Evaluative Judgement in Teaching AI-based Translation: A Class-room Case Study of AI-Mediated Translation and Post-Editing

In three linesClassroom case study of 23 student projects in machine translation and post-editing. Students compared general-purpose LLMs and online MT systems, evaluated outputs using automatic metrics and human adequacy/fluency assessment, then justified selections. Results: automatic metrics did not determine final choices; students prioritized adequacy, fluency, and post-editing effort over metric rankings.

Summary generated by Claude — human-verified