CAREER: With Age Comes Wisdom: Leveraging Older Adults' Crystallized Decision-Making Abilities to Develop Adaptive Human-Automation Interfaces for Dynamic Environments

65岁及以上的成年人现在是世界上增长最快的年龄组。与此同时，人工智能（AI）和自动化继续渗透到人类生活的方方面面。然而，尽管几十年来取得了进步，但在复杂的交通、工作和医疗环境中，目前的技术无法解释老年人通过生活经验积累的丰富知识，而这些知识往往会影响他们与人工智能和自动化的互动方式。这项研究旨在通过采用一种新颖的视角来克服这一挑战，该视角关注随着年龄增长而提高的认知能力，而不是那些随着年龄增长而下降的认知能力。具体而言，该项目将建立一个框架：(1)评估老年人如何应用他们的结晶知识来决定是否以及何时在安全关键的动态环境中使用自动化；(2)开发定制的决策支持技术，智能地定制信息以满足该人群的特定需求。项目成果将有助于改变对老年人口的看法，从无能力到有能力，为人工智能的设计和能力提供信息，以更好地符合老年人的思维模式，促进更多的技术使用，提高老年人在工作、交通和休闲环境中安全、独立地使用复杂技术的能力，并减少总体技术培训需求。通过开发互动式在线学习社区和技术平台、以社区为基础的“老龄设计”研究项目以及年度“老龄数字设计”竞赛，一个紧密整合的研究和教育项目将吸引老年人以及本科生、研究生和代表性不足的学生参与并赋予他们权力。在接下来的几十年里，自动驾驶系统将在许多环境中变得更加广泛。这项研究将为老年人的先验知识和经验如何塑造他们与自动化的（预期的）互动提供一个基本的理解，并将这种理解纳入新兴智能系统的设计中。该项目有两个主要研究重点：(1)确定结晶知识应用的人口统计、行为和绩效指标，这些指标可以预测不确定情况下自动化的使用/停用决策；(2)开发一个自适应多模式决策支持系统，为老年人提供实时指导，指导他们何时使用自动化，以最大限度地提高安全性和绩效。一系列的人体实验将在驾驶中进行，这是一项复杂的任务，老年人有多年的经验，人工智能和自动化有望扩展他们的执行能力。将使用参与式设计方法，并从较年轻和较年长的年龄组收集数据，以检查使用先验知识和决策策略如何随着年龄的变化而变化。将构建计算模型来预测自动化使用决策并触发定制反馈。这项工作将有助于推进几个领域的理论、框架和方法，包括老龄化、人机交互、多模式信息表示和决策支持系统。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。

Adults 65 years and older are now the fastest-growing age group worldwide. At the same time, artificial intelligence (AI) and automation continue to penetrate every aspect of human life. Yet despite decades of advances, current technologies in complex transportation, work, and healthcare environments fail to account for the wealth of knowledge that older adults accumulate through their life experiences, which often shapes how they interact with AI and automation. This research aims to overcome this challenge by employing a novel perspective that focuses on the cognitive abilities that improve with age as opposed to those that decline. Specifically, the project will build a framework to: (1) evaluate how older adults apply their crystallized knowledge to decide whether and when to use automation in safety-critical, dynamic settings; and, (2) develop customized decision support techniques that intelligently tailor information to this population's specific needs. Project outcomes will help to change the narrative on older populations from incapable to empowered, inform the design and capabilities of AI to better align with older adults’ mental models, promote increased technology usage, extend older individuals’ ability to safely and independently collaborate with sophisticated technologies in work, transportation, and leisure settings, and reduce overall technology training needs. A tightly integrated research and education program will engage and empower older adults as well as undergraduate, graduate, and underrepresented students through the development of an interactive online learning community and technology platform, community-based ‘design for aging’ research projects, and annual ‘digital design for aging’ competitions. Over the next several decades, the presence of autonomous systems will become even more widespread across many environments. This research will provide a fundamental understanding of how older adults’ prior knowledge and experiences shape their (intended) interactions with automation and incorporate this understanding into the design of emerging intelligent systems. The project has two major research thrusts: (1) identify demographic, behavioral, and performance indicators of application of crystallized knowledge that can predict automation use/disuse decisions in uncertain situations, and (2) develop an adaptive multimodal decision support system that provides older adults with real-time guidance on when to use automation to maximize safety and performance. A series of human-subject experiments will be conducted in driving, a complex task that older adults have years of experience performing and for which AI and automation promise to extend their ability to perform. Participatory design methods will be used, and data will be collected from younger and older age groups to examine how use of prior knowledge and decision-making strategies change as a function of age. Computational models will be constructed to predict automation use decisions and trigger customized feedback. This work will contribute to advancing theories, frameworks, and methods in several areas including aging, human-automation interaction, multimodal information presentation, and decision support systems.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.