October 7, 2022
In May, Dr Timnit Gebru was named as one of Time Magazine 100 Most Influential People 2022 in a list packed with global A-list stars of film, music, sports and politics. Ahead of her headline appearance at BTF 22 in October, we are profiling her leading role in the world of AI ethics, and uncovering how she has gone beyond the tech community and into the pages of Time alongside the likes of Oprah Winfrey, Zendaya and Joe Biden.
Gebru holds a PhD from Stanford University where she studied computer vision, having been born and raised in Ethiopia by Eritrean parents. She researched demographic patterns and the possibilities of predicting behaviours using large scale publicly available images. For example, predicting voting patterns in neighbourhoods based on Google Streetview images showing the makes of cars in those neighbourhoods. Gebru’s experience of being only one of a handful of black researchers at the largest and most prestigious AI conference led her to co-found Black in AI, and what started as a workshop is now a thriving community.
Next Gebru joined the Fairness Transparency Accountability and Ethics (FATE) in AI group at Microsoft Research, collaborating with Dr Joy Boulamwini at Massachusetts Institute of Technology (MIT) on a groundbreaking and now famous 2018 research paper “Gender Shades”. In 2020 this research project became the focus of the award-winning documentary Coded Bias, now available on Netflix, with reviews liberally using words like “chilling”, “terrifying”, “engaging”, “dismaying”, “explosive”, “thought-provoking”, “inspirational”.
The Gender Shades project uncovered that various facial recognition software, used by many big tech companies were systemically biased. Not only did the software identify men better than women, it performed poorly on darker-skinned individuals, and the worst for darker-skinned females.
Gebru later joined Google, as a co-lead on the Ethical AI team and was catapulted to the centre of an unprecedented public controversy. The catalyst was her co-authoring of a paper on the possible risks of large language models (LLMs), which are machine learning algorithms that can recognise, predict, and generate language based on very large text-based data sets). The paper ‘On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?’ highlighted that LLMs, like the ones used at Google, not only have huge environmental and financial costs, but also ethical costs. Gebru cited wider societal costs such as perpetuating negative stereotypes, promoting racist, misogynist and ableist viewpoints, increasing extremist ideology and enshrining in code dominant normative narratives and potential outcomes from use of the tech.
Google asked her to remove herself from the paper, which had yet to be published, and any others affiliated with Google. When Gebru pushed back, she was fired, although Google still maintains that Gebru quit. Timnit’s firing from Google fueled notable controversy and had the world of AI research in uproar so much so that letter signed by 2695 ‘Googlers’ and 4302 academic, industry, and civil society supporters condemned the move.
In the past, many big tech companies have in the past been called out for the mistreatment of women and people of colour, as well as the power to silence and harass researchers. Big tech employs thousands of the world's top researchers, and often donates significant capital to universities. Many critics suggest that in exchange they have significant influence over the research agenda and what does and does not make it to publication.
Since Google, Gebru continues to shine a light on racial disparities in the tech industry, pioneer the field of ethical AI, and has recently started her own research organisation. Distributed AI Research Institute (DAIR) is a bold and disruptive initiative to conduct and promote much-needed independent research, out of big tech’s reach.
To quote Dr. Safiya Noble, “It takes courage to speak truth to the most powerful technology companies in the world. Timnit Gebru is a truth teller.”
Timit Gebru will be in conversation with Teanna Barrett to talk about ethics, AI and why she walked away from Big Tech at 16:00 on Tuesday 11th October 2022. Tickets are available at https://www.blacktechfest.com/tickets