Flawed Facial Recognition Software Leads to Wrongful Arrest, Highlighting Racial Bias Concerns While Researchers Study Innate Talent of Super Recognizer Humans
-
Police relied on flawed facial recognition software to arrest an innocent Black man, Randal Reid, showing issues with accuracy and racial bias.
-
Special humans called "super-recognizers" are extraordinarily good at identifying faces, even in low quality images where AI struggles.
-
Super facial recognition ability seems highly heritable, but can potentially be improved with training.
-
Humans still show "other race" effects in facial recognition, being worse at IDs of unfamiliar races.
-
Researchers are seeking the "Einstein" of facial recognition - the top 0.1% in talent - to understand what makes them so skilled.