False Positives , False Negatives , and False Analyses : A Rejoinder to “ Machine Bias : There ’ s Software Used Across the Country to Predict Future Criminals . And It ’ s Biased Against Blacks . ”

By Anthony W. Flores. 2017

Citations in the corpus

• Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead (2018, neutral): “This is possibly the main flaw identified by criminologists [24] in the ProPublica analysis [25, 26] that accused the proprietary COMPAS recidivism model of being racially biased.”
• Interventions over Predictions: Reframing the Ethical Debate for Actuarial Risk Assessment (2017, neutral): “These individuals contend that the most important consideration regarding bias is whether or not the probability estimates provided by the algorithm are well-calibrated, so that the algorithm performs in a way that is equally accurate across all groups [26, 64] This debate has led to a broader discu”
• Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment (2016, neutral): “Some of the follow-up discussion on this dataset can be found at [3, 11].”
• Fairness Constraints: A Flexible Approach for Fair Classification (2019, unknown): “, 2016b) and some of the follow-up discussion on this dataset (Angwin and Larson; Flores et al., 2016).”
• Exploring the cloud of variable importance for the set of all good models (2020, unknown): “6 ).”
• Human Decision Making with Machine Assistance (2019, unknown): “Prior work has chiefly been interested in the fairness of decisions completely delegated to machines [1, 4, 6, 31] and has extensively studied the distributive [26, 29, 39, 43, 49, 50, 56, 65, 73–75] and procedural [36, 38] fairness of machine decision-making [9, 14, 18, 19, 32, 46], as well as the ”
• Disparate Interactions: An Algorithm-in-the-Loop Analysis of Fairness in Risk Assessments (2019, unknown): “We focused on calibration not as an ideal metric for fairness (recognizing that no perfect metric for fairness can exist [32]), but because it is the most commonly-used approach for evaluating risk assessments in practice [16, 25, 44].”
• Human Perceptions of Fairness in Algorithmic Decision Making: A Case Study of Criminal Risk Prediction (2018, unknown): “Given their potential impact on human lives, concerns have been raised about the fairness of the decisions made by algorithms [1, 5, 7, 22].”