When Translating from Arabic to English and vice versa, is Google Bard a Trustworthy Tool?

Abstract

The advancement of artificial intelligence (AI) in translation, particularly for language pairs such as English and Arabic, has facilitated cross-cultural communication. However, AI translation tools often struggle with cultural context and domain-specific terminology, underscoring the need for an evaluation of their accuracy (Keen, 2020). This study aims to assess the translation accuracy of Google Bard, a free AI translation tool, by comparing its outputs to human translations across various categories, including historical, literary, media, legal, and scientific texts. The evaluation framework is based on Williams' (2004) four qualitative metrics for translation quality, which examine the morphological, syntactic, semantic, and pragmatic levels of translated texts. By analyzing six texts translated by Google Bard and comparing them with human translations, this research identifies the limitations of AI in handling complex content that typically requires human expertise. While Google Bard is generally adequate for more straightforward texts, its performance diminishes with more technical documents, such as legal and medical reports, scientific studies, and literary works. Consequently, users should be cautious when relying on Google Bard for these genres, as human intervention is essential to ensure accuracy and appropriateness