AI Behind Virtual Humans: Communicating the Capabilities and Impact of Artificial Intelligence to the Public through an Interactive Virtual Human Exhibit

虚拟人类展览背后的人工智能旨在通过虚拟人类中使用的人工智能技术（包括面部识别和自然语言处理）向公众传达人工智能（AI）的能力和影响。人工智能已经并将继续深刻影响美国和地球仪的社会。重要的是要让国家的年轻人和未来的劳动力具备人工智能的基本知识。博物馆等非正式环境提供了开放和灵活的机会，帮助青少年和公众了解人工智能。虚拟人提供了一个理想的工具来说明人工智能的许多领域，因为人工智能可以说是构建像人类一样思考和行动的智能的科学。在南加州大学创意技术研究所和加州大学伯克利分校劳伦斯科学大厅的多学科研究人员团队的领导下，该项目将开发一个虚拟人展览，通过与虚拟人的结构化对话吸引游客，同时展示人工智能如何在幕后驱动虚拟人的行为。该展览将为参观者（如亲子、兄弟姐妹和同龄人）提供协作学习体验，以探索AI是什么和不是什么，AI是什么和不能够做什么，以及它将对他们的生活产生什么影响。该项目将探讨三个研究问题：（1）如何设计博物馆展览，让参观者对AI进行协作学习？(2)如何以公众可以理解的方式传达虚拟人背后的复杂AI概念？（3）虚拟人如何以及在多大程度上增加知识并减少对AI的误解？该项目利用了通过数十年的人工智能合作研究开发的现有会话虚拟人技术，包括机器视觉，自然语言处理，自动推理，角色动画和机器学习。设置在博物馆的非正式环境中，展览将根据计算机支持的协作学习中的循证研究进行设计。项目团队将采用混合方法设计，借鉴基于设计的研究方法和实验研究。研究团队将对游客的观察和访谈进行分析，以进行迭代的形成性改进。随机实验研究将在实验室和自然环境中进行，以评估访问者对人工智能的了解。将进行准实验分析，以研究与展品特征和人工智能知识之间的关系。该项目将制作一个互动展览，在劳伦斯科学馆和其他参与博物馆安装一个虚拟人，并测量人工智能学习的仪器。该项目还将建立一个网站，让参观者可以在线体验部分展览，并继续深入了解人工智能和虚拟人技术。该项目具有产生理论和实践进步的潜力，可以帮助公众了解人工智能的能力和道德，在青少年学习者发展人工智能知识的过程中推进知识，并制定在非正式环境中创建协作学习体验的设计原则。研究结果将通过会议报告、学术出版物和社交媒体传播。虚拟人展览将被设计用于传播，并提供给非正式的科学教育社区。该奖项反映了NSF的法定使命，并被认为是值得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的支持。

The AI behind Virtual Humans Exhibit aims to communicate to the public about the capabilities and impact of artificial intelligence (AI) through AI technologies used in Virtual Humans including facial recognition and natural language processing. AI has and will continue to profoundly impact society in the United States and around the globe. It is important to prepare the nation’s youth and the future workforce with fundamental knowledge of AI. Informal settings, such as museums, offer open and flexible opportunities in helping youth and the general public learn about AI. Virtual Humans provide an ideal vehicle to illustrate many fields of AI, as AI is arguably the science of building intelligence that thinks and acts like humans. Led by a multidisciplinary team of researchers with expertise in AI, learning design, and assessment from the Institute for Creative Technologies at University of Southern California and the Lawrence Hall of Science at University of California, Berkeley, this project will develop a Virtual Human exhibit to engage visitors through structured conversations with a Virtual Human, while showcasing how AI drives the Virtual Human’s behavior behind the scenes. The exhibit will include collaborative learning experiences for visitors such as parent-child, siblings and peers to explore what AI is and is not, what AI is and is not capable of, and what impact it will have on their lives.The project will investigate three research questions: (1) How can a museum exhibit be designed to engage visitor dyads in collaborative learning about AI? (2) How can complex AI concepts underlying the Virtual Human be communicated in a way that is understandable by the general public? And (3) How does and to what extent the Virtual Human exhibit increase knowledge and reduce misconceptions about AI? The project leverages existing conversational Virtual Human technology developed through decades of collaborative research in AI, including machine vision, natural language processing, automated reasoning, character animation, and machine learning. Set in the informal setting of a museum, the exhibit will be designed following evidence-based research in Computer Supported Collaborative Learning. The project team will use a mixed methods design, drawing on design-based research methodologies and experimental studies. The research team will conduct analysis of visitor observations and interviews for iterative formative improvement. Randomized experimental studies will be conducted in both lab and naturalistic environments to gauge visitor knowledge about AI. Quasi-experimental analyses will be performed to study the relationship between engagement with exhibit features and AI knowledge. The project will produce an interactive exhibit with a Virtual Human installed at the Lawrence Hall of Science and other participating museums, and instruments to measure AI learning. The project will also produce a website where visitors can experience parts of the exhibit online and continue more in-depth learning about AI and the Virtual Human technology. The project holds the potential for producing theoretical and practical advances in helping the general public develop an understanding of AI capability and ethics, advancing knowledge in the process through which young learners develop knowledge about AI, and formulating design principles for creating collaborative learning experiences in informal settings. The results will be disseminated through conference presentations, scholarly publications, and social media. The Virtual Human exhibit will be designed for dissemination and made available for installations at informal science education communities.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.