The application of Artificial Intelligence (AI) is a wicked problem facing the government at present, and there is a certain contradiction between the technical rationality of artificial intelligence and the value rationality of administrative behavior. Reasonable utilization of AI will bring benefits to the government, while misbehaving AI application will cause risks. In order to explore the possible risks of AI application in government: 1) this paper retrieves 74 highly relevant articles in a systematic literature review from China National Knowledge Infrastructure (CNKI), 2) by analyzing the contents of the 74 articles, five major categories of risks that may arise from the application of AI to the government are formed and (3) we propose a future research agenda and limitations in the discussion section and then conclude the paper key ideas. We find that existing literature is mostly theoretical and forward-looking and lacks empirical research. The technical, political, administrative, legal and organizational risks derived from the literature review are macro view in nature, and the next experiment should be directed to specific case studies to improve the persuasiveness of the research.

artificial intelligence, AI, risks, literature review