Fair but impersonal: student responses to AI-led evaluation of reflective writing in translator and interpreter training

In translation and interpreting education, reflection is an important pedagogical device through which students can develop their metacognition and professional judgement. The subjectivity of reflection and openness of a task present several challenges to this work’s automatic assessment. Further, the ability of generative artificial intelligence (AI) to assess students’ competencies beyond merely their technical skills are increasingly being questioned. This study examines students’ perceptions of AI-generated assessments and feedback within the framework of translator education. Accordingly, 43 students from interpreting and subtitling courses within the Minor in Translation programme at Qatar University submitted reflective assignments, each of which was evaluated by ChatGPT (GPT-4.1) against a standardized rubric. Following the evaluation, students completed an online survey measuring their perceptions of the fairness, accuracy, and quality of AI-generated feedback. Results revealed that they considered the AI-based evaluations fair and technologically competent; however, they were less likely to perceive AI-generated feedback as an emotionally involved response or as acknowledging their individual contributions compared with their experiences with human evaluators. These results contribute to the debates on AI’s scope and limitations in translation education and provide insights into AI systems’ capability to recognize the reflective, culturally situated, and voice-based elements inherent in student writing.