Multidisciplinary Research on Social Media, Artificial Intelligence and Mental Well-Being

Abstract

Social media and artificial intelligence increasingly occupy the same space in everyday emotional life. People encounter algorithmically selected content, discuss personal difficulties online, and may seek support from conversational systems. Yet these activities are often examined in separate academic fields. This paper develops a multidisciplinary critical review connecting social media research, psychological measurement, computational inference, and digital mental health interventions. A purposive selection of verified publications from 2020 to 2026 provides the evidence base. No primary survey, clinical experiment, or participant-level dataset is reported. The synthesis distinguishes three roles of artificial intelligence: organizing digital exposure, inferring possible mental health signals, and providing interactive support. Evidence on social media emphasizes variation between individuals and the limits of time-based measures. Research on conversational agents suggests potential benefits for selected symptoms, while broader well-being, sustained effectiveness, and safety require separate evaluation. These findings motivate an original conceptual framework linking digital experiences with social comparison, perceived support, emotional regulation, and personal agency. Four propositions specify relationships for future testing. The paper argues that a technically accurate model cannot establish the psychological value of an intervention without valid outcomes, appropriate comparison conditions, and attention to consequences. It proposes collaborative research practices for psychology, computer science, communication studies, education, and public health. The contribution is a structured explanation of how these disciplines can investigate the same digital experience without treating prediction, engagement, and well-being as interchangeable outcomes.