近年来，以DeepSeek为代表的大语言模型，凭借算法、算力与成本优势，深度介入教育领域，重塑了教育生态。人工智能的“一体两面”性决定了其赋能教育实践时，既展示出解放劳动力、释放创造力，推动教育平权的潜能，又暴露了学习者智能降维、教育工作者职业胜任力下降等隐忧，由此引发了新一轮的“智能悖论”争鸣。回应争鸣，解决问题的关键在于厘清人工智能与人的智能“和而不同”的关系，构建人机协同的理论框架与实践路径，推动两种智能形态和谐共生，服务于数智教育的高质量发展，真正实现从“学会”到“会学”、从“知识”到“智慧”的升级跃迁。

In recent years, large language models represented by DeepSeek have deeply penetrated the education sector, leveraging their advantages in algorithms, computing power, and cost, thereby reshaping the educational ecosystem. The "one body, two faces" nature of artificial intelligence determines that when it empowers educational practice, it not only demonstrates the potential to liberate labor, unleash creativity, and promote educational equality, but also exposes hidden concerns such as the dimensionality reduction of learners' intelligence and the decline in educators' professional competence, thus triggering a new round of contention over the "intelligence paradox". To respond to this contention and solve the problem, the key lies in clarifying the "harmonious yet distinct" relationship between artificial intelligence and human intelligence, constructing a theoretical framework and practical path for human-machine collaboration, promoting the harmonious coexistence of the two forms of intelligence, serving the high-quality development of digital and intelligent education, and truly achieving the upgrade transition from "learning" to "learning how to learn" and from "knowledge" to "wisdom".