Driverless cars are making waves as the future of transportation, but a new study released this week demonstrates yet another reason why they may not be the ideal solution. According to the research, due to camera-based sensors and facial recognition software, driverless cars are far less capable of recognizing kids and dark-skinned people, potentially leading to dangerous outcomes on the roads.
Discrimination in cars is nothing new. In 1971, researchers noted that Ford Pintos identified black pedestrians less accurately than white ones. Now, sensing technology in autonomous cars is opening up a new avenue of discrimination that could prove to be equally dangerous.
The study, which was conducted by the RAND Corporation and the University of Colorado, tested the facial recognition capabilities of two existing driverless car technologies. They found that, overall, the sensing technologies detected dark-skinned individuals about 10 percent less accurately than light-skinned individuals, and children about 20 percent less accurately than adults.
While these systems did not do as well in recognizing dark-skinned individuals, they still tended to be more accurate when it came to other features – such as being able to distinguish between male and female pedestrians.
The researchers noted that while facial recognition technology is still developing, these systems must become more accurate in order to create vehicles that are safe for all people. Without this accuracy, driverless cars could fail to detect pedestrians unless they fit a certain profile, leading to disastrous circumstances.
As driverless cars continue to become more prevalent and widespread, these numbers become increasingly concerning. Only by recognizing the potential for discrimination in these systems and working to fix them can we ensure that these cars are safe and effective for all pedestrians.
