Coded Bias Parent Guide

When MIT researcher Joy Buolamwini began work on a mirror that would superimpose inspiring images on user’s faces, she was disappointed to discover that the software had a hard time recognizing her face…that is, until she wore a white mask. This led the Ghanaian-American computer scientist to launch a study of facial recognition software. Using a sample set of 1000 faces, Buolamwini discovered that while facial recognition programs are very good at recognizing white male faces, they are less effective at assessing black male faces and they are even worse with white female faces and abysmally bad at categorizing black female faces. This discovery led the young researcher to examine the algorithms that power so many of the software programs that dominate our lives – from social media to loan applications to resume sorters to security programs.

It is no surprise to discover that social and racial biases are embedded in so many of these algorithms. After all, algorithms are created and software is designed by human beings who have biases of their own, often unacknowledged. As Buolamwini states, “AI (artificial intelligence) looks to the future but our data reflects our past.” The field of artificial intelligence was established by a small coterie of white men and their perspectives continue to infuse the discipline, as Coded Bias demonstrates. In a twist that parents will appreciate, this documentary makes its point by interviewing experts in AI and computer science, most of whom are female. It’s heartening that teenage girls with an interest in tech will have the all-too-rare chance to watch women with technical expertise on the big screen. That’s a bit of bias correction all on its own.

While this documentary looks at issues of social bias built into software, it isn’t a screed of racial invective. Rather, it’s an examination of how big data and powerful machine learning systems erode privacy and equity for all citizens, especially those who are poor and/or visible minorities. In the words of big data expert Cathy O’Neil, “Power: it’s all about who owns the code. People who own it deploy it against others. People are suffering algorithmic harm and there’s no accountability.”

Coded Bias devotes a fair bit of its runtime to examining the issue of algorithmic harm – from misuse of inaccurate facial recognition software in policing to harassing tenants to the perpetuation of social and economic inequality. And it takes a chilling look at the power corporate interests and governments obtain when we freely hand over our data online. If The Social Dilemma didn’t scare you, Coded Bias certainly will. Hopefully, it will lead to substantive discussions with teens about the responsible use of digital devices and sensible security precautions. A little examination of our own online habits won’t hurt either.