人工智能在數字政府建設中的應用及其法治化路徑，基於DeepSeek在數字行政中應用場景的分析

以DeepSeek為代表的生成式人工智能正成為我國數字政府建設中新的技術變量。各層級政府都將人工智能應用作為數字政府建設的重要切入點。當前政府應用Deep-Seek等主要在內部辦公、智能問答、政策推送和輔助行政管理等場景。區別於企業和個人，政府應用人工智能技術在目標與任務、數據處理、算法應用的要求、技術應用產生的影響等方面具有特殊性，同時也隱含著合法性、政務數據安全、加深偏見與差別化待遇以及技術避責等風險。人工智能在數字政府建設中的應用及其法治化，應當遵循安全優先原則、合法原則、透明原則和責任原則。應當積極對行政系統的內部治理進行調適，提供以“軟法”為先導、兼容“硬法”的法律規則供給、推行監管沙盒策略、採取風險規制的措施並強化對採購和部署的監督。

Generative AI represented by DeepSeek, is becoming a new technical variable in the construction of digital government in China. Governments at all levels regard the application of AI as an important entry point for the construction of digital government. At present, government applications such as DeepSeek are mainly used in scenarios such as internal office, intelligent Q&A, policy push, and auxiliary administrative management. Different from enterprises and individuals, the application of AI technology by the government has particularities in terms of goals and tasks, data processing, requirements for algorithm application, and impact of technology application, as well as risks such as legitimacy, government data security, deepening bias and differential treatment, and technical evasion of responsibility. The application of AI in the construction of digital government and its legalization shall follow the principles of security primacy, legality, transparency, and accountability. The internal governance of the administrative system should be actively adjusted, providing a supply of legal rules that are guided by“soft law”and compatible with“hard law”, implementing regulatory sandbox strategy, adopting risk control measures, and strengthening supervision of procurement and deployment.