Interdisciplinary learning requires students to integrate knowledge and methods from different disciplines, prompting advanced skills development and knowledge construction by connecting diverse subjects. Artificial Intelligence (AI), with its rich information base and abilities to track learning and provide personalized support, has the potential to support interdisciplinary teaching and learning. Previous literature reviews either focus on AI in education in general or Science, Technology, Engineering, and Mathematics (STEM) education, leaving the research trends, challenges, and opportunities of AI for interdisciplinary learning insufficiently examined. To address this gap, this systematic review synthesizes research on AI for interdisciplinary learning published from 2010 to 2023. Following the PRISMA guidelines, we conducted a meticulous analysis of 71 publications. We found that AI for interdisciplinary learning was primarily manifested in four forms: models, robots, systems, and conceptual frameworks, with varied frequencies. We summarized the diverse roles of AI in interdisciplinary learning: it not only enhances interactive, immersive, and personalized learning experiences, but also supports institutional and administrative services. AI has holistic influences on students and plays a significant role in supporting teachers' assessment and professional development. Moreover, the review uncovered the imbalance and limitations of current studies and suggested future research trends. This review provides a thorough understanding of AI for interdisciplinary learning, offering valuable insights for future research, teaching, and technological designs.