Joy Buolamwini
Last reviewed
Jun 5, 2026
Sources
31 citations
Review status
Source-backed
Revision
v2 ยท 2,617 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 5, 2026
Sources
31 citations
Review status
Source-backed
Revision
v2 ยท 2,617 words
Add missing citations, update stale details, or suggest a clearer explanation.
Joy Buolamwini is a Canadian-American computer scientist and digital activist known for research on bias in commercial facial recognition and facial-analysis systems. She founded the Algorithmic Justice League, a nonprofit that combines research, art, and advocacy on the social effects of artificial intelligence. As a researcher at the MIT Media Lab she led the 2018 Gender Shades study, written with Timnit Gebru, which found that commercial gender-classification products from IBM, Microsoft, and Face++ misclassified darker-skinned women far more often than lighter-skinned men. She describes herself as a poet of code, uses the phrase the coded gaze for encoded discrimination in computing systems, appears in the 2020 documentary Coded Bias, and wrote the 2023 book Unmasking AI: My Mission to Protect What Is Human in a World of Machines. [1][2][3]
| Field | Detail |
|---|---|
| Full name | Joy Adowaa Buolamwini |
| Born | January 23, 1990, Edmonton, Alberta, Canada |
| Citizenship | Canadian-American |
| Heritage | Ghanaian descent |
| Fields | Computer science, AI ethics, facial-analysis auditing |
| Education | Georgia Institute of Technology (BS, 2012); University of Oxford (MS); MIT (MS, 2017; PhD, 2022) |
| Known for | Gender Shades study; Algorithmic Justice League; the coded gaze |
| Organization | Algorithmic Justice League (founder, 2016) |
| Book | Unmasking AI (2023) |
| Honors | Rhodes Scholar; Fulbright Fellow; Time 100 AI (2023) |
Buolamwini was born on January 23, 1990, in Edmonton, Alberta, Canada, to parents of Ghanaian descent, and she grew up partly in Mississippi and in the Memphis area of Tennessee. Her mother is an artist and her father is an academic in pharmaceutical sciences who was completing doctoral study at the University of Alberta around the time she was born. The family spent part of her early childhood in Ghana and moved to the United States when she was about four years old, after her father took a post at the University of Mississippi. She attended Cordova High School near Memphis and was a competitive pole vaulter and basketball player. She has described an early interest in technology and art, including exposure as a child to Kismet, a sociable robot built at MIT, and she taught herself web languages such as HTML, JavaScript, and PHP while still young. [1][2][17]
She studied computer science at the Georgia Institute of Technology, graduating in 2012, where she held a Stamps President's Scholarship, worked on health informatics, and was a finalist in the university's InVenture Prize competition. In 2011 she worked with the Carter Center on an Android-based assessment tool used in a trachoma-prevention program in Ethiopia. She won a Rhodes Scholarship and earned a master's degree in learning and technology at the University of Oxford, where she was a member of Jesus College. She also held a Fulbright fellowship and worked on mobile technology in Zambia, helping young people there learn to build software, an interest she had earlier formalized through an initiative she called Code4Rights aimed at teaching technology for human-rights causes. She later joined the MIT Media Lab, completing a master's in media arts and sciences in 2017 and a PhD in 2022. Her doctoral thesis, supervised by Ethan Zuckerman, was titled Facing the Coded Gaze with Evocative Audits and Algorithmic Audits. [1][2][4][18][19]
Buolamwini's interest in facial-analysis bias grew out of an art project at the MIT Media Lab called the Aspire Mirror, a device meant to project an inspiring image, such as the face of a person the user admires, onto the user's reflection. While building it she found that off-the-shelf face-detection software struggled to register her own dark-skinned face, and that it tracked her more reliably when she held up a plain white mask. She used this experience to coin the phrase the coded gaze, her term for the way the priorities, preferences, and blind spots of the people who build computing systems become embedded in technology. She also popularized the related idea of being excoded, her word for people who are effectively coded out of systems and left vulnerable as automated tools spread into identification, sorting, and selection tasks. [2][3][20]
Gender Shades is the study that established Buolamwini's reputation in algorithmic fairness. The work began when she noticed that some face-detection software did not register her own dark-skinned face reliably, in one case detecting it only when she held up a white mask. She turned the observation into a structured audit of commercial facial-analysis products. [3][5]
Much of the public benchmark data used to evaluate such systems was skewed toward lighter-skinned and male subjects, so Buolamwini assembled a more balanced test set of more than 1,200 images, which she named the Pilot Parliaments Benchmark and drew from members of national parliaments in Africa and Europe. Working with a dermatologic surgeon, she labeled the images by skin tone using the six-point Fitzpatrick scale, and she grouped subjects by perceived gender as well, allowing an intersectional analysis across both attributes. [3][5][6]
The study evaluated the gender-classification functions of three commercial systems, from IBM, Microsoft, and the Chinese company Megvii, whose product is known as Face++. The paper, Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification, was written with Timnit Gebru, then a graduate student at Stanford University, and presented at the first Conference on Fairness, Accountability, and Transparency in 2018, appearing in Proceedings of Machine Learning Research, volume 81. [5][6]
The results showed a consistent pattern. Error rates for classifying lighter-skinned men were as low as 0.8 percent, while error rates for darker-skinned women reached as high as 34.7 percent, a gap of about 33.9 percentage points. For the darkest-skinned women in the test set, classified as Fitzpatrick type VI, error rates rose further, to between roughly 46 and 47 percent on some systems. The paper argued that aggregate accuracy figures can hide large disparities for subgroups, and it called for accuracy to be reported across skin tone and gender groups, along with greater transparency and accountability in commercial systems. [5][6]
| Subgroup | Reported classification error |
|---|---|
| Lighter-skinned men | as low as 0.8 percent |
| Lighter-skinned women | higher than lighter-skinned men |
| Darker-skinned men | higher than lighter-skinned men |
| Darker-skinned women | up to 34.7 percent |
| Darkest-skinned women (Fitzpatrick VI) | up to about 46 to 47 percent on some systems |
IBM and Microsoft responded to the findings, and both published updates intended to reduce the disparities the study had measured. Gender Shades became one of the most cited papers in AI ethics, accumulating several thousand citations, and it is widely credited with helping to establish algorithmic auditing as a field and with shaping later policy debate over facial recognition. [3][5][21]
Buolamwini founded the Algorithmic Justice League in 2016. The organization states that its mission is to combat the harms and biases of AI and to raise the voices of the communities most affected by automated systems. It combines academic research with art and public advocacy, and Buolamwini often frames its concern in terms of the coded gaze, her phrase for the way the priorities and biases of system creators become embedded in technology. [1][2][3]
In 2019 Buolamwini and Inioluwa Deborah Raji published a follow-up study, Actionable Auditing, at the AAAI/ACM Conference on Artificial Intelligence, Ethics, and Society. It re-examined the targeted companies and added Amazon's Rekognition service, reporting that Rekognition had a large gap between gender-classification error rates for darker-skinned women and lighter-skinned men. Amazon publicly disputed the methodology and conclusions, and the exchange drew wide attention to the practice of auditing commercial systems and naming the results. [7][8]
The group has run several public campaigns. The Safe Face Pledge, launched in 2018 with the Georgetown Center on Privacy and Technology, asked organizations to commit to limits on harmful uses of facial-analysis technology, and it was later retired. In 2020 the group released Voicing Erasure, a spoken-word piece about racial bias in automatic speech recognition. In 2021 it worked with the consulting firm O'Neil Risk Consulting and Algorithmic Auditing and the company Olay on Decode the Bias, an audit of a skin-analysis tool that reported higher accuracy for lighter-skinned users. The league also runs an initiative called the Community Reporting of Algorithmic System Harms, or CRASH, which gathers accounts of harm from automated systems and works on tools and norms for accountability. [3][9][10][22]
Buolamwini's work reached a broad audience through the documentary Coded Bias, directed by Shalini Kantayya. The film premiered at the Sundance Film Festival in January 2020, follows Buolamwini's research and the wider debate over facial recognition and civil liberties, and began streaming on Netflix on April 5, 2021, after a national broadcast on the PBS series Independent Lens. It was nominated for awards including an NAACP Image Award for Outstanding Documentary. [11][12][23]
In 2023 Buolamwini published Unmasking AI: My Mission to Protect What Is Human in a World of Machines, issued by Random House. The book recounts how she uncovered what she calls the coded gaze and how she built the Algorithmic Justice League into a movement against AI harms. It combines memoir with an account of her research and advocacy, applies an intersectional lens to both the technology industry and the research sector, and became a national bestseller, with recognition that included selection as a Los Angeles Times Book Prize finalist and a Los Angeles Times best book of the year. [13][14][24]
Buolamwini's audits became reference points in policy debates over facial recognition in the United States. She testified before Congress in 2019, appearing before the House Committee on Oversight and Reform on May 22 and before the House Committee on Science, Space, and Technology on June 26, where she presented findings on accuracy disparities and argued for stronger oversight of the technology. [15][16]
Her work is often cited alongside a series of decisions in 2020, when several technology companies stepped back from selling facial recognition for policing. IBM announced that it would stop offering general-purpose facial recognition, while Amazon and Microsoft announced moratoria on sales of the technology to police, the latter pending federal regulation. The Gender Shades and Actionable Auditing studies are frequently named as part of the evidence base that informed those moratoria and the broader push for restrictions at the city and state level. [3][7][8]
Buolamwini was among the experts the Biden administration consulted while preparing its October 30, 2023, executive order on the safe, secure, and trustworthy development of artificial intelligence, the directive numbered Executive Order 14110. She publicly welcomed parts of the order while arguing that it did not go far enough on redress for people harmed by automated systems in areas such as hiring, housing, and criminal justice, a continuing theme in her advocacy on AI governance. [25][26]
Buolamwini has continued to lead the Algorithmic Justice League and to publish audits and reports on deployed systems. In 2025 the league released findings from a multi-year study of the use of facial recognition by the United States Transportation Security Administration at airports, reporting that most travelers were not given clear verbal notice of a right to opt out and that signage was often missed. She has remained an active public speaker, and her TED talk on algorithmic bias has been viewed well over one million times. In 2025 she joined the board of directors of the Legal Defense Fund, a civil-rights legal organization. [27][28][29]
Buolamwini's academic honors include a Rhodes Scholarship, a Fulbright fellowship, a Google Anita Borg Memorial Scholarship, an Astronaut Scholarship, and a Stamps President's Scholarship from Georgia Tech. Her TED talk on algorithmic bias has been viewed widely, and she has spoken on the subject at venues such as the World Economic Forum and the United Nations. She was named to the Time 100 AI list in 2023 and to Fortune's list of the world's greatest leaders in 2019. [1][2][13]
She has received a range of further honors. She appeared on the BBC's 100 Women list in 2018 and on the inaugural Time 100 Next list in 2019, was recognized by the Carnegie Corporation of New York as a Great Immigrant in 2020, and received the American Society for Quality's Hutchens Medal in 2022. In 2024 Dartmouth College awarded her an honorary Doctor of Science degree at its June commencement, and she received a digital civil rights award presented jointly by the NAACP and the Archewell Foundation. She is also listed among the recipients of the MIT Media Lab Disobedience Award and the Morals and Machines Prize. [17][27][30][31]
| Work or role | Year | Notes |
|---|---|---|
| Gender Shades paper (with Timnit Gebru) | 2018 | Intersectional audit of commercial gender classification |
| Algorithmic Justice League | 2016 | Founder; research, art, and advocacy nonprofit |
| Actionable Auditing (with Inioluwa Deborah Raji) | 2019 | Follow-up audit adding Amazon Rekognition |
| Coded Bias (documentary) | 2020 | Directed by Shalini Kantayya; on Netflix 2021 |
| Unmasking AI (book) | 2023 | Random House; national bestseller |
| Congressional testimony | 2019 | House Oversight and Science committees |
| Legal Defense Fund board | 2025 | Board of directors |