From post-editing to prompting

Abstract This study compares traditional translation and post-editing with generative-AI-assisted translation in Arabic–English translation education. Drawing on 68 task records completed by 58 undergraduate students through EduApp, it examines observable effort, textual intervention, and implications for translation competence. The findings indicate that AI-assisted tasks involved significantly more written or entered characters and a higher length ratio, but no statistically significant difference in task completion time. Qualitative analysis suggests a shift from local correction to broader reformulation, including more formal institutional phrasing and syntactic restructuring. The study concludes that AI-assisted translation redistributes rather than eliminates student effort and requires evaluative and interactive skills alongside established translation abilities.