This study explores the application of the Three-Level Analysis of Language Use (Xu & Liu, 2024) in human-machine collaborative translation of culture-loaded words. By examining the translation of terms such as xietian and sheguang, the study analyzes the effectiveness of the three-level analysis in enhancing translation accuracy and cultural adaptability. The first-level analysis focuses on literal translation, the second-level analysis integrates contextual background for refinement, and the third-level analysis delves into cultural and academic significance to ensure the transmission of deep cultural connotations. Findings indicate that the combination of human translators' in-depth understanding with ChatGPT’s efficiency significantly improves translation quality, particularly for complex terms involving cultural and historical contexts. This approach provides new insights into human-machine collaborative translation and highlights its potential in handling culture-loaded words.