legit new study finds potential risk with self-driving cars: failure to detect black pedestrians

[basquebromance]

Study finds a potential risk with self-driving cars: failure to detect dark-skinned pedestrians

I keep trying to warn people about this rush to automation and robotics: The biases of the developer, conscious or not, are being built into the logarithms. The machines may turn out to be just as racist as the men who make them.

 

[Likkmee]

Circumcise all black babys lips so the teeth are always exposed. Problem over !

 

[MarathonMike]

Study finds a potential risk with self-driving cars: failure to detect dark-skinned pedestrians

I keep trying to warn people about this rush to automation and robotics: The biases of the developer, conscious or not, are being built into the logarithms. The machines may turn out to be just as racist as the men who make them.

I think you mean algorithms not logarithms. This "study" fails the common sense test. First of all, the amount of exposed skin in the typical pedestrian is going to be very small compared to the amount and color of clothes. Dark clothes would be far more dangerous than dark skin. Secondly, any difference in a sensor's ability to detect skin difference is PURELY based on the amount of light reflected back to the sensor. There is no way they build racial characteristics into detection algorithms. If the AI is allegedly hitting dark skinned people 5% more than light skinned people, it is the difference in reflected light or other factors, not racially biased algorithms.