مشخصات پژوهش

صفحه نخست /Evaluating Human-AI ...
عنوان Evaluating Human-AI Collaboration in Translation: A Comparative Study on Quality, Synergy, and Context-Specific Performance
نوع پژوهش مقاله ارائه شده کنفرانسی
کلیدواژه‌ها AI Translation Quality Assessment, Comparative Translation Studies, Context-Specific AI Tools, Human-AI Collaboration, Synergistic Translation Workflows
چکیده Abstract: While AI-driven translation tools offer significant benefits, debates persist about their output quality. This study evaluates human, AI-generated, and human-AI collaborative translations through a comprehensive comparative analysis. A corpus of English texts—drawn from comparative legal studies and chemistry textbooks, and collections of short stories—was translated into Persian by six professional translators (MA-level), three AI systems (ChatGPT, Claude, Google Gemini), and three translators collaborating with each AI tool. Translations were assessed by three PhD translation scholars using a rubric and Microsoft Copilot, which analyzed submissions against the same criteria. Results revealed that human-AI collaboration yielded higher quality than standalone AI outputs. Claude-assisted translations achieved the highest scores, followed by ChatGPT and Gemini collaborations. Among AI-only translations, Claude also ranked first, surpassing ChatGPT and Gemini. The findings underscore the superiority of synergistic human-AI workflows, particularly with Claude, while highlighting variability in AI performance across content types. This study contributes empirical insights into optimizing translation practices by integrating human expertise with AI capabilities, advocating for context-specific tool selection to enhance accuracy and reliability.
پژوهشگران رضا خانی (Reza Khany) (نفر اول)، محمدمهدی معادی خواه (نفر دوم)، پوریا برزن (نفر سوم)