Abstract: Artificial intelligence involves various fields of human society, and interdisciplinarity is its basic characteristic. Studying the interdisciplinary structure of artificial intelligence and the relationship between subdisciplines is important for the government to formulate the development strategy of artificial intelligence. In this paper, we used the data of AI-related literature resulting from Web of Science between 2017 and 2020, and by way of bibliometrics and social network analysis, analyzed data from multiple perspectives, such as disciplinary networks, disciplinary clusters, disciplinary individuals and disciplinary keywords, to reveal the interdisciplinary characteristics of AI. The study found that: 1) AI-related literature contains 237 disciplines, accounting for 94% of the core collection of Web of Science, reflecting the strong interdisciplinary character of AI. 2) The disciplines of AI are naturally clustered into 15 main disciplinary clusters, and 5 major disciplinary clusters are in a mature, core position in the disciplinary ecology. 3) The five major disciplines of computing and materials have the largest number of AI research results and the greatest influence on AI in other disciplines. Life science such as medicine and pharmacy AI literature has the largest number of citations, and features high-quality AI research or applications. 4) The relationship between AI sub-disciplines based on disciplinary core terms shows that two-thirds of the dichotomous disciplines are relatively close, but overall, there exists a huge difference, with the phenomenon of “those who are close are extremely close and those who are distant are extremely distant”.