Generative AI in C-E Translation of The Roots of Wisdom: A Comparative Study with Human Translations

Abstract

Generative AI offers a new approach to the cross-cultural translation of Chinese classical philosophical texts, yet its performance in English translation of aphoristic works integrating Confucian, Buddhist and Taoist connotations with poetic and philosophical features demands targeted examination. Taking The Roots of Wisdom as the research subject, this study constructs a three-dimensional evaluation framework of fidelity, readability and aesthetic quality. With the generative AI Doubao as the core object and the classic English translations by Zhou Wenbiao and Paul White as references, it systematically analyzes Doubao’s C-E translation performance from three dimensions: cultural element transformation and interpretation, linguistic style and expression, and cultural value transmission, with theoretical interpretation based on literary translation’s defamiliarization theory and cultural translation perspective. The findings reveal that Doubao’s translations have good readability with standardized grammar and natural English expression, but they suffer from simplified cultural images and comprehension deviations in translation. Restricted by the algorithm’s routinization logic, it fails to restore the original’s concise classical charm, weakening literariness. Moreover, it cannot deeply convey the core Confucian, Buddhist and Taoist philosophies like human translators, leading to superficial cultural connotation transmission. This study holds that Doubao can act as an efficient auxiliary translation tool for classical philosophical texts, but its translations require targeted human revision in cultural connotation, linguistic style and value transmission for cross-cultural communication. This study further provides case references and optimization ideas for the practical application and quality improvement of generative AI in C-E translation of classical literary texts.