Integrating Explainable AI in Chess Coaching Systems for Beginners.

While today’s AI chess programs are extremely capable, they do a poor job of teaching new chess players. These systems often provide a player with the best action to take but rarely explain its relevance or how it was determined. By providing the best move without context, these new players cannot take away the important foundational learning that should occur while they are improving their game. This paper introduces a new chess coaching system that utilizes Explainable AI (XAI) to solve this issue. Our intelligent coaching system is capable of providing a simple, understandable explanation for the complex reasoning done by a chess engine for both expert and novice players: for a beginner on a novice chess platform, the intelligent coaching system would not only tell the player to move their knight forward but also inform them that it was a good action because it ‘develops’ the knight, placing it into a better position, while allowing for future options to control the centre of the board, being on an open file, and creating a threat to their opponent’s rook. In this paper we provide a description of how the coaching system was designed, offer examples of the explanations that would be provided, and explore how this approach enables AI to be much more than an analysis tool but also a mentor. This work is designed to explore how AI-enabled learning could be made more affordable and feasible for new learners interested in playing chess.