Examining Human Over-Reliance on AI-Enabled Decision Aids

As world events, volatility, and innovation propel us into the industry 5.0 era, businesses must understand how AI-enabled decision tools are used in their workplaces. Anchored to the three-stage human-AI-enabled decision aids interaction process, this study identifies the role of technology-oriented, psychological, and cognitive factors in human over-reliance on AI-enabled decision aids. Designed in two stages, the study’s extensive literature review generated a list of seven key factors (intrinsic AI opacity, modality effect, perfect automation schema, availability AI negativity heuristics, need for cognition, psychological ownership and power autonomy) that potentially influence human over-reliance on AI. This informed the second stage of survey development in empirically testing the relationships in a scenario-based vignette survey experiment. Using Smart PLS-SEM, the significant results yielded new knowledge by showing that the modality effect and perfect automation schema had a significant positive relationship with over-reliance on AI, while the intrinsic AI opacity had a significant negative relationship. The remaining cognitive and psychological factors did not have a significant negative relationship, but the direction tended to show a negative direction. This novel study assists in enhancing the understanding of human over-reliance on AI-enabled decision aids and taking necessary steps to mitigate the undue reliance on AI.