DeepSeek as Generative IA Tool in Specialized Text Translation and Post-Editing

Abstract

This study explores the impact of Generative artificial intelligence (GAI) in specialized translation processes, using DeepSeek-V3 as the primary tool. The research, with an empirical-exploratory and mixed-method approach, focuses on two objectives: first, to analyze the translation skills and strategies used by students employing DeepSeek to translate texts in legal, medical, and scientific fields; second, to describe the post-editing process, evaluating the techniques used to improve the accuracy and cultural adaptation of the final text. The central research question guiding the study is how DeepSeek influences the development of translation skills and what post-editing strategies students apply to enhance their translations.

The methodology involves translation practices with a group of 30 advanced students from the Bachelor’s Degree in Translation at UABC in Mexicali, who translated and post-edited three specialized texts with DeepSeek. These exercises, conducted in three-hour sessions, also included the creation of specific terminological glossaries. A specialized rubric was used to measure translation quality, and Translog-II software recorded the time spent on each activity, assessing the efficiency and accuracy of the process.

Preliminary results indicate improved efficiency in the translation process and the final product quality through AI tools and post-editing. In the final survey, students also reported a positive perception of DeepSeek, highlighting its utility in developing specific translation competencies.