The ChessAIThon project (2025-1-ES01-KA220-VET-000354329) is co-funded by the European Union. The views and opinions expressed in this publication are those of the author(s) only and do not necessarily reflect those of the European Union or the Spanish Service for the Internationalisation of Education (SEPIE). Neither the European Union nor the National Agency SEPIE can be held responsible for them.
Lesson plan examples proposed by the project partners
Students will investigate popular chess openings used by the global chess community, analyze massive real-world play data using Python, and ultimately use their own chess intelligence to fine-tune the ChessAIThon AI. By the end of this challenge, students will have hands-on experience in the complete machine learning lifecycle: data collection, filtering, training, and model deployment.
Download the Lesson Plan (PDF)Students will play critical checkmate scenarios from famous games against ChessAIThon's AI and document the most challenging case. By analyzing server logs, they will compare the priority move suggested by the neural network (CNN) with the move ultimately chosen after the Monte Carlo simulation (MCTS). They will contrast these two AI metrics with their own game decision and the actual historical solution of the famous chess game. Through this lesson, students will develop algorithmic auditing and critical analysis competencies by comparing the ChessAIThon neural network's static policy recommendations against the Monte Carlo Tree Search (MCTS) simulation decisions within the server logs, while evaluating their own technical decision-making against historical chess solutions.
Download the Lesson Plan (PDF)This lesson plan introduces students to a classical machine learning regression challenge: predicting the continuous winning probability of a chess board state using Leela Chess Zero datasets. Students transform FEN strings into tabular numerical features and train Linear Regression and Decision Tree models with Scikit-Learn. Learners leverage generative AI tools responsibly to research feature engineering and generate training code, while documenting and auditing all AI-assisted blocks to maintain full explainability and intellectual ownership. Acquired Competencies: Feature Engineering (transforming string-based chess FEN data into tabular numerical formats); Classical ML Regression (training and evaluating models using MSE and R² metrics); Explainable AI & Auditing (critically managing AI coding tools while understanding the underlying model mathematics and feature extraction logic).
Download the Lesson Plan (PDF)This lesson is about a grandparent of the ChessAIThon AI and its victory against Kasparov (https://en.wikipedia.org/wiki/Deep_Blue_versus_Garry_Kasparov). Students will investigate these games and the computing principles of Deep Blue. They will discuss whether its architecture is better than neural networks, and the advantages and issues of each solution. Students will transform every game to CSV format to load it on the platform (they can use AI to make a small Python script to do it). Then, with the games loaded, they will check if the decisions of our AI are the same as Kasparov or Deep Blue, and compare them with Stockfish decisions. This lesson will focus on middle game scenarios, because they are the most difficult to predict and the most decisive for the final result. Finally, they can complete moves with policy and value and make a dataset to train a new version of the ChessAIThon neural network.
Download the Lesson Plan (PDF)Students build a custom web chess frontend using chess.js and connect it to a ChessAIthon backend via asynchronous HTTP requests to render real-time AI move recommendations in FEN format. Students are encouraged to accelerate their workflow using responsible coding assistants like OpenCode, provided they document all generated code, verify its security and maintain full intellectual ownership by being able to explain and debug the asynchronous communication logic themselves. Through this lesson, learners acquire vital full-stack AI engineering competencies, master interoperability standards using FEN and UCI communication protocols, and develop explainable software development skills by bridging lightweight client-side tools with active neural network endpoints.
Download the Lesson Plan (PDF)