AI-Human Collaboration: Augmentation vs. Verification in AI-Assisted Performance Feedback

Drawing on Cognitive Appraisal Theory, this study examines how technology skepticism shapes responses to different human–AI configurations in performance feedback. Using a between-subjects experiment, we compare human-only, AI-only, augmentation (human + AI), and verification (AI + human) source attributions for identical review content. As technology skepticism increases, perceived feedback quality declines across conditions, but the decline is steepest when feedback is attributed to AI alone. In contrast, augmentation largely buffers this negative relationship, suggesting that AI is more acceptable when framed as supporting a human manager who retains authorship and accountability. Verification falls between these extremes, indicating that post-hoc human review provides only partial mitigation. Overall, the findings suggest that employees respond not only to whether AI is involved, but to how human and AI roles are structured in the evaluative process.