faith in its abilities is asking for trouble.
- This is a system that still requires supervision from the human driver.
With this in mind, Tesla’s endless positivity around Autopilot and its abilities could do more harm than good. The impressive statistics Tesla has been promoting do not tell the whole story as this article has shown, and they could cause Tesla drivers to have too much confidence in this system to the point of recklessness.
This is why it is vital to put statistics in context. Autopilot may look ten times safer than a human driver, but a human driver relying on Autopilot’s abilities alone is certainly not safer.
Tesla Autopilot and Fatal Crashes
Fender benders are one thing, but a crash that results in a death is an entirely different situation. One thing that Tesla’s crash statistics do not address is the severity of the crashes. Is Autopilot mainly responsible for minor dents or wrecks with totaled cars and severe injuries?
This information is not easy to obtain, but you do know of several fatal crashes in which Tesla Autopilot was engaged. Those are:
- 2016 – A Model S on Autopilot crashed into a semi-truck in Florida
- 2016 – A Model S on Autopilot hit a street sweeper in China
- 2018 – A Model X on Autopilot crashed into the highway divider in California
- 2021 – A Model 3 on Autopilot hit an overturned truck in California
- 2021 – A Model 2019 on Autopilot hit a tree in Texas
The recent 2021 crashes have increased concern about the safety of Tesla’s Autopilot. The most recent crash statistics say that Teslas on Autopilot only have 1 crash per 4.19 million miles, but if these crashes are more severe then those low numbers are not as encouraging.
The recent fatal crashes may not be the fault of Tesla’s software though. In the incident in Texas, two men were killed when a Tesla hit a tree, but no one was in the driver’s seat. Tesla has stated that the Autopilot function still requires an active driver, but some drivers are putting a bit too much faith in their Tesla.
These new fatal crashes make it quite clear that Tesla Autopilot is not safer than a human driver. It may allow a human driver to drive safer, but using Autopilot as a substitute for a human driver is not safe. The human pilot must remain aware and conscious.
The Potential of Tesla Autopilot
One important thing to remember is that Tesla Autopilot is not a finished invention. Tesla continues to work on and improve this software.
We can see this development in the crash statistics. More recent numbers from 2020 look overall better than the initial figures from 2018. The general trend seems to be that Tesla is improving their Autopilot, and that may mean in time it will be far safer than a human driver.
Completely autonomous self-driving cars are in the works, so you have every reason to think that this type of technology will continue to improve and that Tesla Autopilot will get better and better.
However, the improvement of Tesla Autopilot has clear dangers as drivers relying too much on the software leads to reckless and endangering behavior. The technology is far from being totally self-reliant. Tesla may have boasted overmuch about their program’s current capability, and that boasting may be leading to dangerous consumer behavior.
Tesla’s Autopilot crash statistics only tell part of the story, and these impressive figures have given some drivers far too much confidence in their Tesla’s ability to self-drive. In light of two recent fatal crashes, Tesla may need to re-examine their claim that Tesla Autopilot is 10 times safer than a human pilot.
It is also just as important for Tesla consumers to keep in mind that Tesla’s Autopilot feature is not meant to be used as a completely hands-free, driverless experience. Autopilot still requires a driver’s attention and action while in use. Until driverless technology is completely safe, Teslas still require the driver to remain aware of the car and its surroundings.