AI in Post Editing of News: A Study with Arabic News Articles

The accelerated evolution of artificial intelligence (AI) tools has greatly influenced the practice of translation, specifically challenging their capacity to localize material in a manner that complies with the cultural and contextual demands of the target audiences. The study examines the localization capacity and precision of two of the most widespread AI models—Google Translate and ChatGPT—in translating Arabic news stories into English. The study collected 15 news stories from major Saudi newspapers to act as the baseline dataset. AI­generated translations are compared with professionally created human translations to determine the level of localization achieved. All translations were evaluated by a panel of professional language experts according to predefined localization measures. Their evaluation is quantitatively analyzed via SPSS to determine statistical differences in localization quality. The research seeks to uncover how far AI­based tools can localize culturally unique content, identify challenges for such models, and provide genre­based differences in performance. The results provide insights into the limitations of AI in localizing news content today and the potential for future expansion. Furthermore, the results revealed stark differences in the quality of localization between AI and human translations, identifying key challenges and noting differences in the effectiveness of localization across various news types. ChatGPT emerged as the most accurate AI model for localizing the news data. This study adds to the body of literature on emphasizing the importance of post­editing and AI­aided translation in the case of Arabic­English language pairs.