When AI enables and undermines: dual mechanisms linking AI usage to task performance

<title>Abstract</title> Purpose. The growing adoption of AI is reshaping organizational workflows and employee experiences, delivering benefits while presenting novel challenges, reflecting a double-edged sword effect. Although prior studies have identified potential impacts of AI usage on employees, the underlying mechanisms remain insufficiently examined. Guided by self-determination theory, this study investigates the dual pathways—facilitative and inhibitive—by which AI usage impacts employee task performance via motivational and behavioral mechanisms. It further investigates the boundary role of core self-evaluations (CSE) in moderating these effects. Methodology. A three-wave, multi-source field study was conducted involving 409 employee-supervisor pairs from AI-intensive industries in China. Data were collected using leader–employee matched questionnaires. Key constructs—AI usage, motivation types, job crafting, task performance, and CSE—were measured using validated scales. Hypotheses were tested via hierarchical regression, bootstrapping, and moderated mediation analyses using SPSS and Mplus. Findings. Results revealed a dual-chain mediation mechanism: AI usage enhances task performance via autonomous motivation and promotion-focused job crafting, but simultaneously impairs it through controlled motivation and prevention-focused job crafting. Furthermore, CSE significantly moderates both pathways, amplifying the positive and buffering the negative effects. Originality. This study provides understanding of AI usage’s “double-edged sword” effect by identifying parallel motivational-behavioral pathways and the boundary condition of core self-evaluations. The findings enrich self-determination theory in the digital context and offer actionable insights for designing inclusive and personalized AI integration strategies.