生成式人工智能在教育领域的应用正逐步深化，其在中学地理教学中展现出独特的技术优势与实践潜
力。DeepSeek作为该类工具的典型代表，在多项技术维度上表现突出，被视为推动教学智能化转型的重要力量。然
而，由于模型本身在数据质量、学科理解与教学适配性方面存在局限，其生成内容往往难以完全契合课程目标与教
学需求，亟须教师发挥专业判断力，对生成结果进行审慎识别与内容纠偏。本文以DeepSeek为研究对象，结合高中
地理“服务业区位因素及其变化”具体课例，系统剖析其在教学场景中可能出现的典型偏离现象，进而探索适用于
地理学科的人机协同纠偏路径，以期为生成式人工智能的教育应用规范与教师专业发展提供参考。

The application of Generative Artifi cial Intelligence in the fi eld of education is gradually deepening, demonstrating unique technical advantages and practical potential in high school geography teaching.As a typical representative of such tools, DeepSeek has shown outstanding performance across multiple technical dimensions and is regarded as a signifi cant force in promoting the intelligent transformation of teaching. However, due to inherent limitations of the model in terms of data quality, comprehension of subject, and teaching adaptability, its generated content often fails to fully align with curriculum objectives and practical teaching needs.This necessitates teachers to exercise professional judgment in carefully identifying and correcting deviations in the generated results. This paper takes DeepSeek as the research subject and incorporating a specific case study from high school geography“Locational Factors of Service Industries and Their Changes”.Then, it explores the human-machine collaborative correction paths applicable to the geography discipline, with the aim of providing references for the standardization of generative artifi cial intelligence's educational application and the professional development of teachers.