Skip to content
ARKAI Readiness Kollective
AI Literate · Grades 6-8Module 1

Facilitator guide: How AI Actually Works

60 minutesReview status: Being checked
Open the lessonsSlide deck (Google Drive)

Materials

  • A device or screen per small group, or one shared screen
  • Index cards for the unplugged "be the model" sort
  • Printed or projected quick check and pre/post check
  • Paper or a device for Reflect (optional, stays private)

Lesson steps

  1. 1. The guess that felt like mind reading (Explainer, 5 min)
  2. 2. Inputs, patterns, predictions (Explainer, 7 min)
  3. 3. The pet-tagging app at Pigskin (Scenario, 6 min)
  4. 4. Quick check (Quick check, 4 min)
  5. 5. Your own loop (Reflect, 5 min)
  6. 6. Recap (Recap, 3 min)

Timing

Part 60 min 90 min
Pre-check 4 5
Warm-up: “an app guessed wrong about me” 3 5
Step 1: Hook 6 10
Step 2: Core idea, with the “be the model” round 18 25
Step 3: Scenario 8 12
Step 4: Quick check 5 8
Step 5: Reflect 5 7
Step 6: Recap 3 6
Post-check 4 5
Buffer 4 7
Total 60 90

The “be the model” round: one group sorts cards by a secret rule, and the others guess the pattern.

Before you start

  • The Instagram example is illustrative. It is not a judgment on screen time. Let students swap in the app they actually argue about if Instagram isn’t it.
  • Instagram requires users to be 13 or older in most regions, and learners are 11 to 13. Do not ask who has an account.
  • Reflect answers are private. Do not ask anyone to share their screen or read theirs aloud.
  • Name products only as examples. The Pigskin app is made up.

Discussion prompts

  • Use the three in Step 5.
  • For extra time: “Should apps have to explain why they recommended something to you?”

Answer notes

  • Step 2 sort: the common mix-up is treating prediction and feedback as the same thing. A prediction is the model’s guess. Feedback is the new data created by the student’s reaction to that guess.
  • Step 3: B is the best answer. Accept any choice a learner can explain, and use the feedback lines to steer.
  • Step 4: B, C, B, C, B. For Q5, make sure students see machine learning as one specific kind of algorithm, not a separate thing from algorithms entirely.
  • Pre/post: B, A, B.

Words to know

algorithm
A set of step-by-step instructions that a computer follows.
machine learning
A way of building AI where a program learns patterns from examples, so nobody has to write every rule by hand.
training data
The examples an AI tool learns from.
model
What you get after training: a program that takes an input and makes a prediction.
prediction
A model's best guess for a new input.
feedback loop
When a tool's results become its future data, so a pattern keeps repeating.

Workshop quick checks

Learners can answer these on their own phones with no account. Answers count toward ARK's anonymous totals.