Connecting self-esteem to problematic AI chatbot use: the multiple mediating roles of positive and negative psychological states.

Abstract

The emergence of AI chatbot products has ushered in a new era of human-AI interaction, yet scholars and practitioners have expressed concerns about their use due to potential addictive and adverse effects. Currently, the understanding of problematic AI chatbot use (PACU) remains incomplete and inconclusive. Despite previous findings that indicate negative outcomes associated with the use of AI products, limited studies have explored the underlying factors that drive the complex process leading to the formation of PACU. Furthermore, while existing literature highlights how personal traits influences problematic IT use via evoked psychological states, it largely overlooks that the positive psychological experience may also have a potential influence on problematic outcomes. Incorporating flow experience into the compensatory internet use theory, this study presents a multiple mediation model to investigate how social anxiety, escapism, and AI chatbot flow influence the relationship between self-esteem and PACU. We examine the model using Partial Least Squares Structural Equation Modeling (PLS-SEM) with cross-sectional data collected from 563 online users who have engaged with AI chatbots. Our findings indicate that users with low self-esteem are more likely to conduct problematic behavior when using AI chatbots. This relationship can be mediated by social anxiety, escapism and AI chatbot flow. This study sheds light on how self-esteem negatively affects PACU, unraveling the underlying psychological processes experienced by users with low self-esteem in their interactions with AI chatbots. Also, we provide practical insights for online users and practitioners to mitigate the potential negative impacts of AI product usage.