Too much, too little, or just right? The inverted-U relationship between human-AI collaboration modes and eudaimonic well-being

As human-AI collaboration becomes more pervasive, optimizing the distribution of decision-making authority between humans and AI has become a fundamental concern in human-centered AI design. The impacts of different human-AI collaboration modes (human-dominant, human-AI collaboration, and AI-dominant) on users’ eudaimonic well-being, as well as the underlying psychological transmission mechanisms, remain poorly understood. Grounded in self-determination theory, we propose a serial mediation framework linking human-AI collaboration modes to eudaimonic well-being via the sequential mechanisms of AI dependence and perceived autonomy. This framework is rigorously investigated through two scenario-based experiments in the domains of autonomous driving and online shopping, with a particular focus on the differential effects of collaboration modes and the underlying psychological transmission chain. The results reveal that human-AI collaboration mode exhibits an inverted U-shaped relationship with eudaimonic well-being; the level of individuals’ eudaimonic well-being under the human-AI collaboration mode is significantly higher than that under the human-dominant and AI-dominant modes. Additionally, AI dependence and perceived autonomy play a significant serial mediating role in the above relationship. This study constructs and validates the psychological transmission mechanism of how human-AI collaboration affects deep-level well-being, expands the application scope of self-determination theory in human-AI collaboration scenarios, and provides optimization guidelines for AI product design and user experience management. These findings contribute to the design of human-centered AI systems that simultaneously promote technological advancement and human well-being.