Beyond ChatGPT: The Psychological Implications of Generative Artificial Intelligence for University Students. A Critical Review

Abstract

The rapid diffusion of generative artificial intelligence (GenAI) tools such as ChatGPT, Gemini, Claude, and Copilot has transformed how university students learn, write, seek help, and manage academic demands. While early scholarship concentrated on academic integrity and assessment redesign, the psychological consequences of sustained GenAI use remain comparatively underexamined. This critical narrative review synthesises empirical and conceptual literature published primarily between 2021 and 2026, identified through systematic searches of Scopus, Web of Science, ERIC, PsycINFO, and Google Scholar, to map what is currently known about the psychological implications of GenAI for university students. Four interrelated thematic domains are identified: (1) cognitive functioning, including cognitive offloading, critical thinking, and metacognitive engagement; (2) motivation and the self, including academic self-efficacy, autonomy, and competence; (3) emotional wellbeing, including AI anxiety, technostress, stress relief, loneliness, and chatbot companionship; and (4) academic integrity, moral emotions, and academic identity. The evidence portrays GenAI as a double-edged psychological technology. It can scaffold confidence, reduce help-seeking anxiety, and provide accessible emotional support, yet it can also foster overreliance, erode critical thinking, intensify anxiety about future employability, and destabilise students' sense of authorship and academic self-worth. A dual-pathway conceptual model is proposed in which AI literacy, self-regulation, social support, and institutional policy moderate whether GenAI use follows an enabling or an eroding trajectory. The review concludes that psychological outcomes depend less on the frequency of GenAI exposure than on the mode of engagement. Implications for university policy, counselling services, curriculum design, and assessment reform are discussed, and an agenda for longitudinal, experimental, and cross-cultural research is outlined.