Navigating algorithmic anxiety and ethical ambiguity: psychological adaptation and self-regulation in generative AI-enhanced higher education

Purpose – To investigate how university students navigate AI-induced anxiety and ethical uncertainty while attempting to maintain learner agency and self-regulated learning in Generative AI-enhanced educational environments. Design/methodology/approach – An explanatory concurrent mixed-methods design was utilized. Partial Least Squares Structural Equation Modeling (PLS-SEM) analyzed quantitative data from a sample of humanities and technological education students ($N = 342$). Concurrently, codebook thematic analysis (κ = 0.84) was applied to qualitative reflexive diaries. Findings – The structural model suggests that perceived ethical transparency is associated with reduced AI anxiety. Furthermore, the satisfaction of basic psychological needs appears to function as a buffering mechanism against algorithmic technostress. Autonomy satisfaction moderates the relationship between AI anxiety and psychological adaptation, while competence satisfaction attenuates the negative association between AI anxiety and self-regulated learning. Practical implications – The study provides an empirical basis for the development of institutional ethical AI guidelines and the implementation of psycho-pedagogical adaptation protocols aimed at fostering constructive digital literacy in higher education. Originality/value – This research contributes to the interdisciplinary field of educational technology and learning sciences by formally integrating Self-Determination Theory (SDT) with cognitive and ethical extensions of the informal Unified Theory of Acceptance and Use of Technology (UTAUT3), offering a conceptual model that explains cognitive-behavioral interactions with algorithms under conditions of ethical ambiguity.