Human Factors in AI-Enabled Decision Support for Industry 4.0/5.0: A Scoping Review of Technical and Empirical Validation

As AI systems become increasingly embedded in industrial decision-making, ensuring they support rather than overwhelm human operators has emerged as a critical design challenge. Human factors construct such as cognitive workload, trust calibration, and situation awareness are central to this challenge, yet their empirical evaluation remains inconsistent. This scoping review examined 30 peer-reviewed studies to assess how consistently these constructs, alongside explainability and usability, were evaluated across Industry 4.0, and 5.0 contexts. A co-occurrence analysis revealed a structural misalignment: explainability and trust dominated the evaluation landscape, while cognitive workload and usability remained largely unmeasured. Workload and usability showed zero co-occurrence, and 11 papers (36.7%) reported trust without cognitive or performance validation. No study assessed all five constructs simultaneously. Integrated human-in-the-loop evaluation frameworks incorporating validated measures such as NASA-TLX and SUS are needed to ensure AI-enabled industrial systems are interpretable, cognitively safe, and aligned with Industry 5.0 principles.