Opening: Activating Exercise

NoteDuration: 10 minutes

Welcome!

Before we dive into the technical details, let’s activate our prior knowledge about learning.

Think-Pair-Share: Learning from Examples

NoteTHINK (3 minutes)

Consider these two scenarios for learning how to calculate correlation:

Scenario A: Problem-Solving

You’re given:

“Calculate the correlation between Variable X and Variable Y”

Then you spend time figuring out the formula, looking up procedures, trying different approaches.

Scenario B: Worked Example

You’re shown:

“Calculate the correlation between your weekly running distance and your 5K race times:

Given data: - Running: [25, 30, 20, 35, 28] km/week - Race times: [22, 21, 24, 19, 20] minutes

Step 1: Calculate means… [Complete solution with every step explained]”

Question: Which scenario would help you learn better? Why?

NotePAIR (4 minutes)

Share your thoughts with a neighbor:

  1. Which scenario did you choose?
  2. Have you experienced either approach in your own learning?
  3. Which approach do you use when teaching?
NoteSHARE (3 minutes)

Brief group discussion:

Common responses:

  • “Scenario B is easier to follow”
  • “The personal context makes it more engaging”
  • “I can see the pattern more clearly in the worked example”
  • “But won’t students just copy without understanding?”

The Bridge to Today’s Workshop

What you just experienced touches on two key principles we’ll explore:

  1. The Worked Example Effect: For novice learners, studying complete solutions is more effective than struggling through problems independently

  2. The Personalisation Effect: Familiar contexts (like your own running data) reduce cognitive load and improve learning

Today’s goal: Build a tool that combines both principles at scale using AI.

ImportantThe Workshop IS a Worked Example

Here’s the key insight: This workshop itself is designed as a worked example.

We won’t ask you to build this tool from scratch. Instead:

  1. We’ll show you a complete, working solution (the personalised example generator)
  2. We’ll study it step-by-step, examining how each component works
  3. We’ll explain why it’s designed this way ,connecting code to cognitive science
  4. We’ll gradually reduce scaffolding, from guided study to independent application
  5. You’ll transfer the pattern, applying it to your own teaching domain

Why this structure?

Because you’re learning to build educational AI tools, a relatively new domain where most of you are “novices.” According to Cognitive Load Theory (Sweller 1988), the most effective way for you to learn is to study a worked example, not struggle through unguided problem-solving.

The meta-lesson: By experiencing the worked example effect yourself, you’ll better understand how to design it for your students.

TipMeta-Cognitive Moment

Throughout today, notice:

  • When we show you complete solutions (worked example effect)
  • When we use familiar teaching contexts (personalisation effect)
  • When we gradually reduce guidance (fading)
  • When we ask you to apply independently (transfer)

The workshop structure demonstrates the principles it teaches!

Ready? Let’s move to Theory

Back to top

References

Sweller, John. 1988. “Cognitive Load During Problem Solving: Effects on Learning.” Cognitive Science 12 (2): 257–85. https://doi.org/10.1016/0364-0213(88)90023-7.

Reuse