Using sensitive personal data may be necessary for avoiding discrimination in data-driven decision models

By Indrė Žliobaitė, Bart Custers. 2016

129 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Safety Research, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety Research

In Sociology and Political Science

Citations in the corpus

Fairness in Criminal Justice Risk Assessments: The State of the Art (2017, background): “Under these circumstances, it is possible for a risk instrument to have predictive parity but not be well calibrated.”
Blind Justice: Fairness with Encrypted Sensitive Attributes (2018, background): “In order to check and enforce such requirements, the mod-eler must have access to the sensitive attributes for individuals in the training data—however, this may be undesirable for several reasons (ˇZliobait˙e & Custers, 2016).”
What We Can't Measure, We Can't Understand: Challenges to Demographic Data Procurement in the Pursuit of Fairness (2020, background): “Perceiving this potential conflict between data availability and making systems less discriminatory, legal scholars have argued for regulatory provisions for collecting demographic data to audit algorithmic bias [72, 79, 82, 84, 86].”
Algorithms, Platforms, and Ethnic Bias: An Integrative Essay Forthcoming in Phylon: The Clark Atlanta University Review of Race and Culture (2018, unknown): “…category in the algorithm to overcome possible discrimination due to indirect factors could expose the responsible organization to the charge of using protected data categories for decisions and thus legal violations (d'Alessandro, O'Neil, & LaGatta, 2017; Žliobaitė & Custers, 2016, p. 3).”
Bias in data‐driven artificial intelligence systems—An introductory survey (2020, background): “…neighborhoods in U.S. cities are highly correlated with race, and this fact has been used for systematic denial of services such as bank loans or same-day purchase delivery.4 Rather, including sensitive features in data may be beneficial in the design of fair models (Zliobaite & Custers, 2016).”
Obfuscation via Information Density Estimation (2019, background): “, political preference), and (ii) we have access to a reference dataset from which we can fit the TIDE (though this is difficult to avoid as discussed in [59]).”
Repairing without Retraining: Avoiding Disparate Impact with Counterfactual Distributions (2019, methodology): “Collecting and Predicting Sensitive Attributes : Our approach requires collecting data on sensitive attributes, such as race or gender, which may infringe privacy (though this is diﬃcult to avoid as discussed in Žliobait˙e and Custers, 2016).”
Awareness in practice: tensions in access to sensitive attribute data for antidiscrimination (2019, background): “mplementing fairness. Zliobaite and Custers use theoretical and linear regression-based examples to argue that sensitive data must be included in the modeling process in order to avoid discrimination [79]. Focusing on the European regulatory environment, their study distinguishes between direct and ”