ChatGPT for Arabic-English Translation: Evaluating the Accuracy

Abstract

Abstract Using ChatGPT as a translator between several language pairings, including English and Arabic, has made it an invaluable resource for cross-cultural communication. However, due to the limitations, the translation outputs should be evaluated. The main aim of this article is to evaluate the translation accuracy of the ChatGPT 3.5 model. The Arabic and English texts are translated by ChatGPT, and the outputs are compared with the professional translators translations. The texts are divided into genres such as historical, literary, media, legal, and scientific. ChatGPT outputs are evaluated regarding language accuracy and meaning in the given context. The results showed that ChatGPT can be used as a translator for simple content, but human intervention is still required, especially for complex texts that ChatGPT is unable to correctly interpret. The paper also offers some recommendations for the translators and students who use ChatGPT.