Menú
Shatranj.ai Curriculum — AI Curriculum Based on Historic Board Games (Chess, Shatranj, Qirkat, Dama) for Youth Development
Shatranj.ai is a heritage-first, build-and-experiment AI curriculum. We learn modern AI by rebuilding the intellectual tools that shaped human decision-making for centuries: historic board games. These games traveled across languages and civilizations, carrying patterns of logic, ethics, and creativity. Today, they form one of the clearest laboratories for understanding artificial intelligence—because every move is measurable, every decision is explainable, and every algorithm leaves a trace you can test.
Historic puzzle track (core studies and sources): Across the curriculum we tackle a curated set of historic puzzles and endgame studies: Horse Tour (Knight’s Tour), Eight Queens, the Wheat & Chessboard exponential-growth puzzle, classic mating studies (including Dilaram and early manuscript checkmates), and Suli’s Diamond. Our reconstructions and puzzle narratives draw from two foundational historic chess books: Libro del Acedrez and Kitab ash-Shatranj—two of the most important sources for the cultural and intellectual heritage of chess and Shatranj.
This curriculum goes beyond chess. You will recreate and analyze multiple historic board games such as 3-stone, 9-stone, Mancala, the Royal Game of Ur, and Checkers, plus the richer strategic systems of Qirkat and Shatranj. The goal is to learn AI as a transferable set of ideas, not something tied to a single game.
AI milestones through chess software: We explicitly trace landmark systems and what each one taught the world about “machine intelligence”: The Turk, Deep Blue, AlphaZero, and Stockfish. Learners see how search, evaluation, engineering constraints (chips, memory, speed), and modern learning-based methods shaped the evolution of chess software and AI.
Who this curriculum is for
• Students: learn AI by building systems you can explain, measure, and improve.
• Teachers: a modular course you can teach as a full sequence or standalone units.
• Schools & institutions: a culturally rich AI pathway for clubs, electives, bootcamps, and interdisciplinary programs.
What you will be able to do
• Build board-game environments and implement legal move generation
• Implement classical AI search (DFS, BFS, UCS, A*, minimax, alpha–beta)
• Use dynamic programming to solve a historic endgame study
• Understand how modern chess engines work (representation → movegen → search → evaluation)
• Implement reinforcement learning from tabular Q-learning to deep Q-networks and MCTS
• Understand AlphaZero-style pipelines (policy/value nets + PUCT + self-play)
Curriculum map (25 lessons, grouped into 9 sections)
Section 1 — Foundations & Python Basics (Lessons 1–6)
Lesson 1: Course Scope and Priorities
Lesson 2: Introduction to Computing & Python Setup
Lesson 3: Python Data Types
Lesson 4: Conditionals, Loops, Control Flow
Lesson 5: Functions, Scope, Parameters
Lesson 6: Files, Exceptions, Libraries, Testing
Section 2 — Object-Oriented Programming & Board Game Modeling (Lesson 7)
Lesson 7: OOP, Classes, TicTacToe
Section 3 — Chess Foundations & Engine Code (Lessons 8–9)
Lesson 8: Chess & Shatranj Board Representation
Lesson 9: Piece Movement, Game State Updates, and Terminal Conditions
Section 4 — Classical and Adversarial Search Algorithms (Lessons 10–11)
Lesson 10: Search Problems and Graph Traversal (DFS, BFS, UCS)
Lesson 11: Heuristic Search and Adversarial Game Trees (A*, minimax, expectiminimax, alpha–beta)
Section 5 — Solving Ancient Chess Puzzles with AI Algorithms (Lessons 12–15)
Lesson 12: Horse Tour (Knight’s Tour)
Lesson 13: Eight Queens Puzzle
Lesson 14: Wheat & Chessboard Problem (exponential growth + puzzle mathematics)
Lesson 15: Minimax, Alpha-Beta, Checkmate Logic (with historic checkmates and sources)
Section 6 — Dynamic Programming (Lesson 16)
Lesson 16: Suli’s Diamond (Historic Endgame Study)
Section 7 — The Intertwined History of AI and Modern Chess Software (Lesson 17)
Lesson 17: Stockfish as Modern Chess AI Software (engine architecture, modern search/eval, and adapting engines to historic variants such as Shatranj)
Section 8 — Reinforcement Learning (Lessons 18–22)
Lesson 18: RL Foundations (Gridworld, Dynamic Programming, Complexity)
Lesson 19: Tabular Q-Learning on FrozenLake (Frozen Rook)
Lesson 20: Two Rooks vs Lone King (TD Q-learning)
Lesson 21: Deep Q-Networks (Connect-4, Fox & Hounds, Othello/Reversi)
Lesson 22: Monte Carlo Rollouts and MCTS on Qirkat
Section 9 — AlphaZero (Lessons 23–25)
Lesson 23: AlphaZero on Othello/Reversi (policy/value nets + PUCT + self-play)
Lesson 24: AlphaZero on Qirkat (PUCT, policy/value nets, self-play; path-aware move encoding)
Lesson 25: Turkish Checkers (Dama): Alpha-Beta, PUCT-guided MCTS, AlphaZero comparisons
How to start. If you are new to Python, begin with Lessons 1–6. If you already code, start at Lessons 8–11 to enter engine-building and search. Educators can teach section-by-section as independent units or deliver the full 25-lesson pathway. To access the lessons create a free account and login to our learning management system: lms.shatranj.ai
