AI Institute for Societal Decision Making (AI-SDM)

公共卫生危机或灾害应对等领域的决策具有重大的社会和经济影响。这些领域对决策提出了严峻的挑战，因为它们需要在动态、不确定和资源受限的情况下做出复杂的、可能挽救生命的决策，同时考虑到对决策的接受至关重要的因素，如利益相关者的偏见和对风险、信任和公平的看法。人工智能的进步和数据的可用性可以弥补人类在导航这一复杂决策空间方面的局限性，然而，目前的系统无法考虑利益相关者的心理状态和行为。人工智能社会决策研究所(AI-SDM)将通过开发以人为中心的人工智能进行决策和跨学科培训，将这一机会瞄准社会决策科学和人工智能的融合，以实现应对社会决策挑战的变革性解决方案。通过引入人工智能和社会科学研究人员，AI-SDM将使应急经理、公共卫生官员、急救人员、社区工作者和公众能够快速、数据驱动和资源高效地做出决定，同时还通过考虑管理接受的人为因素来改善结果。AI-SDM的愿景将通过开发新的AI理论和方法、转化研究、培训和推广来实现，不同的大学、政府组织、企业合作伙伴、社区学院、公共图书馆和高中之间的合作伙伴关系将使之成为可能。该研究所将在公共卫生危机和灾难应对中出现的由使用引发的挑战下，确立AI在推进和桥梁人类和自主决策方面的作用。具体地说，基础研究将开发(1)人类决策过程的计算表示，(2)用于集体决策的稳健聚集方法，(3)多目标自主决策支持工具，以及(4)因果推理和反事实推理的相应创新。这些基本重点的灵感来自于并将应用于公平的资源分配以改善公共卫生和灾难结果，通过人类决策提供及时的有针对性的干预以鼓励遵守政策建议，以及通过了解如何通过不同的使用模式来调整采用人工智能决策支持。这项研究将以计算认知科学、社会选择理论、无分布统计、博弈论、随意和反事实推理以及交互和自主机器学习的理论进步为指导。除了通过广泛的合作伙伴网络影响用例域之外，AI-SDM还将通过更广泛的影响努力培养接受过以人为中心的AI培训的下一代劳动力和具有AI意识的公众，包括针对高中教育工作者的专业发展研讨会、针对未被充分代表的学生的丰富和领导活动、跨学科学位和课程、与社区学院和教育合作伙伴的课程合作设计、劳动力培训和公众参与活动。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。

Decision making in domains such as a public health crisis or disaster response has a significant societal and economic impact. These domains present critical challenges for decision-making as they require complex, potentially life-saving, decisions to be made under dynamic, uncertain and resource-constrained scenarios, while accounting for factors that are key to acceptance of the decisions, such as stakeholders' biases and perception of risk, trust, and equity. AI advancements and data availability can complement human limitations in navigating this complex decision space, however, current systems fail to account for the stakeholders' mental states and behavior. The AI institute for Societal Decision Making (AI-SDM) will target this opportunity at the confluence of social decision sciences and AI by developing human-centric AI for decision-making and inter-disciplinary training, to enable transformative solutions to societal decision challenges. By bringing AI and social science researchers, AI-SDM will enable emergency managers, public health officials, first responders, community workers, and the public to make quick, data-driven, and resource-efficient decisions, while also improving outcomes by accounting for human factors governing acceptance. The vision of AI-SDM will be realized via development of novel AI theory and methods, translational research, training, and outreach, enabled by partnerships among diverse universities, government organizations, corporate partners, community colleges, public libraries, and high schools.The institute will establish the role of AI in advancing and bridging human and autonomous decision-making, under the use-inspired challenges of working in environments that are dynamic, uncertain, resource constrained, and require societal acceptance arising in public health crisis and disaster response. Specifically, the foundational research will develop (1) computational representations of human decision processes, (2) robust aggregation methods for collective decision-making, (3) multi-objective autonomous decision support tools, and corresponding innovations in (4) causal and counterfactual reasoning. These foundational foci are inspired by, and will be applied to, equitable resource allocation to improve public health and disaster outcomes, timely targeted interventions informed by human decision-making to encourage adherence to policy recommendations, and adoption of AI decision support by understanding how adoption can be modulated by different use patterns. The research will be guided by theoretical advances in computational cognitive science, social-choice theory, distribution-free statistics, game theory, casual and counterfactual reasoning, and interactive and autonomous machine learning. In addition to impacting use-case domains via a wide network of partners, AI-SDM will develop the next generation of workforce trained on human-centric AI and an AI-aware public via broader impact efforts including professional development workshops for high school educators, enrichment and leadership activities for under-represented students, inter-disciplinary degrees and courses, curriculum co-design with community college and educational partners, workforce training, and public engagement activities.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.