HUMAN REVIEW PRIORITIES IN AI-ASSISTED TECHNICAL TRANSLATION: AN EXPLORA TORY PERSIAN–AZERBAIJANI CASE STUDY

This article examines how human review can be focused on terminology and meaning in AI-generated technical translations. It combines recent post-editing research with an exploratory reading of fifteen purpose-built Persian software instructions and their provisional Azerbaijani translations. The bilingual pairs were prepared with AI assistance; neither independent human annotation nor a controlled revision experiment has been completed. Selected examples indicate potential problems involving technical specificity, interface wording, and target-language conventions, rather than a measured error rate. A three-level review procedure is proposed to distinguish verifiable inaccuracies, terminology risks, and stylistic preferences. The contribution is a transparent and manageable protocol for subsequent human-guided revision in this language pair, not evidence that an AI system improves after feedback.