AI-mediated feedback in English learning: Moroccan graduate students’ accounts of motivation, autonomy, and trust conditions

This article examines how Moroccan graduate students describe using generative-AI tools (e.g., ChatGPT) as sources of feedback for English learning, with attention to motivation and autonomy as enacted through feedback uptake and trust calibration. Sixty-two graduate students at Sidi Mohamed Ben Abdellah University (Fez) completed an online qualitative survey (Oct 2024-Feb 2025) comprising seven open-ended questions; prompts were delivered in French, and responses were provided in French, Arabic, or English, then translated into English under a meaning-preservation rule and analysed through reflexive thematic coding in MAXQDA 24. Using the Technology Acceptance Model as a heuristic frame, perceived ease of use is linked to rapid access and low-friction interaction, while perceived usefulness is described as condition-dependent across task stakes, access stability, and pragmatic-cultural fit. Participants frame "improvement" through task-situated performance dimensions (accuracy, fluency, controlled complexity, genre/register appropriacy) and through a feedback-use cycle in which tasks are specified, outputs inspected for fit, weaknesses diagnosed, prompts refined, and suggestions then adopted, edited, verified, or rejected. The analysis supports an assessment-for-learning interpretation in which educational value is associated with mediation routines and calibrated trust, rather than automated judgement.