Computing · Year 8

Active learning ideas

AI Applications: Image and Voice Recognition

Active learning works because AI applications in image and voice recognition rely on hands-on pattern recognition, where students directly experience how algorithms interpret data. When students train models or test recognition systems themselves, they move from abstract concepts to concrete understanding of AI’s strengths and limits.

• National Curriculum Attainment Targets
• KS3: Computing - Artificial Intelligence
• KS3: Computing - Societal and Ethical Impacts

• 25–40 min
• Pairs → Whole Class
• 4 activities

Activity 01

• Concept Mapping
• 35 min · Small Groups

Demo Lab: Train Image Recognizer

Direct students to Teachable Machine website. They gather 20 images each for two classes, such as happy or sad faces, train the model, then test live with webcam inputs. Groups compare accuracy across different lighting conditions and suggest tweaks.

Explain how AI systems recognize faces or voices.

Facilitation Tip

During Demo Lab: Train Image Recognizer, circulate to ensure students intentionally select varied images to test their model’s robustness against rotation and lighting changes.

What to look for

Pose the question: 'Imagine a smart doorbell that uses facial recognition to identify visitors. What are two benefits and two risks of this technology?' Facilitate a class discussion, encouraging students to consider privacy, security, and potential misuse.

• Understand
• Analyze
• Create
• Self-Awareness
• Self-Management

Activity 02

• Concept Mapping
• 40 min · Pairs

Debate Pairs: Surveillance Pros and Cons

Pair students to research one side: benefits like crime prevention or risks like bias and privacy loss. Pairs present arguments to the class, followed by a whole-class vote and reflection on balanced views.

Analyze the benefits and risks of AI in surveillance and security.

Facilitation Tip

For Debate Pairs: Surveillance Pros and Cons, provide sentence stems to guide structured arguments and time checks to keep the debate focused on evidence rather than opinion.

What to look for

Provide students with a short paragraph describing a scenario involving AI voice recognition, such as a customer service chatbot. Ask them to identify one specific way the AI is 'hearing' and one potential challenge it might face in understanding the user.

• Understand
• Analyze
• Create
• Self-Awareness
• Self-Management

Activity 03

• Concept Mapping
• 30 min · Small Groups

Scenario Board: Helpful or Harmful AI

In small groups, students select a setting like schools or shops, design an AI recognition scenario with sketches showing steps and outcomes. Groups pitch ideas, class votes on most realistic ethical dilemma.

Design a simple scenario where AI image recognition could be helpful or harmful.

Facilitation Tip

In Scenario Board: Helpful or Harmful AI, ask students to justify their categorizations with at least one piece of evidence from the scenario before moving to the next case.

What to look for

On an index card, ask students to write one sentence explaining how AI 'sees' an image and one sentence explaining how AI 'hears' a voice. They should use at least one vocabulary term from the lesson.

• Understand
• Analyze
• Create
• Self-Awareness
• Self-Management

Activity 04

• Concept Mapping
• 25 min · Individual

Voice Test Challenge: Individual Logs

Students record five phrases in varied tones or accents using phone apps or online tools, test recognition rates, and log results in a table. Share patterns in pairs to discuss influencing factors.

Explain how AI systems recognize faces or voices.

Facilitation Tip

During Voice Test Challenge: Individual Logs, remind students to record not just accuracy but also environmental factors like background noise that affect voice recognition performance.

What to look for

Pose the question: 'Imagine a smart doorbell that uses facial recognition to identify visitors. What are two benefits and two risks of this technology?' Facilitate a class discussion, encouraging students to consider privacy, security, and potential misuse.

• Understand
• Analyze
• Create
• Self-Awareness
• Self-Management

A few notes on teaching this unit

Start with concrete examples to ground abstract concepts, such as showing how a simple edge-detection filter works on an image before introducing deep learning models. Use misconception-focused questions to uncover prior beliefs, such as asking students to predict how an AI would label a blurry or upside-down image. Research suggests that students grasp pattern recognition more deeply when they actively manipulate data and observe failures, so prioritize hands-on testing over theoretical lectures.

Successful learning looks like students explaining how image and voice recognition systems process data, identifying real-world limitations, and discussing ethical implications using specific examples from their activities. They should articulate why AI models succeed or fail and how these outcomes connect to broader computing principles.

---

Watch Out for These Misconceptions

• During Demo Lab: Train Image Recognizer, watch for students assuming the AI understands images the way humans do.

  Have students intentionally test rotated, mirrored, or partially obscured images in their training set and document the model’s failures, then discuss why these errors reveal the AI’s reliance on pattern matching rather than true comprehension.

• During Demo Lab: Train Image Recognizer, watch for students believing facial recognition works equally well for all people.

  Guide students to test images with varied lighting, angles, and facial features from their own sets, then compare error rates across different conditions to highlight how biased or incomplete training data reduces accuracy.

• During Voice Test Challenge: Individual Logs, watch for students assuming voice recognition depends only on spoken words.

  Ask students to record identical phrases spoken by different classmates and analyze how pitch, timbre, and background noise affect recognition, then discuss how these factors create unique voice prints.

---

Methods used in this brief

• Concept Mapping