Research on the Auxiliary Effectiveness of Translation Software Driven by Artificial Intelligence Large Language Models in the Rapid Comprehension of Cross-Lingual Academic Papers

在全球学术交流需求激增的背景下，本研究基于认知负荷理论与信息处理模型，运用准实验设计和用户问卷，实证分析人工智能大模型驱动的翻译软件在跨语言学术论文快速理解中的辅助效能。研究表明，该类软件通过自动化语义处理和逻辑图谱构建，显著提升文献筛选与表层信息提取效率，在术语解析和句式转化中有效减轻认知负荷。然而，其在处理图表、公式、算法伪代码等非文本元素的语义对齐，以及跨学科知识关联和长段落逻辑连贯性还原方面存在不足，易导致用户对文献深层方法论和创新逻辑的理解缺失，还可能引发过度依赖、削弱批判性思维。研究明确其作为“表层信息加速处理工具”的定位，指出在低阶认知任务中效能显著，但高阶逻辑整合仍需人工干预。Against the backdrop of growing global academic exchanges, this study empirically explores the auxiliary effectiveness of AI large model-driven translation software in cross-lingual academic paper comprehension. Using quasi-experimental design and user surveys, grounded in cognitive load theory and information processing models, it finds that the software boosts literature screening and surface-level information extraction efficiency via automated semantic processing and logical graph construction, reducing cognitive load in term parsing and sentence transformation. However, it struggles with semantic alignment of non-text elements (charts, formulas, pseudocode), cross-disciplinary knowledge linking, and long passage logic restoration, causing gaps in understanding deep methodologies and innovation logics. Over-reliance may also undermine critical thinking. The study positions it as a “surface-level information accelerator,” effective for low-level tasks but requiring human input for high-level logical integration.