Timnit Gebru

Timnit Gebru is one of the most prominent and consequential critics of how the technology industry builds and deploys artificial intelligence, particularly its treatment of data used to train large-scale models. Born in Ethiopia and educated in the United States, she earned a PhD in computer vision at Stanford under Fei-Fei Li before turning her research toward the social consequences of the systems she had helped build.

Her 2018 paper 'Gender Shades', co-authored with Joy Buolamwini, exposed how commercial facial recognition systems from IBM, Microsoft and Face++ performed dramatically worse on darker-skinned women than on lighter-skinned men, an early and widely cited demonstration that biased training data produces biased, discriminatory outcomes at scale. The paper reshaped industry and regulatory debate over facial recognition and helped catalyse bans and moratoria on its use by police in several US cities.

Gebru's departure from Google in December 2020 became one of the defining controversies in AI ethics. As co-lead of Google's Ethical AI team, she co-authored 'On the Dangers of Stochastic Parrots', a paper warning that ever-larger language models trained on scraped internet data encode bias, consume enormous energy and environmental resources, and can mislead users into perceiving coherence where none exists. Google's objection to the paper and her subsequent exit, which she has described as a firing, drew international attention to the pressure large technology companies place on internal critics of their own data practices.

In 2021 she founded DAIR, an independent research institute explicitly structured to study AI's effects on marginalised communities without the commercial pressures of Big Tech, funding subsequent work on topics including surveillance, labour exploitation in data annotation, and the environmental costs of large-scale data infrastructure. She also co-founded Black in AI, an organisation supporting Black researchers in the field.

Gebru continues to be one of the most cited and most quoted critics of generative AI's data practices, warning regularly about the concentration of power in a small number of data-rich companies and the human and environmental costs embedded in the data supply chains behind modern AI.