AI Literacy Lesson Plans for K-8: 6 Ready-to-Teach Lessons

AI literacy lesson plans for K-8 work best when most of the lesson happens away from the screen. Children learn what AI is, how it learns from examples, why it makes mistakes and who is responsible for it, then test those ideas on a real tool for ten minutes. Below are six ready-to-teach lessons, from kindergarten to grade 8, that you can run this week with paper, sticky notes and one supervised device.

What AI literacy means for 5- to 14-year-olds

AI literacy is not "how to write a good prompt". For primary and middle school students it is four ideas, in this order:

1. AI is a tool people built. It is not alive, it does not have feelings, and someone is responsible for it.
2. AI learns patterns from examples. Change the examples and you change what it does.
3. AI makes mistakes, confidently. Checking is a habit, not an insult.
4. AI affects people unequally. Bias, privacy and fairness are part of the topic from the start.

These map onto the established frameworks, so your lessons line up with whatever your district or trust adopts later:

The OECD link matters for school leaders: AI literacy is heading into international benchmarking, so it is worth building a K-8 thread now rather than bolting it on in high school.

How to use these AI literacy lesson plans

• Length: 30-45 minutes each. Lessons 1-3 suit K-2 or grades 1-3; lessons 4-6 suit grades 3-8, with notes for stretching older students.
• Devices: Five of the six lessons are fully unplugged or need only a teacher-led screen. Where students use AI, use a child-safe, school-approved tool, not a consumer chatbot with an age limit your students do not meet. See ChatGPT in schools for why.
• Order: Teach lesson 1 or 4 first. Every later lesson assumes students already know AI is a tool, not a friend.

Lesson 1: Is it AI or not? (K-2, 30 minutes)

Objective: Students can sort everyday objects into "follows a fixed rule" and "learns from examples", and say that AI is made by people.

Big idea: Perception, Societal Impact.

1. Hook (5 min): Show pictures of a light switch, a calculator, a smart speaker, a robot vacuum, a teddy bear and a phone camera that recognizes faces.
2. Sort (10 min): In pairs, students sort the cards into three piles: "not a computer", "computer that follows one rule", "computer that learned from lots of examples". There will be arguments. Good.
3. Discuss (10 min): Ask "How does the camera know it is a face?" Draw out: it was shown thousands of face pictures by people.
4. Exit ticket (5 min): Students finish the sentence: "AI is made by ___ and it learns from ___."

Watch for: Children saying the speaker "is my friend". Gently reframe: "It is a tool that is good at talking. Your friends are people."

Lesson 2: Teach the robot (K-3, 40 minutes)

Objective: Students can explain that AI learns from examples and that bad examples make bad AI.

Big idea: Learning.

1. Setup (5 min): One volunteer is "the robot". The robot leaves the room.
2. Training (10 min): The class picks a secret rule, for example "animals that live in water". They show the robot 10 picture cards, saying "yes" or "no" for each.
3. Testing (10 min): Show the robot five new cards. Can it guess? Count how many it gets right.
4. Bad data round (10 min): Repeat, but only show water animals that are fish. Then test with a whale and a duck. The robot gets confused. Ask why.
5. Exit ticket (5 min): "If you only show the robot fish, it thinks water animals are ___."

Stretch for grade 3: Ask what would happen if the training cards only showed one kind of person. This is the doorway to bias.

Lesson 3: Spot the mistake (grades 1-3, 35 minutes)

Objective: Students can find a confident mistake in an AI answer and check it against a book or the teacher.

Big idea: Natural Interaction.

1. Teacher demo (10 min): On the board, ask a child-safe AI tool three questions about your current topic. Before class, prepare one answer you know contains an error, or ask about something very local the AI cannot know (your school's lunch menu).
2. Detective work (15 min): Groups get the printed answers and a "detective sheet" with three columns: "Sounds right", "Sounds wrong", "How can we check?"
3. Share (5 min): Each group reports one thing they would check.
4. Exit ticket (5 min): "When AI tells me something, I can check it by ___."

Key line to repeat: "AI can sound sure and still be wrong."

Lesson 4: Prediction machine (grades 3-6, 40 minutes)

Objective: Students can explain that language AI predicts likely next words rather than looking up the truth.

Big idea: Representation and Reasoning.

1. Warm-up (5 min): Read sentence starters aloud and let the class shout the next word: "Once upon a ___", "Peanut butter and ___".
2. Paper model (15 min): Groups get a short paragraph about your topic with every fifth word blanked. They fill in the most likely word, then compare. Most groups pick the same words. Discuss why.
3. The twist (10 min): Give a paragraph about a made-up animal. Groups still fill the blanks fluently, even though none of it is true. That is a hallucination, built by hand.
4. Exit ticket (10 min): In two sentences, explain to a younger student why AI sometimes makes things up.

Stretch for grades 6-8: Compare two AI answers to the same question phrased differently, and discuss why wording changes the output.

Lesson 5: Fair or unfair? (grades 4-8, 45 minutes)

Objective: Students can identify how training data and design choices make AI systems work better for some people than others.

Big idea: Societal Impact.

1. Scenario cards (10 min): Give groups real-world-style scenarios: a voice assistant that understands some accents better than others; a photo app trained mostly on adult faces; a homework helper that only works in one language.
2. Root cause (15 min): For each, groups answer: Who was in the training examples? Who was left out? Who is affected?
3. Redesign (15 min): Groups write three changes the designers should make.
4. Exit ticket (5 min): "One question I will ask about any new AI tool is ___."

This lesson is where the UNESCO "human-centred mindset" and the AILit "Shape AI" domain come alive. Keep it concrete and avoid making any individual student's accent or language the example.

Lesson 6: My AI rules (grades 3-8, 40 minutes)

Objective: Students co-write a class charter for when and how AI may be used, including privacy and honesty.

Big idea: all five, as a capstone.

1. Recap (5 min): Quick-fire review of lessons 1-5.
2. Draft (15 min): Groups write rules under four headings: "When we can use AI", "What we never type into AI" (full name, address, school, photos of people), "How we say we used AI", "What we do when AI says something upsetting".
3. Merge (15 min): Combine into one class charter. Students sign it.
4. Display (5 min): Post it next to the board and send a copy home.

Older students can compare their charter with your school's actual AI policy. If you do not have one yet, our AI policy for schools template is a good starting point, and a letter to parents about AI can carry the charter home.

Assessing AI literacy without a test

Use the exit tickets as your evidence. By the end of the six lessons, most students should be able to:

• Say that AI is made by people and learns from examples.
• Give one reason AI can be wrong and one way to check.
• Name one thing they never share with an AI tool.
• Describe one way an AI system could be unfair.

A simple three-point rubric (not yet, getting there, secure) against those four statements is enough for a report comment.

Where students should try AI for real

The unplugged lessons build the concepts; ten minutes of real use makes them stick. When students do use AI, three conditions matter:

• The tool is built for their age. Answers adapt to a 6-year-old differently than to a 13-year-old.
• You can see what they asked. Lesson 3 and lesson 4 rely on reviewing real conversations afterwards.
• No student personal data is needed to sign in.

Askie for Schools was built for this setting: you paste a class list, each student gets a private access code, answers adapt to each student's age (Askie serves ages 4-15), and you can read the conversation history afterwards. You can preview it as a student before your class touches it. For broader free curriculum, Common Sense Education also publishes short AI literacy lessons for grades 6-12 that adapt well for upper primary.

Frequently asked questions

What age should AI literacy start?

Kindergarten. The AI4K12 guidelines start at K-2 with ideas like "computers see and hear through sensors". Young children already talk to smart speakers, so the question is whether they get an accurate mental model or an invented one.

Do I need devices to teach AI literacy?

No. Five of these six lessons work with paper and one teacher screen. Concepts like learning from examples and prediction are often clearer unplugged.

How often should I teach AI literacy?

One lesson every two to three weeks works well, threaded through computing, PSHE or social studies. Short, repeated exposure beats a one-off assembly.

Is AI literacy the same as digital citizenship?

They overlap on privacy and online safety, but AI literacy adds how systems learn, why they make mistakes, and who is accountable. Teach them together where you can. For a parent-facing take, share how to talk to kids about AI with families.