Low-temperature plasma (LTP) is a form of plasma that is widely used in many industrial fields. Numerical simulation is an important means of studying and analyzing LTP. In recent years, with the advancement of artificial intelligence (AI) technology, AI-driven numerical simulation methods have gradually been applied in the field of LTP, which is ex-pected to overcome the shortcomings of traditional numerical simulation methods. This paper focuses on LTP and first introduces mainstream LTP simulation models, including kinetic models, fluid models, chemical dynamics models, and hybrid models. It analyzes the problems faced by traditional LTP numerical simulation methods from four aspects: model complexity, numerical computation, parameter consistency, and result reliability. Then, using data-driven methods, physics-informed data-driven methods, and numerical simulation acceleration strategies as classification criteria, the current research status of AI-driven LTP numerical simulation is introduced and analyzed in detail. Finally, from the perspective of convergence and generalization, this paper summarizes the challenges faced by relevant research and proposes further development directions.