How Do Machines Learn?

Approximate Total Time: 2 hours

Summary:

This hands-on unit introduces students to the foundational ideas behind artificial intelligence and machine learning through playful, age-appropriate exploration. Learners act as “AI trainers,” teaching machines to recognize patterns and make predictions using real-world examples. Through guided experiments with Google’s Teachable Machine and classroom games, students see how data is used to train AI systems, what happens when the data is incomplete or biased, and why fairness matters in technology design.

Lesson Flow:

Lesson 1 – Teachable Machine (60 minutes)
Students explore how machines learn from humans using a process called supervised learning. They practice training a “robot” to distinguish between categories (like crocodiles vs. alligators), then use Google’s Teachable Machine to train a real AI model. By comparing results, they see how diverse data leads to more accurate predictions.

Lesson 2 – Bias in Artificial Intelligence (45–60 minutes)
Building on their machine learning experience, students investigate how AI systems can make unfair or inaccurate decisions due to biased training data. Through activities such as Can a Robot Sort Fairly?, students examine the role of perspective, learn about algorithmic bias, and brainstorm how to make AI systems more fair and responsible.

Approximate Total Time: 2–3 hours

Summary:

In this project-based unit, students explore how machines learn from data through hands-on experimentation with supervised learning and real-world applications. They train and test image-based AI models using Google’s Teachable Machine, analyzing how input data shapes the model’s accuracy and fairness. By engaging with activities, discussions, and media examples—including the story of researcher Joy Buolamwini—students uncover how algorithmic bias forms and how it affects individuals and communities. The unit challenges learners to consider ethical questions around fairness, perspective, and responsible AI design while building foundational skills in computational thinking and data literacy.

Lesson Flow:

Lesson 1 – How Do Machines Learn? (60 minutes)
Students explore supervised learning by training an AI model to recognize patterns in images using Teachable Machine. Working in pairs or small groups, they collect labeled data, test their models, and reflect on how dataset quality, variety, and labeling choices affect performance. Optional extensions invite students to apply machine learning to solve a classroom or community problem.

Lesson 2 – Bias in Artificial Intelligence (45–60 minutes)
Building on their machine learning experiments, students examine how bias enters algorithms and datasets. Through case studies, videos, and online simulations such as The Most Likely Machine, they analyze how imbalanced data can lead to unfair outcomes. Students then design posters to teach others about algorithmic bias and propose ways to create fairer, more inclusive AI systems.

Approximate Total Time: 2–3 hours

Summary:

In this high school unit, students dive into the core principles of machine learning and artificial intelligence. Through interactive lessons and simulations, they explore how computers learn from data, how neural networks process information, and how bias can shape AI outcomes. Students distinguish between supervised, unsupervised, and reinforcement learning, act out the flow of data through a neural network, and investigate how algorithmic bias affects fairness in real-world AI systems. By the end of the unit, learners understand how machines make decisions—and their own role in ensuring AI is developed responsibly and equitably.

Lesson Flow:

Lesson 1 – How Do Machines Learn? (45 minutes)
Students explore the three major types of machine learning—supervised, unsupervised, and reinforcement learning—through real-world examples and scenario-based activities. They identify how data, feedback, and decision-making drive each approach and discuss which type powers familiar AI tools.

Lesson 2 – Inside a Neural Network (45 minutes)
Students step inside the “brain” of an AI by simulating a neural network. Acting as nodes in input, hidden, and output layers, they experience how data flows, how mistakes lead to adjustments, and how connections strengthen learning over time. Optional digital extensions let students train a simple image model using Google’s Teachable Machine.

Lesson 3 – Bias in Artificial Intelligence (30 minutes)
Students examine how bias enters AI systems through data and design. Using videos, discussions, and the “Bias Detective” activity, they uncover examples of algorithmic bias, analyze its human impact, and brainstorm ways to build fairer, more inclusive technologies.