Abstract:
Automation provides the opportunity for many tasks to be done more effectively and with greater safety. However, these benefits are unlikely to be attained if an automated system is designed without the human user in mind. Many characteristics of the human and automation, such as trust and reliability, have been rigorously examined in the literature in an attempt to move towards a comprehensive understanding of the interaction between human and machine. However, workload has primarily been examined solely as an outcome variable, rather than as a predictor of compliance, reliance, and performance. This study was designed to gain a deeper understanding of whether workload experienced by human operators influences compliance with and reliance on an automated warehouse management system, as well to assess whether age-related differences exist in this interaction.
As workload increased, performance on the Receiving Packages task decreased among younger and older adults. Although younger adults also experienced a negative effect of workload on Dispatching Trucks performance, older adults did not demonstrate a significant effect. The compliance data showed that as workload increased, younger adults complied with the automation to a greater degree, and this was true regardless of whether the automation was correct or incorrect. Older adults did not demonstrate a reliable effect of workload on compliance behavior. Regarding reliance behavior, as workload increased, reliance on the automation increased, but this effect was only observed among older adults. Again, this was true regardless of whether the automation as correct or incorrect. The finding that individuals may be more likely to comply with or rely on faulty automation if they are in high workload state compared to a low workload state suggests that an operator's ability to detect automation errors may be compromised in high workload situations.
Overall, younger adults outperformed older adults on the task. Additionally, older adults complied with the system more than younger adults when the system erred, which may have contributed to their poorer performance. When older adults verified the instructions given by the automation, they spent longer doing so than younger adults, suggesting that older adults may experience a greater cost of verification. Further, older adults reported higher workload and greater trust in the system than younger adults, but both age groups perceived the reliability of the system quite accurately.
Understanding how workload and age influence automation use has implications for the way in which individuals are trained to interact with complex systems, as well as the situations in which automation implementation is determined to be appropriate.