Translating Legal Texts: Human vs. Machine and AI

Abstract

This research aims to investigate the accuracy and acceptability of legal translations produced by Google Translate and ChatGPT compared to standard human translations. Additionally, the study examines the translation strategies used by both Google Translate and GPT and highlights their strengths and weaknesses in rendering legal texts between Arabic and English. Subsequently, three legal Arabic texts along with their standard translations were selected and given to Google Translate and GPT to translate into English. Then their outputs, related to legal terms and expressions, were analyzed and compared to the standard/human translations. Depending on the comparative and descriptive analysis of human translation, Google translation and GPT translation, it is found that: human translation is the most accurate and legally reliable; ensuring precise terminology and structural integrity, ChatGPT performs better than Google Translate; offering more context-aware translations, but it still lacks full legal precision and Google Translate, relying on literal translation, is the least reliable, often misinterpreting legal terms. Furthermore, although Al tools can assist in translation, they should not replace human expertise in legal contexts.