AI Explorables for Schools

Welcome to the course AI Explorables for Schools!

In this course, (prospective) teachers will playfully explore two central areas of machine learning: artificial neural networks and reinforcement learning.

• Modules 1 and 2 introduce artificial neural networks, their structure, training, and underlying mathematical methods.
• Modules 3 and 4 focus on reinforcement learning (also known as reward-based learning), with a special emphasis on understanding and balancing different strategies.

As the name suggests, this course emphasizes playful and exploratory learning. To support this, we provide "Explorables" – small, interactive online games that run directly in your browser without the need for additional software. You will also find suggestions on how to integrate these Explorables into your teaching. In addition, short exercises throughout the course give you the chance to test and consolidate your knowledge.

This course is designed primarily for teachers and student teachers in STEM subjects, with a focus on computer science and mathematics teaching. However, the content is also well-suited for extracurricular contexts such as workshops, clubs, or project days. Beyond that, the course can also be taken by interested students (independently or with guidance), parents, and a broader audience.

• Overview of the fundamental topics of machine learning, particularly neural networks, gradient descent, exploration vs exploitation, and reinforcement learning.
• Introduction to theoretical functioning and mathematical relationships.
• Teaching content, objectives, and methods for using games and experiments to create contemporary, exploratory learning experiences.

After completing the course, I can

• explain the structure and functioning of artificial neural networks.
• describe the gradient method and the “exploration vs. exploitation” concept in machine learning.
• explain the basic principles of reinforcement learning.
• develop my own game strategies based on heuristics (e.g. “experience-based decision rules”).
• compare human learning with artificial learning.
• teach basic terms, methods and aims of machine learning with a focus on artificial neural networks and reinforcement learning, creatively and at different levels in the classroom, using the corresponding AI Explorables.

• No prerequisites

IMAGINARY is a hub for modern mathematics communication. As a non-profit organization, we aim to make mathematical knowledge freely accessible and share it with people around the world. A distinctive feature of IMAGINARY is its interactive, open-source content, which actively involves scientists and stays closely connected to current research.

We would like to thank the whole team involved in the implementation of the course. In addition to the listed authors Andreas Matt, Bianca Violet, Eric Londaits, Fabian Graap, Antonia Mey and Christian Stussak, Juliane Sperling, Elisabeth Schaber, Oliver Schön, Karla Schön, Daniel Ramos and Johanna Marschall were also involved.

The Explorables are based on exhibits from the exhibition “I AM A.I. – Explaining Artificial Intelligence”, which was developed by IMAGINARY and funded by the Carl Zeiss Foundation.