Addressing Human Factors Related to Artificial Intelligence Integrated Visual Cueing

A variety of assistive extended reality (XR) visual cueing techniques have been explored over the years. Many of which provide significant benefits to tasks such as visual search. However, when the cueing system is erroneous, performance may instead suffer. Factors such as automation bias, where an individual trusts the cueing system despite errors in the cueing may affect task efficacy (i.e. completion time, accuracy, etc.). In some cases, such as with automation bias, these hindrances may be the product of artificial intelligence (AI) integration. Despite this, there may be benefits to using adaptive AI-based cueing systems for XR tasks. However, aspects such as the flow of information, automation accuracy, communication of confidence, or the refusal of output must be considered to build effective AI adaptive cueing systems. In this paper, we discuss four studies conducted by our group that explore visual cueing and AI. We then discuss potential future avenues for integrating AI into cueing techniques to minimize automation bias and cognitive demand on users, as well as, improve overall cueing outcomes.