This paper is aimed to construct an intelligent teaching system supported by data mining, clustering, personalized recommendation and other technologies based on an in-depth study of the theoretical knowledge of new liberal arts teaching and the technology in the field of artificial intelligence. As a result, an intelligent interactive teaching mode is proposed. And the research results can be obtained by investigating the influence of intelligent interactive teaching modes on English listening, reading, and so on. The data of students’ online learning behaviors were characterized and clustered into 3 classes using the K-means method. More than 50% of the students think that the system’s exercise recommendation function can basically meet their needs. Before and after the intelligent interactive teaching, students’ language perception, language expression, language comprehension, information acquisition, affective attitudes and reading ability were significantly improved (Sig. = 0.001). After the teaching, the listening and speaking skills were all over level 6.