Understanding human interaction with automated and algorithmic systems

自主系统、高级算法和人工智能是新兴技术。 虽然这些技术旨在减少人类的脑力工作量，但如果它们的设计没有考虑到用户，这些技术可能会增加工作量并降低态势感知能力。 该提案探索使用人因工程的方法，认知工作分析（CWA），以帮助人类操作员并改善人类决策的方式定义自主，算法或人工智能系统的要求。 CWA非常适合这个问题，因为CWA的开发是为了帮助人们更好地处理像这里提出的复杂系统。 特别是，CWA利用了强大的推理结构和人类信息处理模型，推进了当前的人类交互方法。 Burns博士作为CWA的世界领导者，在新的方向上开发了该方法，并成功地将其应用于许多不同的领域。 这里提出的研究是新颖和创新的，这是第一次将认知工作分析应用于生成人工智能对操作员的解释，直接预测信任将如何随着系统透明度而变化，预测人类操作员的工作量将如何变化，并向人类操作员传达意图。 博士Burns建议开发和演示CWA是否可以在AI、自主或算法系统中实现以下目标： O1）提供一个框架，以改进向操作员提供的解释。 O2）预测人类操作员的表现如何随着系统透明度而变化。 O3）提高对系统意图的识别。 这项研究应用于几个不同的背景下，从金融系统到防御系统再到自动驾驶汽车。 在不同背景下的应用增加了工作的能力，对工业和人因工程的方法产生影响。 这项工作很重要，将影响所有使用算法或人工智能来做出决策的先进技术的新应用。 这项工作建立在伯恩斯博士在人因工程方面的世界领先计划的基础上，并提高了加拿大作为高效先进技术领导者的地位。

Autonomous systems, advanced algorithms, and artificial intelligence are emerging technologies. While these technologies aim to reduce human mental workload, if they are not designed with the user in mind these technologies can increase workload and decrease situation awareness. This proposal explores using a method from human factors engineering, Cognitive Work Analysis (CWA), to define the requirements for autonomous, algorithmic or artificially intelligent systems in ways that they assist human operators and improve human decision making. CWA is well-suited for this problem as CWA was developed to help people work better with complex system like the ones proposed here. In particular, CWA takes advantage of a strong reasoning structure and models of human information processing that advance current approaches to human interaction. Dr. Burns, as a world leader in CWA has developed the method in new directions and applied it successfully to many different domains. The research proposed here is novel and innovative, and this is the first time Cognitive Work Analysis will be applied to generate explanations from AI to operators, to directly predict how trust will change with system transparency, to predict how human operator workload will change, and to communicate intention to human operators. Dr. Burns proposes to develop and demonstrate whether CWA can achieve the following in AI, autonomous or algorithmic systems: O1)Provide a framework for improving the explanations provided to operators. O2) Predict how human operator performance changes with system transparency. O3) Improve the recognition of system intent. This research is applied in several different contexts, from financial systems, to defence systems to autonomous vehicles. The application in different contexts increases the ability of the work to have impact to industry and to the methods of human factors engineering. This work is important and will influence all new applications of advanced technology, that use algorithms or AI to make decisions. This work builds on Dr. Burns' world leading program in human factors engineering and advances Canada's position as a leader in highly effective advanced technologies.