In view of the development needs of oil and gas reservoir fracturing stimulation technology, the research status of artificial intelligence (AI) in fracturing stimulation is described, the key theoretical problems in the development of AI in fracturing are analyzed, and the future main research directions and application scenarios are prospected. At home and abroad, some progress has been made in fracturing design optimization, fracturing condition diagnosis and risk pre-warning,fracturing flowback optimization control. But it is still in the transition stage from academic research to industrial application, facing with the key theoretical problems including small sample size and lack of label data, poor model interpretability and the requirement of deep integration of datadriven and physical models. Based on the existing problems, the main research directions of artificial intelligence for fracturing in the future are prospected in this paper, such as data management and feature engineering, deep mining of fracturing data in small sample learning scenarios, interpretable fracturing intelligent algorithms based on knowledge embedding and knowledge discovery, and dynamic optimization of fracturing parameters and risk pre-warning and control methods based on reinforcement learning, etc. Three application scenarios are suggested, including intelligent optimization of fracturing design, closed-loop control of fracturing construction, and intelligent control of fracturing flowback, aiming for the high-quality balanced fracture formation and safe fracturing.