Poetry is the Last Resort of Translation in the Age of Artificial Intelligence

This study investigates the potential and limitations of artificial intelligence (AI) in translating poetry from English to Arabic compared to human translations. Focusing on Khalil Gibran’s Sand and Foam, the research compares a human-translated version by Antonious Basheer with an AI-generated version produced by ChatGPT-4. Using Multidimensional Quality Metrics (MQM) and qualitative assessments, the study analyzes accuracy and fluency, reflecting the poetic fidelity of both translations. Findings reveal that while ChatGPT-4 excels in grammatical consistency and structural clarity, it often lacks emotional depth, stylistic features, and cultural resonance in human translation. Conversely, although more expressive, the human version contains notable inconsistencies and errors. The results highlight both AI's promise and current limitations in literary reproduction, advocating for a hybrid approach that combines AI efficiency with human creativity. Ultimately, this research contributes to a deeper understanding of poetic translation and offers pathways to enhance AI’s role in preserving literary artistry.