Autonomous vehicles are in many ways considered the path to the future because of the fact that this is the sort of thing that could potentially end up allowing people to travel with ease. However, researchers in King’s College London and the Peking University collaborated on a study which revealed a significant danger of AI based self driving cares, namely that they are biased against people of color and children.
With all of that having been said and now out of the way, it is important to note that adults were 19.52% more likely to be detected by self driving cars than children. This could make them more dangerous for kids than might have been the case otherwise, suggesting that they are not quite safe enough to rely on in the modern era. As for the racial bias, autonomous vehicles had a 7.52% higher success rate when detected light skinned individuals as compared to those that have a darker skin tone.
The disparity becomes even greater at night with all things having been considered and taken into account, reaching as high as 9.86%. This seems to indicate that people of color are at higher risk of being run over by self driving vehicles that are incapable of detecting them with their AI sensors.
The main issue here might be the data set that is being used to train these vehicles. These data sets contain the inherent biases of humanity, but in spite of the fact that this is the case, not all that much is being done to reduce these biases which might lead to a downturn in self driving vehicles in the future.
Interestingly, there was not all that much of a gender disparity based on the findings in this study, coming in at just over 1%. This just goes to show that some biases are harder to get rid of than others, and children in particular would not be able to stay safe if this is not corrected. It will be crucial to tackle this problem head on instead of allowing it to become more significant as the industry grows.
Read next: Electrifying Growth: US Electric Vehicle Sales Surged by 79% in the Initial Quarter of 2023, Propelled by Tax Rebate Support
With all of that having been said and now out of the way, it is important to note that adults were 19.52% more likely to be detected by self driving cars than children. This could make them more dangerous for kids than might have been the case otherwise, suggesting that they are not quite safe enough to rely on in the modern era. As for the racial bias, autonomous vehicles had a 7.52% higher success rate when detected light skinned individuals as compared to those that have a darker skin tone.
The disparity becomes even greater at night with all things having been considered and taken into account, reaching as high as 9.86%. This seems to indicate that people of color are at higher risk of being run over by self driving vehicles that are incapable of detecting them with their AI sensors.
The main issue here might be the data set that is being used to train these vehicles. These data sets contain the inherent biases of humanity, but in spite of the fact that this is the case, not all that much is being done to reduce these biases which might lead to a downturn in self driving vehicles in the future.
Interestingly, there was not all that much of a gender disparity based on the findings in this study, coming in at just over 1%. This just goes to show that some biases are harder to get rid of than others, and children in particular would not be able to stay safe if this is not corrected. It will be crucial to tackle this problem head on instead of allowing it to become more significant as the industry grows.
Read next: Electrifying Growth: US Electric Vehicle Sales Surged by 79% in the Initial Quarter of 2023, Propelled by Tax Rebate Support
