Return to main page.
Approaches and Practices
| Overview | Cloud-Based Translation Seminars 雲譯場 | Regulated Use of AI as Translation Aids | Summer Institute |
Regulated Use of AI as Translation Aids
The Baiqian Project Committee acknowledge that AI assistance can greatly accelerate the translation process—making large projects, such as 200,000-word texts, more manageable. However, should a translator decide to employ artificial intelligence (AI) models to assist in their translation process, they should obtain the Translation Initiative Committee’s consent and adhere to the Initiative’s regulated guidelines set as follows (subject to change):
The translator is required to submit all AI-generated drafts—including various versions produced by different models—simultaneously with the draft translation, final translation, rectified drafts, and other relevant work products at each stage. Additionally, for a segment of no less than 1,000 words from the original Chinese text, the translator shall perform a comparative analysis between the AI-generated draft and the version submitted to the Translation Initiative Committee. This analysis should include a detailed explanation of the methodology used to incorporate AI, as well as any modifications and optimizations made to enhance the translation quality.
While AI has shown increasing effectiveness with modern Chinese 白話文, its reliability remains limited for Classical Chinese 文言文 and other technically complex materials. This brings us to a major academic issue that this translation project aims to address: how the Humanities can respond thoughtfully and critically to the rapid advances of AI, and how the future of Digital Humanities (DH) might be reshaped in response. Rather than treating AI merely as a tool for efficiency or as a threat to traditional scholarship, the Baiqian Initiative offers an opportunity to reconsider the evolving roles of human expertise, interpretation, and scholarly judgment—especially in translating classical texts and technical materials, where nuances of meaning, context, and intellectual rigor remain dependent on human insight and cannot be hastily replaced nor fully automated in the foreseeable future.
