Buolamwini, Joy Adowaa

Impact on the world

Discovered as an MIT Media Lab graduate student that commercial facial recognition software repeatedly failed to detect her own face, succeeding only after she put on a white mask, an experience that directly led to her landmark 2018 study 'Gender Shades,' co-authored with Timnit Gebru, already profiled in this collection, which tested facial recognition systems from IBM, Microsoft, and Face++ and found error rates as high as 34.7 percent for darker-skinned women compared to under 1 percent for lighter-skinned men; the research prompted all three companies to substantially revise their systems, with IBM ultimately discontinuing its facial recognition product altogether, and a 2019 follow-up audit she co-authored examining Amazon's facial recognition technology led the company to attempt to publicly discredit her work before more than 70 researchers, including a Turing Award winner, signed a letter defending her findings.

Life & background

Born in 1989 and educated at Georgia Tech, where she earned her computer science degree, before completing two master's degrees at Oxford and MIT as a Rhodes Scholar and Fulbright Fellow, Buolamwini first encountered algorithmic bias directly while working on a facial-recognition art project at the MIT Media Lab, when the software could not consistently detect her darker-skinned face even as it readily recognized her lighter-skinned classmates; testing whether the failure was specific to her, she found the system immediately recognized a simple face she drew on her own palm, while continuing to fail on her actual face until she covered it with a plain white mask. Determined to investigate whether this was an isolated glitch or a systemic pattern, she built a benchmark dataset of 1,000 faces deliberately balanced across skin tone and gender, in direct contrast to the facial recognition industry's existing training data, which she found was more than 75 percent male and over 80 percent lighter-skinned, and used it to formally audit major commercial systems for her MIT thesis, work that became the foundation of 'Gender Shades.' She coined the term 'coded gaze' to describe how algorithmic systems inevitably reflect the blind spots and biases of the people who build them, founded the Algorithmic Justice League to advance more accountable and inclusive technology, and has paired her technical research with creative work, including a widely exhibited spoken-word piece titled 'AI, Ain't I a Woman?' depicting AI systems' documented failures to correctly recognize the faces of prominent Black women including Oprah Winfrey, Michelle Obama, and Serena Williams. Her continued advocacy includes testimony before Congress on facial recognition regulation, recognition on Time's 100 Most Influential People in AI and Fortune's naming her 'the conscience of the AI revolution,' and her 2025 election to the board of the Legal Defense Fund, a major civil rights legal organization; she continues to call for mandatory, independent auditing of AI systems before deployment rather than the largely self-regulated industry practices that prevail today.

Sources

1. MIT Technology Review; Algorithmic Justice League official site; Out of the Box (Gender Shades exhibit)