Towards a Critical Race Methodology in Algorithmic Fairness

Abstract

We examine the way race and racial categories are adopted in algorithmic fairness frameworks. Current methodologies fail to adequately account for the socially constructed nature of race, instead adopting a conceptualization of race as a fixed attribute. Treating race as an attribute, rather than a structural, institutional, and relational phenomenon, can serve to minimize the structural aspects of algorithmic unfairness. In this work, we focus on the history of racial categories and turn to critical race theory and sociological work on race and ethnicity to ground conceptualizations of race for fairness research, drawing on lessons from public health, biomedical research, and social survey research. We argue that algorithmic fairness researchers need to take into account the multidimensionality of race, take seriously the processes of conceptualizing and operationalizing race, focus on social processes which produce racial inequality, and consider perspectives of those most affected by sociotechnical systems.

Conference on Fairness, Accountability, and Transparency (FAT* '20), January 27-30, 2020, Barcelona, Spain

References in corpus (5)

• Model Cards for Model Reporting
• Equality of Opportunity in Supervised Learning
• The Measure and Mismeasure of Fairness
• Racial categories in machine learning
• Interventions over Predictions: Reframing the Ethical Debate for Actuarial Risk Assessment

Cited by in corpus (66)

• On the Opportunities and Risks of Foundation Models
• Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence
• Measurement and Fairness
• An Empirical Characterization of Fair Machine Learning For Clinical Risk Prediction
• Do Datasets Have Politics? Disciplinary Values in Computer Vision Dataset Development
• Rethinking Fairness: An Interdisciplinary Survey of Critiques of Hegemonic ML Fairness Approaches
• Fairness for Unobserved Characteristics: Insights from Technological Impacts on Queer Communities
• Interactive Model Cards: A Human-Centered Approach to Model Documentation
• Conservative AI and social inequality: Conceptualizing alternatives to bias through social theory
• Towards Intersectionality in Machine Learning: Including More Identities, Handling Underrepresentation, and Performing Evaluation
• AI's Regimes of Representation: A Community-centered Study of Text-to-Image Models in South Asia
• "Garbage In, Garbage Out" Revisited: What Do Machine Learning Application Papers Report About Human-Labeled Training Data?
• Robots Enact Malignant Stereotypes
• Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency
• Four Years of FAccT: A Reflexive, Mixed-Methods Analysis of Research Contributions, Shortcomings, and Future Prospects
• One Label, One Billion Faces: Usage and Consistency of Racial Categories in Computer Vision
• Where Are We? Using Scopus to Map the Literature at the Intersection Between Artificial Intelligence and Research on Crime
• Gender and Racial Bias in Visual Question Answering Datasets
• It's COMPASlicated: The Messy Relationship between RAI Datasets and Algorithmic Fairness Benchmarks
• Between Subjectivity and Imposition: Power Dynamics in Data Annotation for Computer Vision
• Robustness Disparities in Commercial Face Detection
• Perfect density models cannot guarantee anomaly detection
• Can Fairness be Automated? Guidelines and Opportunities for Fairness-aware AutoML
• "What We Can't Measure, We Can't Understand": Challenges to Demographic Data Procurement in the Pursuit of Fairness
• Evaluating the Impact of Social Determinants on Health Prediction in the Intensive Care Unit
• Subverting machines, fluctuating identities: Re-learning human categorization
• Much Ado About Gender: Current Practices and Future Recommendations for Appropriate Gender-Aware Information Access
• An Empirical Analysis of Racial Categories in the Algorithmic Fairness Literature
• From Explanation to Recommendation: Ethical Standards for Algorithmic Recourse
• Controlled Analyses of Social Biases in Wikipedia Bios
• Fairness Through Robustness: Investigating Robustness Disparity in Deep Learning
• Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyond
• Potential for allocative harm in an environmental justice data tool
• ARTiST: Automated Text Simplification for Task Guidance in Augmented Reality
• Extending the Machine Learning Abstraction Boundary: A Complex Systems Approach to Incorporate Societal Context
• Beyond Model Interpretability: Socio-Structural Explanations in Machine Learning
• Causal Modeling for Fairness in Dynamical Systems
• Fairness Hacking: The Malicious Practice of Shrouding Unfairness in Algorithms
• Environment Inference for Invariant Learning
• An Agent-based Model to Evaluate Interventions on Online Dating Platforms to Decrease Racial Homogamy
• Social Norm Bias: Residual Harms of Fairness-Aware Algorithms
• The Neutrality Fallacy: When Algorithmic Fairness Interventions are (Not) Positive Action
• Recommenders with a mission: assessing diversity in newsrecommendations
• Beyond "Fairness:" Structural (In)justice Lenses on AI for Education
• ProportionNet: Balancing Fairness and Revenue for Auction Design with Deep Learning
• Mitigating Dataset Harms Requires Stewardship: Lessons from 1000 Papers
• Measurement as governance in and for responsible AI
• Unlawful Proxy Discrimination: A Framework for Challenging Inherently Discriminatory Algorithms
• Disaggregated Interventions to Reduce Inequality
• Laissez-Faire Harms: Algorithmic Biases in Generative Language Models
• An Introduction to Algorithmic Fairness
• Ethnography and Machine Learning: Synergies and New Directions
• Identities are not Interchangeable: The Problem of Overgeneralization in Fair Machine Learning
• Simplicity Bias Leads to Amplified Performance Disparities
• The Social Construction of Visualizations: Practitioner Challenges and Experiences of Visualizing Race and Gender Demographic Data
• Algorithms in the Stacks: Investigating automated, for-profit diversity audits in public libraries
• Counterfactual Situation Testing: From Single to Multidimensional Discrimination
• Everything is Relative: Understanding Fairness with Optimal Transport
• Toward Substantive Intersectional Algorithmic Fairness: Desiderata for a Feminist Approach
• Fairness without Regret
• A method of supervised learning from conflicting data with hidden contexts
• Mutatis Mutandis: Revisiting the Comparator in Discrimination Testing
• Fairness is in the details: Face Dataset Auditing
• Text as Causal Mediators: Research Design for Causal Estimates of Differential Treatment of Social Groups via Language Aspects
• A Sociotechnical View of Algorithmic Fairness
• Theory In, Theory Out: The uses of social theory in machine learning for social science