The double-edged sword ai as a tool for mental well-being enhancement or addiction?

Abstract

This study explores how AI tools like ChatGPT are becoming a big help for learning, productivity, and emotional support, but it also looks at the concerns about overuse and AI addiction. The research found that AI can be a double-edged sword - it can enhance resilience and well-being, but it might also lead to compulsive behaviors. Using ideas from Compensatory Internet Use Theory (CIU) and Flow Theory, and examined data from 343 people in Vietnam who regularly use AI platforms. The results show that depression often leads people to use AI for entertainment or stress relief, which then helps them get lost in the flow and form emotional bonds with the technology. While flow can improve mental health, it also increases the risk of AI addiction. Emotional attachment appears to facilitate mental well-being; however, it does not necessarily result in addiction. These insights reveal a sort of paradox: engaging with AI for comfort might unintentionally cause overreliance. Overall, this research helps us better understand how people interact with AI, shedding light on psychological aspects, and offers useful guidance for developers to create emotionally smart and ethically sound AI platforms. Research purpose: This study aims to investigate the relationship between mental health conditions and AI addiction by examining how situational vulnerabilities such as perceived stress, depression, and fear of social judgment influence users’ mental well-being and susceptibility to AI addiction, and how psychological mechanisms including entertainment-seeking, pressure relief, flow experience, and emotional attachment mediate this relationship. The study endeavors to provide useful implications for developers and policymakers to design emotionally intelligent and ethically responsible AI platforms that promote users’ mental well-being. Research motivation: The increasing prevalence of mental health issues and the rapid adoption of AI technologies have created a paradoxical situation: while AI can enhance psychological well-being, it may also foster AI addictive behaviors that exacerbate mental health problems. However, existing literature remains limited in examining the relationship between psychological vulnerabilities, mental well-being, and AI addiction. In particular, few studies have explored the underlying psychological mechanisms, such as entertainment, pressure relief, flow experience, and emotional attachment as potential mediators that explain how positive mental health conditions influence users’ AI addiction behavior. Research design, approach, and method: This study adopted a quantitative research design using a structured questionnaire comprising two sections: the first section collected respondents’ demographic information, while the second section measured the relationships among the study variables through 33 items across 9 variables, rated on a 5-point Likert scale. Data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) via SmartPLS software. The data analysis process included assessing internal consistency reliability, convergent and discriminant validity, multicollinearity, and evaluating model fit through R2, effect size (f2), and predictive relevance (Q2). Main findings: The findings reveal that depression positively influences entertainment and releasing pressure while perceived stress and fear of social judgment show no significant relationship with these compensatory motivations. Moreover, both compensatory motivations have an impact on flow experience, which subsequently enhances emotional attachment to AI. Emotional attachment, interestingly, contributes positively to psychological well-being without directly leading to addictive behaviors. The study also identifies a positive correlation between mental well-being and AI addiction, suggesting that a positive mental state could lead to the overuse and dependence on AI of individuals with existing psychological vulnerabilities. Practical/managerial implications: The study highlights AI’s potential in promoting emotional well-being and supporting mental health interventions. Developers are encouraged to integrate personalization features that adapt to users’ emotional states and incorporate flow-enhancing elements to foster positive engagement. Furthermore, safeguards like usage limits and reflective prompts should be implemented to prevent overuse and overreliance. Collaboration with mental health professionals remains essential to ensure ethical, clinically informed design and enhance public trust in AI-based emotional support systems.