Human-Human v.s. Human-AI Collaboration: The Double-Edged Sword Effect of Moral Disengagement

The widespread use of artificial intelligence (AI) is changing teamwork in organizations, shifting collaboration from human–human to human–AI arrangements. Although task failure is common in human–AI collaboration, its effects on employees’ psychological states and behaviors remain underexplored. Drawing on social cognitive theory, this study examines the double-edged effects of moral disengagement following task failure in human–AI collaboration. Across three scenario-based experiments and one field survey, we find that task failure in human–AI collaboration is more likely to trigger employees’ moral disengagement than failure in human–human collaboration. Moral disengagement reduces employees’ negative affect and enhances work-related well-being, with leaders’ emotional support strengthening this effect. However, moral disengagement also suppresses employees’ job reflective learning and lowers task performance, with leaders’ instrumental support weakening this effect. By focusing on task failure in human–AI collaboration as a critical context, this study reveals the double-edged nature of moral disengagement and offers insights into leadership interventions following failure.