The Safety of Self-Driving Cars

Perhaps one of the biggest concerns about the migration over to self-driving cars is the safety factor. Many questions are being raised about the issue of ethics. While a computer can make lightning-fast decisions based on data and logic, the recurring fear is that, as a machine, it cannot possibly make the same type of life-affecting moral decision that a human can. For this reason, many studies are being done to determine whether it really is safer to have an AI system behind the wheel.

Research published by the National Highway Traffic Safety Administration division of the U.S. Department of Transportation (USDOT NHTSA) found that the majority of all car crashes between 2005 and 2007 were the result of driver error, as opposed to a malfunction. Driver distraction and/or impairment caused 94% of the crashes. In order to be deemed safe, therefore, an automated system would just have to have better numbers than humans.

Testing has been a bit of a contentious issue since real-world scenarios are hard to employ without putting people in danger. There are test runs being done by companies such as Google, who are showing huge strides with their focus being on driver assistance, rather than car manufacturing. What began as a self-driving car project, their offshoot company, Waymo, has partnered with several major car developers to enhance their own cars for automation. 

If data analysts focus on the issue of driver error due to impairment and/or distraction, there will undoubtedly be fewer instances of alcohol- or drug-impaired driving when the human is removed from the equation. Within the next few years, safety testing will advance to real-world situations in order to help with the machine learning process. Tesla’s latest model has received high marks for safety tests. In order to earn its five-start rating, the Model 3 not only had to show good overall performance in crash protection but also display superior and robust crash avoidance technology.

That being said, all computing systems are vulnerable to hacks, and cyber protection will have to be at unprecedented levels in order to prevent malicious external cyberattacks from occurring.