<![CDATA[The integration of Artificial Intelligence (AI) into educational environments has intensified scholarly interest in its cognitive and behavioral implications for learners. At the same time, the term “AI psychosis” has appeared in non-academic discourse to describe perceived psychological effects of AI use. However, this terminology is not recognized in established clinical classification systems, including the DSM-5-TR and ICD-11, and its use in education more accurately reflects misinterpretation in human–AI interaction rather than clinical pathology. This study presents a narrative review at the intersection of artificial intelligence in education, cognitive psychology, and human–AI interaction. Drawing on literature related to AI literacy, anthropomorphism, cognitive bias, and learner behavior, the review synthesizes key mechanisms shaping how learners perceive and engage with AI systems. A structured thematic approach was used to organize evidence from prior empirical and conceptual studies in educational technology and cognitive science. The synthesis identifies recurring mechanisms, including anthropomorphism in AI perception, authority bias toward algorithmic outputs, confirmation bias in AI-assisted inquiry, and cognitive offloading that may reduce independent critical evaluation. These mechanisms are especially pronounced when AI literacy is limited, influencing learner trust, dependency, and evaluative judgment. Based on the synthesis, the AI Misinterpretation Model (AIMM) is proposed as a conceptual framework organizing these mechanisms into three layers: perception, interaction, and integration. The review emphasizes strengthening AI literacy to promote critical engagement with AI-generated outputs and reframes “AI psychosis” as a misinterpretation phenomenon rather than a clinical construct.]]>
