The emergence of artificial intelligence technology has changed translation techniques in the English as a Foreign Language (EFL) learning, particularly through ChatGPT, Google Translate and DeepL. The aim of this study was to find the differences in the translation quality of EFL students when translating texts without AI support and with it, and to find out how far AI can help support the process of academic translation. This study employed qualitative descriptive design with 10 students of the sixth semester in the English Literature Study Program at Universitas Bumigora West Nusa Tenggara. Data was acquired via two translation tasks, one using the traditional way, and the other using AI aid. Larson’s translation quality metrics were then used to assess the data. The results indicated that both strategies yielded relatively satisfactory translations. However, AI assisted translation scored higher in all respects. The best score in category 4 earned an overall score of 78.6% in the traditional way, with an accuracy of 82.9%, clarity of 78.6%, and naturalness of 74.3%. Meanwhile, the AI-assisted translation obtained an overall score of 83.33%, with an accuracy of 87.14%, clarity of 82.86% and naturalness of 80.0%. The results showed that the AI-assisted condition obtained higher observed scores than the conventional condition across accuracy, clarity, and naturalness. However, these differences should be interpreted cautiously because the two tasks were not fully equivalent and the AI-assisted task followed the conventional task. The findings therefore indicate an observed difference in translation quality rather than a definitive causal effect of AI assistance. These findings suggest that AI-assisted translation may serve as a useful learning resource when accompanied by critical evaluation and human post-editing.
Copyrights © 2026