多數電腦遊戲中電腦控制的對手的行為模式，通常只是在數個固定的策略中作切換，因此當玩家玩了一段時間之後，便能察覺其行為模式的重覆性，於是感覺失去挑戰性而覺得無趣。因此最近幾年來，遊戲開發業者已開始注意到提供一個具有挑戰經驗的遊戲才能具有市場的競爭力。對大部分的遊戲而言，所謂具有挑戰經驗的遊戲，就是指具有高品質的遊戲人工智慧(Game AI)。而所謂的遊戲人工智慧，是指電腦控制的對手的所有決策程序。若在遊戲過程中能自動分析玩家的行為，而修正缺點及更改策略，則稱為自適應遊戲人工智慧(adaptive game AI)。現階段自適應遊戲人工智慧的研究尚屬萌芽階段，本計畫將以三年時間完整地開發在不同遊戲階段的自適應遊戲人工智慧技術。首先我們將結合離線自適應遊戲人工智慧與玩家建模技術，自動產生多組適合不同程度玩家之策略。並研發線上自適應遊戲人工智慧技術，搭配玩家建模，即時地根據玩家行為變換或改進策略。更將導入動態難度調整技術，以提供玩家具有適當挑戰性的遊戲經驗同時不致於產生挫折感。接著將進一步探討如何將此經驗與技術應用於教育型遊戲，以增加教育型遊戲之娛樂性、挑戰性與學習成效。最後我們將開發遊戲娛樂性與學習成效評估機制，作為自適應遊戲人工智慧之有效性評估。本計畫對於自適應遊戲人工智慧之完整研究，將有助於未來學者及業者對於電腦遊戲與教育型遊戲之進ㄧ步研發。
Computer-controlled opponents in most games change their tactics in a finite manner. After playing several times, the human player will recognize the repetition of tactics adopted by the computer-controlled opponents, and then the player will get bored at the game. Hence, in recent years game developers have paid more resources on developing challenging games. For most games, high-quality game AI is the key to providing challenging gameplay. Game AI is defined as the decision-making process of computer-controlled opponents in computer games. It is called adaptive if the computer-controlled opponents can automatically fix weaknesses in the game AI and respond to changes in player’s tactics during gameplay. Adaptive game AI is currently in its infancy. This three-year project aims to develop adaptive game AI for both online and offline adaptation of tactics. First, in this project we will develop various tactics to fit human players with different gameplay experience by using offline adaptive game AI and player modeling. Next, we will develop new online adaptive game AI to improve the tactics of the computer-controlled opponents based on the player’s behaviors, and enhance challenge and entertainment value by using dynamic-difficulty-adjustment techniques. Third, our accumulated experience at adaptive game AI will be applied to educational games to enhance their entertaining and learning effects. Finally, an evaluation of the developed adaptive game AI on educational games will be performed. The research results of this project will forward the developments of adaptive game AI in both entertaining and educational games.