2020, NR, 90 min. Directed by Shalini Kantayya.
REVIEWED By Jenny Nulf, Fri., Nov. 20, 2020
Code is supposed to be neutral. It’s not that lifelike, non-human, and binary codes and algorithms can’t be racist or sexist – they just are.
But what if the algorithm written reflected the writer’s worldview?
Coded Bias explores the world of new tech, opening with MIT Media Lab researcher Joy Buolamwini’s discovery that facial recognition software has trouble detecting dark-skinned faces, especially those of Black women. This finding sets the tone of the documentary, a dark looming cloud that hovers as director Kantayya hops around the globe, uncovering how countries and institutions utilize facial recognition software, from the UK police force to the Chinese government. If these programs aren’t accurate, and cannot even detect a Black woman’s face properly, aren’t they therefore problematic and potentially dangerous if they become the standard?
Code learns from humans. It doesn’t just pop up in a neutral space, ready to serve its purpose. A human has to write it, and who are the humans that historically are the leaders of new tech? White men, and their worldview reflects the systems they build, whether intentionally or not. Kantayya stresses how this unbalance and lack of diversity in leadership affects jobs throughout Coded Bias. Amazon uses a recruiting tool that weeds out applications that mention if a candidate attended a women’s university or participated in a women’s sports club; a teacher of the year’s lifelong career in Houston is threatened by a system that flags him as a poor educator; and a woman has to have weekly check-ins based on a software that identifies her as a potential weak link.
Kantayya wants to point out these harmful gaps so it improves the coding space. Her film concludes with a promise for a brighter future: Buolamwini presenting her findings on facial recognition during a White House summit on Computer Science for All (featuring political fan-favorite Alexandria Ocasio-Cortez). Buolamwini expresses optimism that the system can be changed and improved, demonstrating that companies like IBM took her research and fixed their software’s algorithm so it doesn’t have trouble detecting any kind of face regardless of gender or skin tone.
Coded Bias is not interested in wallowing in despair for the future, like many tech-infused documentaries like to do. Kantayya wants to inform and inspire change. New tech is not inherently evil or totally lost in its bias, and because it is code it can ultimately be improved upon and changed rapidly, unlike humanity. For subjects like Buolamwini, science fiction brought her hope and comfort, and Kantayya wants to infect others with those same aspirations.
Coded Bias is available now as a Virtual Cinema release.