May 1, 2024 – August 31, 2024
The concept of “algorithmic reparation” brings together theories of intersectionality with acts of repair, with the goal of recognizing and rectifying structural inequity. Reparative AI applies algorithmic reparation to critical cultural studies of AI and “smart” technologies to formulate avenues for algorithmic and AI reform. A reparative approach to AI provides both a theoretical framework and toolbox to fix AI – emphasizing questions of power and sociohistorical context to name, unmask, and provide redress for algorithmic harms. Each team member brings a unique perspective on intersections between digital media, historically marginalized communities, and the everyday experience of algorithmic systems. While all members share a common understanding of what reparative AI might be, we will work together to further solidify a definition of what reparative AI is.
Collectively, the project team is focused on research questions that encompass the lifecycle of AI, from its historicization and conception, industry practices and regulation, and everyday use. Our approach, arising from a cultural studies tradition of tracking historical and contemporary networks of power and culture, sees each of these moments as sites of struggle where the meaning of AI – and the socio-technical power to construct that meaning – is enacted and contested. In addition, our reparative approach recognizes the entanglements of AI within the rhythms of everyday life as well as the institutional logics and biases within which they are embedded. During Summer 2024, we will collaboratively conceptualize this theoretical framework further, and work to identify and disambiguate global practices of AI repair focusing on the questions: What are existing practices of repair, both locally and globally? What is an interdisciplinary theoretical framework for reparative AI?
Reparative AI is led by PI Germaine Halegoua, Associate Professor of Communication and Media.
Image: Mimi Onuoha, The Library of Missing Datasets, mixed media installation (2016).