Gebru, Timnit

Impact on the world

Co-authored the 2018 'Gender Shades' study with Joy Buolamwini, already profiled in this collection, demonstrating that commercial facial recognition systems from IBM, Microsoft, and Face++ misclassified darker-skinned women's gender at error rates as high as 34.7 percent, compared to under 1 percent for lighter-skinned men, research that directly led all three companies to overhaul their systems and contributed to IBM discontinuing its facial recognition product entirely in 2020; she co-founded Black in AI in 2017 to increase Black researcher representation in the field, and her December 2020 departure from Google, where she co-led the company's Ethical AI team, became a major public controversy after Google's head of AI research stated a paper she had co-authored on the risks of large language models did not meet internal publication standards, prompting roughly 2,700 Google employees and over 4,300 outside academics and civil society supporters to sign letters protesting what they characterized as her firing, and leading nine members of Congress to formally request that Google clarify the circumstances of her exit.

Life & background

Born in Addis Ababa, Ethiopia around 1982 to an Eritrean family, Gebru fled political conflict and eventually settled in the United States as a teenager, going on to earn her PhD from the Stanford Artificial Intelligence Laboratory studying computer vision under Fei-Fei Li, with a dissertation examining how large-scale image data could reveal sociological patterns. She completed a postdoctoral fellowship in Microsoft Research's Fairness, Accountability, Transparency, and Ethics in AI group before her widely cited 2018 collaboration with Joy Buolamwini exposed severe racial and gender disparities in commercial facial recognition accuracy, a finding she has said reflects a deeper pattern in the field, that algorithms trained on biased or unrepresentative data will reliably reproduce and sometimes amplify that bias regardless of how neutral the underlying mathematics may appear. Google hired her in 2018 specifically to help ensure its AI products did not perpetuate discrimination, a role in which she recruited prominent researchers of color, published influential papers on algorithmic risk, and grew increasingly vocal internally about experiences of racism and sexism within the company; her December 2020 departure, following internal conflict over a co-authored paper examining the environmental and bias risks of increasingly large language models, drew sustained public attention as a flashpoint in the broader debate over whether large technology companies can credibly police their own AI ethics research when that research threatens their commercial interests. She went on to found the Distributed Artificial Intelligence Research Institute, an organization explicitly structured outside major tech company influence, and has continued publishing widely on algorithmic bias and the societal risks of large-scale AI systems; she was named to the BBC's 100 Women list in 2023 and received Carnegie Corporation's Great Immigrants Award the same year.

Sources

1. TIME Magazine; Klover.ai; Out of the Box (Gender Shades exhibit); Algorithmic Justice League official site