Algorithmic Reparation Workshop

University of Michigan September 30-October 1, 2022

How to Apply ↗

No longer accepting applications. Registration is closed.

Wired Magazine Feature ↗

A Move for ‘Algorithmic Reparation’ Calls for Racial Justice in AI

Background and Purpose

Machine learning has an inequality problem that is now widespread and well known. The field of “fair machine learning” (FML) has emerged in response, positing mathematical correctives to account for and remove direct and proxy indicators of protected class attributes–race, class, gender, disability etc–within machine learning models. Although FML predominates and continues to thrive, its effects have been wanting, and thinkers are beginning to challenge the “fairness” value standard (Birhane and Guest 2020; Bui and Noble 2020; Davis et al. 2021; Hanna et al. 2020; Hoffmann 2019; Mohamed et al. 2020; So et al. 2022). 

Fairness models seek to erase demographic differences and achieve unbiased outputs. Such aspirational neutrality is intrinsically flawed, ignoring the ways history, identity, and social systems entwine. In this way, “fairness” approximates colorblind racism and its gendered, heteronormative, and ableist cousins.

Algorithmic Reparation is a response and alternative to FML, one that centralizes rather than obviates levers of inequality in machine learning systems. Rooted in theories of Intersectionality (Cho et al. 2013; Crenshaw 1990; Collins 2002, 2019) and movements for reparation (Bittker, 1972; Coates, 2014; Henry, 2009), this approach is committed to empowerment at the margins and systemic redress. First introduced in an article published by Big Data & Society (Davis, Williams and Yang 2021), we invite participants to begin actioning algorithmic reparation in a 2-day workshop at the University of Michigan, September 30-October 1, 2022.

With funding from the National Center for Institutional Diversity, co-hosted by the Digital Studies Institute and the Center for Ethics, Society, & Computing at the University of Michigan, the Humanising Machine Intelligence Project at the Australian National University, and the Tech Ethics Center at the University of Notre Dame, the workshop will combine efforts from social scientists, computer scientists, activist leaders, and industry representatives. The workshop includes invited panel presentations and hands-on exercises, featuring Algowritten , TheirTube, and others, that attend to machine learning across domains and within social and institutional contexts.

Application deadline has passed

Co Directors