2021/02/08 by Jesse Russell, Russell, Jesse
Computer Science · Social Sciences · #Adversarial Robustness in Machine Learning #Applications (stat.AP) #Artificial Intelligence in Law #Criminal Justice and Corrections Analysis #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.2102.04342
openalex publication_date 2021/02/08 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
This paper explores how different ideas of racial equity in machine learning,\nin justice settings in particular, can present trade-offs that are difficult to\nsolve computationally. Machine learning is often used in justice settings to\ncreate risk assessments, which are used to determine interventions, resources,\nand punitive actions. Overall aspects and performance of these machine\nlearning-based tools, such as distributions of scores, outcome rates by levels,\nand the frequency of false positives and true positives, can be problematic\nwhen examined by racial group. Models that produce different distributions of\nscores or produce a different relationship between level and outcome are\nproblematic when those scores and levels are directly linked to the restriction\nof individual liberty and to the broader context of racial inequity. While\ncomputation can help highlight these aspects, data and computation are unlikely\nto solve them. This paper explores where values and mission might have to fill\nthe spaces computation leaves.\n