Thread by @jovialjoy

venturebeat.com/2020/08/23/the…
This analysis of the Gender Shades paper seems to miss how it has been employed to shift policy + industry practices contribute to legislative wins against harmful tech, & expose the lie of technical superiority of AI powered systems and their makers.1/n

Gender Shades and the following Actionable Auditing paper were positioned in a manner that opened up space for computer scientists and AI researchers to actively oppose the police use of facial recognition technology. medium.com/@bu64dcjrytwit… 2/n

GS and AA papers authored by young black women challenged norms of evaluation in computer vision, ML, & biometric tech by exemplifying why intersectionality cannot be ignored. The take away is not "optimize oppressive systems" but to rethink the technologies evaluated 3/n

The Gender Shades project was not just a series of academic papers. The follow on NYT oped called for specific action in 2018- a federal moratorium on police use of facial recognition and led to invitations to brief elected officials on AI harms. nytimes.com/2018/06/21/opi… 4/n

This was followed by testimonies I gave at the federal, state, and local levels which have contributed to ongoing efforts to resist surveillance technology. bostonglobe.com/2020/06/24/met… 5/n

If third wave "AI ethics" is characterized by addressing power in both words and actions, @AJLUnited -using not just research but also poetry and storytelling- is certainly not an organization focused on mere data representation as a cheap substitute for systemic change. 6/n

The Gender Shades paper also paved the way for AI, Ain't I A Woman? to move from performance metrics to performance arts to again challenge the notion of the technical sophistication of AI products from leading tech companies. 7/n

In Actionable Auditing, @rajiinio and I wrote about moving from notions of algorithmic fairness to the urgent need for algorithmic justice which I reiterated in my @Stanford Keynote Fall 2019: "The Coded Gaze and the Dangers of Supremely White Data."

I am hopeful future pieces will approach Genders Shades with a contextual awareness about how it has been employed well beyond the ML fairness space. I am glad I have an opportunity to write a book where I can add more color to the work. hollywoodreporter.com/news/joy-buola…