Collaborative Research: Standard Grant: Artificial Intelligence and Predictive Policing: An Ethical Analysis

合作研究：标准拨款：人工智能和预测性警务：伦理分析

• 批准号: 1917712
• 负责人: Duncan Purves
• 金额: $ 26.96万
• 依托单位: University of Florida
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-01 至 2023-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1917712&HistoricalAwards=false
• 关键词: Collaborative; Research; Standard; Grant; Artificial

This project is a study of the use of artificial intelligence to assist police on patrol by using massive troves of historical crime data to train machine learning algorithms to anticipate the timing and location of criminal activity. Predictive policing has come under intense criticisms from civil rights groups, academics, and communities that have been subject to predictive policing. These criticisms include its recapitulation of racially biased patterns of policing, its further burdening of marginalized communities, and its infringement on the liberties of targeted communities. The researchers propose to examine these claims and to develop best practices for the development and deployment of algorithmic policing programs. The results of this project promise to benefit police departments, communities patrolled based on algorithmic crime predictions, and public understanding of the societal and ethical implications of predictive policing. The aims of this grant are to examine and develop viable ethical frameworks for the assessment of predictive policing practices. In doing so, the researchers will propose and evaluate novel considerations that might bear on the ethics of predictive policing. They will also develop empirically grounded recommendations for the ethically sensitive and effective development and use of predictive policing. The results of this project promise to advance understanding and illuminate fertile new areas of research in moral philosophy, technology ethics, sociology, and criminology. They also will have clear relevance for other algorithmic technologies with similar implications for justice including algorithms that predict criminal sentences, make healthcare diagnoses, and serve as gatekeepers to government benefits (all of which are currently in use).This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

这个项目是关于使用人工智能来帮助巡逻警察的研究，方法是使用大量的历史犯罪数据来训练机器学习算法，以预测犯罪活动的时间和地点。预测性警务受到了民权组织、学者和社区的强烈批评，这些团体一直受到预测性警务的影响。这些批评包括重复带有种族偏见的治安模式，进一步加重边缘化社区的负担，侵犯目标社区的自由。研究人员建议对这些说法进行审查，并为开发和部署算法警务程序开发最佳实践。该项目的成果有望惠及警察部门、基于算法犯罪预测进行巡逻的社区，以及公众对预测性警务的社会和伦理影响的理解。这笔赠款的目的是审查和制定可行的道德框架，以评估预测性警务做法。在这样做的过程中，研究人员将提出并评估可能与预测性警务的伦理有关的新考虑。他们还将为在道德上敏感和有效地发展和使用预测性警务制定有经验的建议。这一项目的结果有望促进理解，并照亮道德哲学、技术伦理学、社会学和犯罪学的新的研究领域。它们还将与其他对司法有类似影响的算法技术具有明显的相关性，包括预测刑事判决、进行医疗诊断和充当政府福利的守门人的算法(所有这些都在使用中)。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

A Machine Learning Evaluation Framework for Place-based Algorithmic Patrol Management

基于地点的算法巡逻管理的机器学习评估框架

• DOI: 10.2139/ssrn.4549861
• 发表时间: 2023
• 期刊: SSRN Electronic Journal
• 作者: Purves, Duncan;Jenkins, Ryan
• 通讯作者: Jenkins, Ryan

Public Trust, Institutional Legitimacy, and the Use of Algorithms in Criminal Justice

公众信任、机构合法性以及刑事司法中算法的使用

• DOI: 10.5406/21520542.36.2.03
• 发表时间: 2022
• 期刊: Public Affairs Quarterly
• 作者: Purves, Duncan;Davis, Jeremy
• 通讯作者: Davis, Jeremy

Should Algorithms that Predict Recidivism Have Access to Race?

预测累犯的算法是否应该考虑种族问题？

• DOI: 10.5406/21521123.60.2.07
• 发表时间: 2023
• 期刊: American Philosophical Quarterly
• 影响因子: 0.8
• 作者: Purves, Duncan;Davis, Jeremy
• 通讯作者: Davis, Jeremy

Fairness in Algorithmic Policing

算法监管的公平性

• DOI: 10.1017/apa.2021.39
• 发表时间: 2022
• 期刊: Journal of the American Philosophical Association
• 影响因子: 1
• 作者: Purves, Duncan

Five ethical challenges facing data-driven policing

数据驱动警务面临的五项道德挑战

• DOI: 10.1007/s43681-021-00105-9
• 发表时间: 2022
• 期刊: AI and Ethics
• 作者: Davis, Jeremy;Purves, Duncan;Gilbert, Juan;Sturm, Schuyler
• 通讯作者: Sturm, Schuyler