Psychological Impact of Artificial Intelligence on Human Cognition, Behavior, and Emotional Well-Being: A Systematic Review

Abstract

The rapid proliferation of artificial intelligence across diverse contexts has generated increasing scholarly attention to its psychological consequences. Despite growing empirical literature, evidence on AI's cognitive, behavioral, and emotional impacts remains fragmented across disciplinary silos, precluding an integrated understanding. This systematic review synthesized evidence across cognition, behavior, and emotional well-being, identifying mechanisms shaping long-term human functioning. Following PRISMA 2020 guidelines, six databases (PsycINFO, PubMed, Scopus, ERIC, Google Scholar, Web of Science) were searched. Out of the 450 identified records, 57 peer-reviewed studies met inclusion criteria and were coded across 28 cognitive, behavioral, and emotional constructs. Thematic synthesis yielded three major clusters. Cluster A (n = 18): cognitive offloading, critical thinking erosion, and metacognitive atrophy were dominant, moderated by contextual and pedagogical factors. Cluster B (n = 19): AI drove adoption and learning gains but eroded agency, fostered algorithmic dependency, and produced context-sensitive trust calibration. Cluster C (n = 20): therapeutic AI reduced depressive symptoms (small-moderate effects), whereas social media AI amplified anxiety and loneliness, and companionship AI increased isolation over 12 months despite short-term comfort. AI functions as a cognitive enhancer, behavioral modifier, emotional support, and dependency mechanism, with effects moderated by design, context, user literacy, and exposure duration. Findings call for AI literacy, human-centered design, and evidence-based policy to ensure AI strengthens rather than supplants human psychological capacity.