Artificial Intelligence in Translating Literary : An Accuracy and Satirical Representation of the Arabic Short Story

The advancement of artificial intelligence (AI) technology has revolutionized translation practices, including the translation of literary texts rich in cultural nuances and stylistic features. This research examines the accuracy and representation of satire in the machine translation of the Arabic short story Juha and the Wooden Donkey by Shaqi Hasan using ChatGPT. The objectives of the research are first, to assess translation accuracy based on Nababan’s evaluation theory and second, to identify the types of satire (Horatian, Juvenalian, and Menippean) in the source text and their preservation in the translation. Using a qualitative descriptive approach, data were collected through reading and note-taking techniques on both the source text and its translation, then analyzed using the Miles, Huberman, and Saldana model. Findings reveal that out of 26 excerpts, 22 (88%) were accurate, 2 fairly accurate, and 2 inaccurate. Ten excerpts were identified as Horatian satire, seven as Juvenalian, and eight as Menippean. These results show that while AI can produce structurally and semantically accurate translations, it still struggles with irony, implicit meaning, and cultural depth. This study emphasizes the human translator’s role in preserving meaning and aesthetics and urges critical evaluation of AI in literary translation.