Adaptive Game AI improves adaptability of opponent AI with the challenge level of the gameplay; as a result the entertainment of game is augmented. Opponent game AI is usually implemented by scripted rules in video games. However MCT (Monte-Carlo for Trees), a most updated algorithm which perform excellent in computer go can also be used to achieve excellent result to control non-player characters (NPCs) in video games. In this paper, the prey and predator game genre of Pac-Man is used as a test-bed, the basic principle of MCT is presented, and the effectiveness of its application to game AI development is demonstrated. Furthermore, in order to reduce the computation intensiveness of Monte-Carlo, ANN (Artificial Neural Network) is used to produce the intelligence of game opponent with the data collected from Monte-Carlo method. The effectiveness and efficiency of the process is proved.