Abstract: In recent years, China has made great progress in water conservancy informatization and is now gearing up to implement smart water conservancy. However, the specific goals, standards and technical routes of its intelligent application systems are still under exploration. Intelligent engineering safety monitoring is an indispensable component of intelligent water conservancy, and the boundary of its realistic goals is determined by the maturity of AI technology. In light of the status quo of AI technology and based on our experience in dam safety monitoring for several decades, in this study the radical challenges of intelligent engineering safety monitoring were analyzed systematically, and the three cardinal ideas in intelligent scheme formulation and implementation, the realistic goals and roadmap, were put forward definitively. First, we have to decide “what should be done” and “what can be done” from the philosophical points of view, evaluate the reasonability of the scheme from the aspects of technology, economy and safety, and maintain a sincere attitude from the very beginning to the end to get it well done. Second, intelligent engineering safety monitoring should be focused on the principal contradiction between engineering risk and structural resistance, besides the demand and effective supply of information related to engineering safety, and currently its realistic goals should be directed towards identifying the existing structural distress in an explainable way, foreseeing the engineering safety risks and keeping them under control. Third, four key steps, including automated monitoring, full informatization, visualization and modelling, are essential to surpassing conventional informationization and realizing the real intelligentization. With these concepts being well implemented, a solid foundation will be laid for the successful implementation of smart water conservancy.