In December of 2017, DeepMind introduced AlphaZero, a machine learning algorithm that taught itself to play chess, shogi, and go. After only four hours of training, it was arguably the most powerful chess engine at the time.

Why not build your own???

No, seriously, let's do it! If you're like the rest of the world, you played at least a few games of chess after binge watching The Queen's Gambit. Whether you've played chess or not, it's much easier than you might think to create your own chess engine. And what better way to learn some AI/ML than to write a program that can beat any human on the planet in a game of chess?

We will start with a simpler, but related, problem to solve: Connect 4. This will lay the groundwork for the chess engine by examining step by step how to approach solving Connect 4 heuristically, then using deep learning to train a neural network to play the game. Once Connect 4 is solved, we will take what we learned and apply it to chess, working to train a neural network that teaches itself.

You should leave this workshop with a working algorithm to solve Connect 4 and chess, a better understanding of how AI/ML can be applied to solve problems, and, with any luck, some ideas on how to use AI/ML in your daily work.