襄阳职业技术学院学报

在会计实务工作中，居民个人劳务报酬个税扣缴计算普遍存在基于人工判断的计算机应用情形，工作效率低下。人工智能正由试验性嵌入转向场景化普及应用，财税领域亦不例外。文章提出“政策—模型—应用”三阶框架：首先，梳理居民个人劳务报酬预扣预缴个税相关政策，确保大语言模型精准理解法规；其次，利用DeepSeek大语言模型生成所需的VBA代码；最后，在Excel中设计功能按钮，实现一键调用自动计算。实践表明，该方法高效准确，可为居民个人劳务报酬支付单位提供可复制的轻量化AI应用范式。

In current accounting practice, the calculation of individual income tax(IIT) withholding for resident remuneration for personal services generally relies on computer-aided processing based on manual judgment, leading to low work efficiency. Artificial intelligence(AI) is transitioning from experimental embedded application to scenario-based widespread adoption, and the field of finance and taxation is no exception to this trend. This paper proposes a three-stage "Policy-Model-Application" framework: first, we sort out relevant policies regarding the prepayment and withholding of IIT on resident personal remuneration for personal services to enable large language models to accurately interpret tax regulations; second, we utilize the DeepSeek large language model to generate the required VBA code; finally, we design function buttons in Excel to enable oneclick invocation for automatic calculation. Practical applications show that this method is efficient and accurate,and can provide a replicable lightweight AI application paradigm for entities that pay remuneration for personal services to residents.