Culture defines us, and preserving it is essential for maintaining our identity and heritage. While recent advancements in AI have focused on digitizing African languages, cultural preservation extends beyond language to include traditional games, many of which face extinction due to the rise of video games. This project takes a step toward preserving African culture by digitizing Ayo, a traditional Nigerian mancala game.

Using Python, with NumPy and PyTorch for algorithm development and Tkinter and FastAPI for the graphical interface and backend, this research implements reinforcement learning (RL) agents capable of mastering Ayo. Three agent types—Heuristic, Minimax, and Pruned Minimax—were trained, with the Pruned Minimax agent demonstrating exceptional performance. It not only outperformed other agents in simulations but also defeated every human player tested so far.

This ongoing work bridges AI and cultural heritage, showcasing Python's versatility in addressing real-world challenges. By digitizing traditional games, we ensure they remain accessible and relevant for future generations. Attendees will gain insights into the technical implementation of reinforcement learning for game AI, explore the challenges of adapting AI to traditional games, and witness a live demonstration of the Ayo game interface. This project serves as a blueprint for preserving and promoting cultural heritage globally, demonstrating the power of Python in advancing AI research and cultural preservation.