What Is AI? Machines That Learn
Rules vs learning from examples, training data, AI you already use, why AI makes mistakes and bias.
⌨️ Start with your 10-minute typing warm-upBy the end, you will be able to:
- Explain in their own words that AI learns patterns from examples (training data), unlike a normal program that follows rules people wrote.
- Give everyday examples of AI that recognises, predicts and creates.
- Describe how training data affects AI, and explain "garbage in, garbage out" and bias with an example.
What you need
No computer at home? Print the PDF and practise on paper.
Follow these steps in order
Tick each step when you finish it. Your ticks are saved on this device, so you can stop and come back later.
Learn the ideas ⏱️ 35 min
Read pages 1–4 of the workbook and try every small task. Type your answers straight into the boxes: they save by themselves.
Explore the 3D lab ⏱️ 15 min
Play with the 3D model below. The 🎯 missions tell you exactly what to do next. When you finish every mission, this step ticks itself.
Do the Lab Mission ⏱️ 25 min
Page 5 is the hands-on task. Do it on a real computer, step by step. You need: a computer with a web browser (no AI account needed).
Check yourself ⏱️ 10 min
Answer the questions on page 6 without looking back, then check your answers below.
Finished? Check your answers
Where have you seen AI? (Page 2): keyboard next word: P. Face unlock: R. "Videos you might like": P. Google Translate: C (it writes a new sentence; accept P with a good reason, since it predicts the most likely translation). Spam folder: R (it recognises scam messages). Chatbot answer: C. Any reasonable Ghanaian idea is fine, for example spotting crop diseases from a phone photo, reading X-rays, predicting rain for farmers, translating local languages, or catching MoMo fraud.
Plantain puzzle (Page 3): Mystery 1: Ripe. Mystery 2: Not ready. Mystery 3: Ripe. The pattern: soft (or yellow, brown or black skin) means ripe; hard and green means not ready. The best answers notice that softness is the most reliable clue, because colour alone fails for the black plantain if you only saw yellow ones.
Think about it (Page 4): it might mark pupils with messy, small or unusual handwriting (or left-handed pupils) as wrong even when their answers are right. Fix: train it on handwriting from many different pupils, and have a teacher check its marks.
Lab table (Page 5): scores vary. Round 1 usually scores lowest, Round 2 usually highest, and Round 3 shows the robot repeating the wrong label on items that look like the one taught wrongly.
Spot AI at home: any three suggested words.
True or False (Page 6)
- AI is alive and thinks and feels like a person: False.
- AI learns patterns from examples called training data: True.
- If an AI sounds very sure, the answer must be correct: False.
- An AI trained mostly on one group can be unfair to other groups: True.
- The keyboard suggesting the next word is AI predicting: True.
Written Review Questions
- A normal program follows exact rules a person wrote (IF this THEN that). An AI learns its own pattern from many examples and then guesses about new things.
- If you train an AI on bad data (wrong, too little or one-sided), it learns a bad pattern and gives bad answers. Example: a bottle labelled "paper" makes the robot put bottles in the paper bin.
Rate yourself and save your work ⏱️ 5 min
Tick how confident you feel on the last page, then press 🖨 Print / Save as PDF in the workbook to keep a copy. On a shared computer, press Clear my answers when you have saved it.
🎉 Session 18 complete!
Brilliant work. Next up: Session 19: Using AI Wisely & Safely.
Go to Session 19 →🧠 Teach the Robot
Show a recycling robot examples of plastic, paper and cans, then watch it sort things it has never seen. Give it bad examples and see how AI gets things wrong. Follow the 🎯 missions on the right: they tell you what to do next.
Stuck? Try this
Most problems are one missed word. Look for the Panic Button or Help box in the workbook.
Refresh the page. On a very old computer it may not show 3D. The workbook still has everything you need.
Answers save on the computer you typed them on. Open the workbook on that same computer and browser.
Ask your mentor, a teacher or a friend. Show them the step number you are on.
Going somewhere without internet? Download this session's study pack (workbook, PDF and README). Last updated 30 October 2026.