AI Insight
This study examined how different forms of social interaction relate to loneliness among university students in AI-mediated environments, using social capital as a mediating factor. Results showed that online social interaction preferences directly increased loneliness and provided only weak indirect benefits through social capital accumulation, while offline embodied social competence strongly reduced loneliness by building social capital. The research reveals that social capital gained through offline interactions offers substantially greater protection against loneliness compared to that derived from online AI-mediated social environments.
Why it matters
These findings suggest that universities and mental health professionals should prioritize interventions that strengthen face-to-face social skills rather than relying primarily on digital social platforms to address student loneliness. The study provides evidence-based guidance for designing campus programs and AI-based social technologies that better support student wellbeing in increasingly digital social environments.
Understand the Science
Based on Social Capital theory, this study explores the mechanisms linking emerging forms of social interaction and individual Loneliness among university students in the age of artificial intelligence (AI), from the perspective of Social Capital accumulation within AI-mediated social environments. Using survey data collected from university students, a Partial Least Squares Structural Equation Modeling (PLS-SEM) approach was employed to analyze the relationships among key variables. Specifically, social interaction preferences (including behavioral preferences and image presentation preferences) and embodied social competence were treated as independent variables. Social Capital was modeled as a higher-order latent construct comprising Bonding, Bridging, and Linking Social Capital, which jointly mediated the associations between social interaction characteristics and Loneliness.; and Loneliness was specified as the dependent variable. The results reveal differentiated mediating mechanisms within the AI-driven social interaction system. Online social interaction preference exhibits a dual-path effect on Loneliness: it directly increases Loneliness, while also exerting an indirect mitigating effect through Social Capital; however, this indirect effect is notably weak, suggesting limited psychological compensation. In contrast, embodied social competence emerges as the strongest predictor of Social Capital and significantly reduces Loneliness through higher levels of Social Capital. Furthermore, different sources of Social Capital demonstrate distinct effects in the context of AI-mediated social environments. The pathway linking online social interaction preferences to Social Capital was comparatively weaker, whereas the pathway from Offline Social Competence to Social Capital was substantially stronger. These findings suggest that different pathways of Social Capital accumulation may be associated with varying levels of protection against Loneliness. Overall, this study identifies differentiated associations among social interaction characteristics, Social Capital, and Loneliness, extends the application of Social Capital theory in AI-enabled socio-technical contexts, and provides important implications for promoting university students’ social development and mental wellbeing.