No French translation of Wang Fuzhi’s Zhouyi Neizhuan (周易内传, “Inner Commentary on the Zhou Changes”) exists. This paper documents an attempt to produce one — undertaken by someone who reads no classical Chinese, holds no expertise in Wang Fuzhi, and possesses no formal training in AI engineering. The attempt was made possible by large language models, primarily DeepSeek, over the course of approximately three months of active work. The paper makes two contributions. First, it offers a documented workflow: how the source text was located and verified, how prompting strategies evolved across twelve rolls, how a content moderation incident was negotiated, and how the subsequent effort to tag the translation for toggleable display of Chinese characters and Pinyin produced only partial success after three distinct technical approaches. Second, and more consequentially, it develops two concepts — friction and delegation — as analytical tools for understanding what AI-assisted scholarship actually looks like from inside the practice. The paper concludes that the translation’s authority is not philological but documentary: it is an artifact of a specific process, at a specific moment, by a specific kind of researcher. That, it argues, is already something worth recording.
François Parlant (Mon,) studied this question.