Abstract This paper proposes a simulation of teaching behavior based on metacellular automata in response to the problems in the teaching reform of college courses on the digital platform. Using the rule grid and binary method, the tuple units in the structure of tuple automata are described and represented, followed by the standard combination of many tuples to form a tuple space. Considering that the definition of neighbors is relatively complex, von Neumann-type, Moore-type, and extended Moore-type represent the neighbors of a two-dimensional tuple automaton, and rules for the evolution of tuple automata are also formulated. Led by the constructivist teaching model, the contextual teaching method is divided into five links, on which the contextual teaching model based on metacellular automata is constructed. Simulation and Analysis are used to analyze the teaching reform of college courses in the era of artificial intelligence, for example. The results show that when the simulation condition is the initial satisfaction rate of the system ξ = 0.425, and the teaching strategy n=2, the teaching quality improves with the increase of the response probability, which is in line with the law of teaching, i.e.The teacher should motivate the enthusiasm of learning, stimulate the interest of learning, and entirely play the role of students in teaching. This study aims to promote the reform of college course teaching in the era of artificial intelligence, which will contribute to the improvement of college teaching quality.