New research shows driverless car software is significantly more accurate with adults and light skinned people than children and dark-skinned people.
New research shows driverless car software is significantly more accurate with adults and light skinned people than children and dark-skinned people.
More like programmers who didn’t consider putting in more folks that don’t look like them in the training datasets. Unconscious bias strikes again.
I would say it’s because dark stuff on dark background is harder to detect than other way around. Roads are dark, shadows are dark, pavement is sometimes dark, houses are quite often dark, so it blends.
I’d rather call for more powerful algorithms than bigger data sets.