NRT-HDR: Computational Research for Equity in the Legal System" (CRELS)

NRT-HDR：法律体系公平的计算研究”（CRELS）

• 批准号: 2243822
• 负责人: David Harding
• 金额: $ 300万
• 依托单位: University of California-Berkeley
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-07-01 至 2028-06-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2243822&HistoricalAwards=false
• 关键词: NRT; HDR; Computational; Research; Equity

The criminal legal system is an important driver of inequities and social and economic polarization, and legal institutions are at the leading edge of use – and misuse – of artificial intelligence. The increasing availability of “big data” from (and about) criminal legal systems – and the people who are enmeshed in them – provides a new opportunity to illuminate inequities and their sources. This National Science Foundation Research Traineeship (NRT) award to the University of California, Berkeley will develop novel interventions to reduce inequities and their resulting harms in criminal legal systems. New scientific knowledge will be generated through the development of tools for large-scale, "human-in-the-loop" analysis of criminal justice data, and will be used in the generation of new insights regarding legal system processes, impacts, and institutions. Faculty and trainees will collaborate across disciplines to simultaneously address social-science and policy questions regarding equity and criminal legal institutions, the development of tools and methods for leveraging newly available data from the criminal legal system, and ethical and social implications of big data and AI in the context of criminal justice. This NRT will train a new generation of researchers interested in computational approaches to equity and legal systems, enabling them to develop and evaluate public policy solutions that can mitigate social and economic polarization. It will also train a diverse workforce with flexible and transferrable computational skills, while also training social and data scientists in ethical AI and its social implications. It will create a transformative, cross-disciplinary model for graduate training at Berkeley and elsewhere, while also developing a broad-based recruiting and mentoring program to enhance training of students from underrepresented groups, which, in turn, helps to diversify the STEM workforce. The project anticipates training 50 PhD students, including 25 funded trainees, from the Social Sciences, Computer Science, and Statistics. Recent public and policy interest in the criminal legal system coupled with new government efforts to make data public and leverage data for public policy creates new opportunities to study the criminal legal system, but only if such data can be made ready for analysis. The criminal legal system is critical terrain for evaluating how pervasive data collection and algorithmic decision-making can be brought into the service of society, while addressing potent challenges that can accompany these approaches. Big data and AI can give us broader and more precise knowledge of the dynamics of social systems and hold potential to increase transparency and support fairer decision-making. At the same time, areas relating to criminal justice have seen a massive expansion of surveillance, data production and reuse, and algorithmic decision-making often without oversight, recourse, or evidence about effectiveness in addressing underlying issues. Data technologies in criminal justice have grounded new social schema of classification and accompanying social hierarchies – from recidivism risk scores to predictive policing – with important implications for opportunity and life chances. Our goal is to develop tools for continuous ingestion, integration, and cleaning of structured and unstructured data, and the analysis of such data. Using a combination of large pre-trained AI models coupled with data management and human-computer interaction techniques, we will develop tools to ingest information from various government and online sources, turning it into structured data for analysis. These efforts will lead to novel and generalizable tools for semi-autonomous and continuous data processing, as well as integration at scale, that also preserves privacy and promotes equity.The NSF Research Traineeship (NRT) Program is designed to encourage the development and implementation of bold, new potentially transformative models for STEM graduate education training. The program is dedicated to effective training of STEM graduate students in high priority interdisciplinary or convergent research areas through comprehensive traineeship models that are innovative, evidence-based, and aligned with changing workforce and research needs.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.

刑事法律的系统是不平等以及社会和经济两极分化的重要驱动力，而法律的机构处于人工智能使用和滥用的前沿。越来越多的来自（和关于）刑事法律的系统的“大数据”--以及卷入其中的人--为阐明不平等及其根源提供了新的机会。这个国家科学基金会研究培训（NRT）奖给加州大学伯克利分校将开发新的干预措施，以减少不平等及其造成的危害，在刑事法律的系统。 将通过开发对刑事司法数据进行大规模“人在回路”分析的工具产生新的科学知识，并将用于产生关于法律的系统程序、影响和机构的新见解。教师和学员将跨学科合作，同时解决有关公平和刑事法律的机构的社会科学和政策问题，开发工具和方法，以利用刑事法律的系统的新数据，以及大数据和人工智能在刑事司法背景下的伦理和社会影响。该NRT将培养对公平和法律的系统的计算方法感兴趣的新一代研究人员，使他们能够开发和评估可以缓解社会和经济两极分化的公共政策解决方案。它还将培养具有灵活和可转移计算技能的多元化劳动力，同时还将培训社会和数据科学家的道德人工智能及其社会影响。它将为伯克利和其他地方的研究生培训创建一个变革性的跨学科模式，同时还将开发一个基础广泛的招聘和指导计划，以加强对代表性不足群体的学生的培训，这反过来又有助于使STEM劳动力多样化。该项目预计将培训50名博士生，包括25名受资助的学员，来自社会科学，计算机科学和统计学。 最近公众和政策对刑事法律的系统的兴趣，加上政府公开数据和利用数据制定公共政策的新努力，为研究刑事法律的系统创造了新的机会，但前提是这些数据可以随时进行分析。刑事法律的系统是评估如何将普遍的数据收集和算法决策带入社会服务的关键领域，同时解决伴随这些方法的潜在挑战。大数据和人工智能可以让我们更广泛、更精确地了解社会系统的动态，并有可能提高透明度，支持更公平的决策。与此同时，与刑事司法有关的领域已经看到监视、数据生产和重用以及算法决策的大规模扩展，而这些决策往往没有监督、追索权或关于解决根本问题的有效性的证据。刑事司法中的数据技术为新的社会分类模式和伴随的社会等级制度奠定了基础-从累犯风险评分到预测性警务-对机会和生活机会具有重要影响。我们的目标是开发用于持续摄取、集成和清理结构化和非结构化数据以及分析此类数据的工具。我们将使用大型预训练人工智能模型与数据管理和人机交互技术相结合，开发从各种政府和在线来源获取信息的工具，将其转化为结构化数据进行分析。这些努力将导致新颖的和可推广的工具，用于半自主和连续的数据处理，以及大规模的集成，同时保护隐私和促进公平。NSF研究培训（NRT）计划旨在鼓励开发和实施大胆的，新的潜在变革模型，用于STEM研究生教育培训。该计划致力于通过创新的、基于证据的、与不断变化的劳动力和研究需求相一致的综合培训模式，在高优先级的跨学科或融合研究领域对STEM研究生进行有效培训。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。