Targeted Infusion Project: Exposing Students to the Social Relevance and Trustworthiness of Artificial Intelligence

历史上的黑人学院和大学本科项目（HBCU- up）通过有针对性的输液项目支持基于证据的创新模式和方法的开发、实施和研究，以提高HBCU本科生的准备和成功，使他们可以追求科学、技术、工程或数学（STEM）研究生课程和/或职业。费耶特维尔州立大学（Fayetteville State University）的这个项目将探讨人工智能（AI）方法对社会的影响及其潜在效益。几乎每个领域的人类知识，从科学和工程，人文和艺术，都可以通过人工智能得到潜在的增强。正在开发和部署计算应用程序，为数据驱动的决策提供人工智能支持，这些决策对人们的生活有实际影响。人工智能算法已被用于解决医疗、执法、军事和政府等领域的不同问题，这仅是几个主要领域。虽然要在这些领域充分发挥其潜力需要更多的时间和努力，但采用人工智能的风险是一个紧迫而重要的问题，因为在进行可信度测试之前过早部署人工智能可能会造成伤害。该项目基于人工智能研究和教育研究，表明计算和非计算课程可以从定制模块中受益，这些模块可以让学生了解人工智能知识和人工智能模型的可信度。人工智能的可信度和社会相关性是一项深刻的多学科努力，该项目正在利用来自不同学科的研究人员和教育工作者的知识。面向过程的指导性探究学习（POGIL）将用于创建基于探究的活动，这些活动将由学生以团队形式完成。大多数活动都涉及在沙箱软件环境中执行的人工智能模型的动手实验工作。通过沉浸在这个过程中，学生将了解人工智能模型是如何工作的，以及如何评估信任，这在根据模型的预测采取行动或部署新模型时至关重要。这个指导过程还提供了对人工智能模型内部工作的见解，学生可以使用它将不可信的模型或预测转换为可信的模型或预测。将建立一个新的教育材料库，以支持在非计算学科中采用人工智能，并通过传统上没有得到充分覆盖的内容来加强计算课程。该项目增加了关于如何激励人工智能研究的知识体系，如何为人工智能技能和领域知识都很重要的现代工作场所做好准备，以及如何培训知情且对社会负责的人工智能创造者和用户。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The Historically Black Colleges and Universities Undergraduate Program (HBCU-UP) through Targeted Infusion Projects supports the development, implementation, and study of evidence-based innovative models and approaches for improving the preparation and success of HBCU undergraduate students so that they may pursue science, technology, engineering, or mathematics (STEM) graduate programs and/or careers. The project at Fayetteville State University will address the impact of Artificial Intelligence (AI) approaches on society and its potential benefits. Almost every area of human knowledge from the sciences and engineering, humanities and the arts, can be potentially enhanced by AI. Computing applications are being developed and deployed to provide AI support for data-driven decisions that have real effects on the lives of people. AI algorithms have been adopted to solve different problems in healthcare, law enforcement, military, and government, to name only a few major areas. While it will take more time and effort to realize its full potential in these areas, the risks of adopting AI is an urgent and important problem because it can potentially cause harm when it is prematurely deployed before being tested for trustworthiness. This project is based on AI research and educational research that suggest computing and non-computing curricula can benefit from customized modules that expose students to the knowledge of AI and the trustworthiness of AI models. The trustworthiness and social relevance of AI is a profoundly multi-disciplinary effort, and the project is leveraging the knowledge of researchers and educators from various disciplines. The Process Oriented Guided Inquiry Learning (POGIL) will be used for creating inquiry-based activities that will be completed by students in teams. Most activities involve hands-on, experimental work with AI models that are executed within a sandbox software environment. Through immersion in this process students will understand how AI models work and how to assess trust, which is essential when taking actions based on the predictions of models, or when deploying a new model. This guided process also provides insights into the inner working of AI models, which students can use to transform an untrustworthy model or prediction into a trustworthy one. A novel repository of educational materials will be created to support the adoption of AI in non-computing disciplines and to strengthen the computing curricula with content that has traditionally not received adequate coverage. This project adds to the growing body of knowledge on how to motivate the study of AI, how to prepare students for the modern-day workplace where both AI skills and domain knowledge are important, and how to train informed and socially responsible creators and users of AI.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.