The Concerns Paradox: An Affective Model of ChatGPT Adoption in ESP Contexts

Purpose – This study examines the psychological determinants of ChatGPT utilisation among English for Specific Purposes (ESP) students in Albanian higher education, applying Self-Determination Theory to identify which factors predict actual adoption behaviour. Design/methodology/approach – A cross-sectional survey was administered to 202 ESP students at the Agricultural University of Tirana and the University of Tirana. The instrument measured Extrinsic Motivation, Competence/Autonomy, Relatedness/Trust, Challenges/Concerns, and Actual Utilisation using 31 Likert-scale items adapted from validated SDT scales. Data were analysed through Multiple Linear Regression and K-Means Cluster Analysis. Findings – Trusting ChatGPT as a safe, non-judgmental partner was the strongest predictor of usage, significantly outweighing digital competence. While extrinsic motivation also drove adoption, awareness of AI risks did not deter use. Cluster analysis revealed a “concerns paradox†: frequent users and hesitant users reported identical levels of worry about AI limitations. Ultimately, affective trust, not risk assessment, separated the adopters from the sceptics. Originality/value – This study provides the first SDT-grounded empirical evidence on ChatGPT adoption in Balkan higher education and documents a “concerns paradox†whereby awareness of AI limitations does not inhibit utilisation. The affective quality of human-AI interaction—not utility alone—is the primary driver of adoption among instrumentally oriented ESP learners.