Efficient AI Strategies with Monte Carlo Tree Search: A Case Study on Tic-Tac-Toe
Session
Computer Science and Communication Engineering
Description
Tic-Tac-Toe, though a simple game, represents a clear environment to study decision-making in artificial intelligence. The challenge lies in enabling an AI to make optimal moves without relying on exhaustive search or handcrafted rules, which often limit adaptability and scalability. Traditional approaches such as Minimax guarantee optimal play but depend on complete state-space evaluation, making them inefficient for larger or more complex problems. To address this, we implement the Monte Carlo Tree Search (MCTS) algorithm, which balances exploration and exploitation through four key stages: selection, expansion, simulation, and backpropagation. By simulating numerous possible plays rather than traversing the entire game tree, MCTS offers a flexible, probabilistic decision-making process that can adapt to uncertain or dynamic domains. In this work, we design and test an MCTS-based AI agent for Tic-Tac-Toe. We evaluate its runtime efficiency, decision quality, and fairness compared to classical approaches. Experimental results show that MCTS consistently produces near-optimal strategies with modest computational requirements, though a noticeable advantage remains for the first player. These findings demonstrate both the strengths and current limitations of MCTS when applied to structured decision problems. Ultimately, this research illustrates how MCTS can provide a scalable foundation for AI decision-making. Beyond Tic-Tac-Toe, the same principles can be extended toward more complex domains such as strategic games, planning tasks, and autonomous systems.
Keywords:
Monte Carlo Tree Search, Tic-Tac-Toe, Simulation-Based Search, Strategic Game Modeling
Proceedings Editor
Edmond Hajrizi
ISBN
978-9951-982-41-2
Location
UBT Lipjan, Kosovo
Start Date
25-10-2025 9:00 AM
End Date
26-10-2025 6:00 PM
DOI
10.33107/ubt-ic.2025.92
Recommended Citation
Azizi, Argjend; Halimi, Gentrit; and Pira, Rigon, "Efficient AI Strategies with Monte Carlo Tree Search: A Case Study on Tic-Tac-Toe" (2025). UBT International Conference. 24.
https://knowledgecenter.ubt-uni.net/conference/2025UBTIC/CS/24
Efficient AI Strategies with Monte Carlo Tree Search: A Case Study on Tic-Tac-Toe
UBT Lipjan, Kosovo
Tic-Tac-Toe, though a simple game, represents a clear environment to study decision-making in artificial intelligence. The challenge lies in enabling an AI to make optimal moves without relying on exhaustive search or handcrafted rules, which often limit adaptability and scalability. Traditional approaches such as Minimax guarantee optimal play but depend on complete state-space evaluation, making them inefficient for larger or more complex problems. To address this, we implement the Monte Carlo Tree Search (MCTS) algorithm, which balances exploration and exploitation through four key stages: selection, expansion, simulation, and backpropagation. By simulating numerous possible plays rather than traversing the entire game tree, MCTS offers a flexible, probabilistic decision-making process that can adapt to uncertain or dynamic domains. In this work, we design and test an MCTS-based AI agent for Tic-Tac-Toe. We evaluate its runtime efficiency, decision quality, and fairness compared to classical approaches. Experimental results show that MCTS consistently produces near-optimal strategies with modest computational requirements, though a noticeable advantage remains for the first player. These findings demonstrate both the strengths and current limitations of MCTS when applied to structured decision problems. Ultimately, this research illustrates how MCTS can provide a scalable foundation for AI decision-making. Beyond Tic-Tac-Toe, the same principles can be extended toward more complex domains such as strategic games, planning tasks, and autonomous systems.
