Unit 11 of 12 · Your future · about 3 weeks

AI Literacy: How does AI work, and how should I use it?

By · Published

Today's children will spend their working lives alongside AI systems. Knowing roughly how they work — that they learn patterns from examples, can be confidently wrong, and reflect the data they were trained on — is becoming as basic as knowing how to search the web.

The unit draws on everything earlier: bias (training data has bias too), metacognition (checking whether you understood or just copied), media literacy (AI-generated content) and decision-making (when to trust a tool).

Teach a machine

  1. Open Google's free Teachable Machine in a browser.
  2. Have the child train it to tell two things apart — thumbs up versus thumbs down, a cat toy versus a ball — with a few dozen webcam pictures each.
  3. Test it. Then deliberately confuse it: different lighting, a different background, someone else's hand.
  4. Ask: why did it fail? Because it only knows the examples it was shown.

Core ideas

  • AI systems learn patterns from examples (training data) rather than following hand-written rules.
  • They are only as good as their data; gaps and biases in the data show up in the answers.
  • Chatbots predict likely text; they can sound certain and be wrong.
  • Use AI to explain, quiz and give feedback — not to do the thinking you are trying to learn.
  • Never share private information with an AI tool; ask a parent first.

Activities

Break the classifier

After training a Teachable Machine model, list every way you found to fool it, and how you would fix each with better data.

AI as tutor, not answer key

With a parent, ask an AI assistant to quiz you on something from an earlier unit instead of explaining it. Compare how much you remember the next day versus reading its explanation.

Catch the mistake

Ask an AI assistant questions you can verify (local facts, a math problem, a book you have read). Record how often it is wrong and how confident it sounded.

One lesson, three outputs

Grades K–2 (ages 5–8)

Sort pictures into groups and explain the rule — that is what a classifier does. Train a Teachable Machine model together with a parent.

Grades 3–5 (ages 8–11)

Train and break a classifier, and keep a log of AI answers checked against real sources.

Grades 6–8 (ages 11–14)

Explain training data, bias and why chatbots can be wrong; write family rules for using AI for schoolwork; complete a free AI curriculum module.

The family question

Did the AI help me think, or think for me?

Thinking Journal prompt

When did an AI tool help me this week? When was it wrong? How did I find out?

Getting it right

Supervise any use of chatbots by younger children, and check each tool's age requirements — many general-purpose AI assistants require users to be 13 or older.

Free resources