U.S. Rep. Rashida Tlaib came under criticism Wednesday for sharing legitimate concerns about facial recognition software: The technology is significantly flawed when used to identify people of color.
Citing research that shows people are better at recognizing faces from their own race, Tlaib suggested the optimal way to limit false arrests based on misidentifications is to hire Black analysts to examine the identities of Black suspects.
“Analysts need to be African-Americans, not people that are not,” Tlaib told Detroit Police Chief James Craig during a tour of his department's Real Time Crime Center, according to a video published by The Detroit News. “I think non-African-Americans think African-Americans all look the same."
Tlaib's reasoning is backed up by data and peer-reviewed scientific research. But the responses to her were anything but.
Craig, who is Black, called Tlaib's suggestion "racist" and "insulting" and even compared it to "racially insensitive comments" made by white Detroit cops who have been fired in the past.
Is trying to minimize the false arrests of African-Americans similar to, say, the white cop who was fired last year for comparing Detroiters to "zoo animals?" Or the white cop who was fired earlier this year for mocking a young Black woman on social media as she walked in subzero temperatures with a caption that read "celebrating black history month?"
Craig even appeared on President Donald Trump's favorite news show, "Fox & Friends: First," to dismiss Tlaib's concerns.
"It's a software. It's biometrics," Craig said. "And, to put race in it ... we're talking about trained professionals. My staff goes through intense training with the FBI, and so they're not looking at race, but it's measurements. We were appalled when she made this statement."
The conservative Detroit News editorial board compared Tlaib's suggestion to Trump telling her and some of her colleagues to "go back to where they came from."
Others called Tlaib "a moron" and "racist" on Twitter, dismissing numerous studies about the racial biases of facial recognition technology.
Tlaib is certainly not alone in her criticism of Detroit's $1 million facial recognition software, which police have used for nearly two years without public input. The department faced a backlash following an alarming study by Georgetown Law’s Center on Privacy & Technology in May. Researchers said Detroit’s extensive surveillance network and its $1 million facial recognition software “risks fundamentally changing the nature of our public spaces.”
Detroit's facial recognition software is especially pervasive because it's used on a quickly expanding surveillance network of high-definition cameras under Mayor Mike Duggan's Project Green Light, a crime-fighting initiative that began in 2016 at gas stations and fast-food restaurants. Since then, the city has installed more than 500 surveillance cameras at parks, schools, low-income housing complexes, immigration centers, gas stations, churches, abortion clinics, hotels, health centers, apartments, and addiction treatment centers. Now, the city is installing high-definition cameras at roughly 500 intersections at a time when other cities are scaling back because of privacy concerns.
I'm sharing some scholarly research on facial recognition w/ @detroitpolice to continue the dialogue. While facial recognition identifies a white man’s gender 99% of the time, it misidentifies darker-skinned females up to 35% of the time. https://t.co/GmvKgMQDML 1/7— Rashida Tlaib (@RashidaTlaib) October 2, 2019
Stay on top of Detroit news and views. Sign up for our weekly issue newsletter delivered each Wednesday.