Computer Science · 11th Grade

Active learning ideas

Introduction to Artificial Intelligence

Active learning works well for artificial intelligence because students need to experience how models learn before they can grasp abstract concepts like training data and accuracy. When students train a classifier or sort data into groups themselves, they move from passive listeners to active participants in the learning process.

• Common Core State Standards
• CSTA: 3B-AP-09

• 20–45 min
• Pairs → Whole Class
• 3 activities

Activity 01

• Inquiry Circle
• 45 min · Small Groups

Inquiry Circle: Training a Classifier

Groups use a simple machine learning tool (e.g., Teachable Machine) to train a model to recognize different objects or gestures. They then test their model's accuracy and discuss how to improve it.

Explain the core concepts and goals of Artificial Intelligence.

Facilitation Tip

During Collaborative Investigation: Training a Classifier, circulate to ask each group how changing one training example affects the model’s predictions.

What to look for

On a slip of paper, ask students to write a one-sentence definition of AI in their own words. Then, have them list one example of weak AI and one hypothetical example of strong AI.

• Analyze
• Evaluate
• Create
• Self-Management
• Self-Awareness

Activity 02

• Simulation Game
• 30 min · Whole Class

Simulation Game: Supervised vs. Unsupervised Learning

Students act as 'learning algorithms.' In one scenario, they are given labeled data (supervised), and in another, they must find patterns in unlabeled data (unsupervised). They discuss the differences in their experience.

Differentiate between various subfields of AI (e.g., machine learning, robotics, natural language processing).

Facilitation Tip

For Simulation: Supervised vs. Unsupervised Learning, ask students to compare the two models aloud before revealing the answer.

What to look for

Present students with a list of AI-related technologies (e.g., self-driving cars, Siri, chess-playing programs, a calculator). Ask them to classify each as an example of strong AI or weak AI and briefly justify their choice.

• Apply
• Analyze
• Evaluate
• Create
• Social Awareness
• Decision-Making

Activity 03

• Think-Pair-Share
• 20 min · Pairs

Think-Pair-Share: The Role of Training Data

Pairs discuss how the quality and quantity of training data affect the performance of a machine learning model. They share examples of how biased or incomplete data could lead to incorrect predictions.

Analyze the historical development of AI and its major milestones.

Facilitation Tip

During Think-Pair-Share: The Role of Training Data, listen for pairs that move from ‘good data means good model’ to ‘good data means reliable patterns’.

What to look for

Facilitate a class discussion using the prompt: 'Considering the historical development of AI, what do you believe are the most significant milestones, and why? How might these milestones influence future AI advancements?'

• Understand
• Apply
• Analyze
• Self-Awareness
• Relationship Skills

A few notes on teaching this unit

Teachers should emphasize hands-on practice with real datasets so students see firsthand how models behave. Avoid overemphasizing human-like intelligence in AI; instead, describe models as pattern-finders that improve with better data. Research shows students retain concepts better when they test models with intentionally noisy or biased data before discussing accuracy.

Students will explain the difference between supervised and unsupervised learning, connect the quality of training data to model accuracy, and recognize that machine learning models are tools that find patterns rather than think like humans. They will use evidence from activities to support their reasoning.

---

Watch Out for These Misconceptions

• During Collaborative Investigation: Training a Classifier, watch for students who say the model is ‘learning like a person.’

  Redirect them to compare the model’s predictions after removing one training example; ask how a human would react to losing one memory and why the model’s performance changes.

• During Simulation: Supervised vs. Unsupervised Learning, watch for students who assume unsupervised learning requires labels to function.

  Have them run the simulation twice: once without labels and once with incorrect labels, then discuss which version produced clearer groupings.

---

Methods used in this brief

• Inquiry Circle
• Simulation Game
• Think-Pair-Share