Abstract:  Cast aluminum alloys possess not only the advantage of lightweight and high strength but also excellent integrated formability for complex thin-walled components, which enables their long-term engineering applications in automotive lightweighting, aerospace parts, and large-scale integrated complex structures. In recent years, application scenarios have put forward increasingly comprehensive and multi-dimensional requirements for aluminum alloy materials. It is urgent to improve the strength and plasticity, as well as the service performance under extreme conditions, while also taking into account factors such as heat treatment avoidance, high thermal conductivity, control of impurity tolerance in recycled aluminum, and a stable casting process window. Traditional casting alloy composition design mainly relies on empirical modification of elemental components, and the interactions among multiple alloying elements exhibit obvious non-linear superposition characteristics. Therefore, the traditional empirical trial-and-error method is no longer suitable for the development needs of composition design. The machine learning method based on artificial intelligence technology provides an effective approach to quantify aforementioned complex interactive relationships and transform them into modelable, interpretable, and screenable composition spaces, and gradually applying it. This work provides an overview of the recent advancements in the application of machine learning technology in the design of casting aluminum alloy compositions, covering aspects such as data foundation, feature engineering, model application, and microstructure property regulation. It also anticipates future development directions and pressing issues that need to be addressed.