AI literacy for
every subject.

Say this to your class

"We are going to watch a drawing game. I will draw what the tool asks for, but I will not tell you the word. Hold up one guess. Then we will write one sentence about the tool's output, with a capital letter, spaces and a full stop. Remember, a guess is not always right."

When it clicks

The concept has landed when a pupil says, "The tool made a guess from patterns, so we checked it."

Say this to your class

"You will help a pretend sentence model. Some pupils will be examples. One pupil will match a new card to the examples. Other pupils will check the sentence. We will use our English rules after the model gives its answer."

When it clicks

The concept has landed when a pupil links the model's guess to the example cards and names the human check.

Say this to your class

"Listen to this short class story. Then you will speak from one role. First, think quietly. Next, talk with your role group. We will hear one idea from each role and use reasons from the story."

When it clicks

The concept has landed when pupils say that an AI output can help but cannot replace a person's check.

Say this to your class

"You will design a paper helper for a child who is writing about a picture. Your helper may give one sentence idea. It must also remind the writer to check the capital letter, spaces and full stop. The writer stays in charge."

When it clicks

The concept has landed when a pupil says that the helper can suggest but the writer chooses and checks.

Say this to your class

"We are going to see if we can teach this computer to be an architect's assistant. I will show it photos of different places of worship, and we will see if it can learn to spot the patterns. Look closely; if I only show it churches with steeples, will it recognise a church that looks like a modern hall? Let's find out how the computer 'thinks' about these buildings."

When it clicks

The concept has landed when a pupil points out a specific feature (like a dome) and says "The computer is looking at that shape to make its guess."

Say this to your class

"We are going to turn our classroom into a giant sorting machine. I am going to give you a secret rule, and you have to sort these symbols using only that rule. Don't think about what the symbols mean to people; just look at the shapes like a computer would. Let's see if our machine works!"

When it clicks

Look for pupils realising that the meaning of the symbol is irrelevant to the sorting task.

Say this to your class

"Listen to this story about a 'Smart Camera' at a mandir. We will then decide what the temple committee should do next."

When it clicks

The concept has landed when pupils suggest that religious experts (not just computer scientists) should help build the AI.

Say this to your class

"Imagine we have discovered a new group of people who worship the 'Spirit of the Forest'. They need a symbol for their sacred space. Your job is to design that symbol, but there is a catch: you also have to write the instructions for an AI so it can find this symbol on a map. What shapes and colours will the AI need to look for?"

When it clicks

The concept has landed when pupils use technical language like "features" or "patterns" to describe their drawing instructions.

Say this to your class

"We are going to use Google Earth to find real volcanoes like Mount Etna. Scientists use AI to look at millions of these pictures to see if a volcano is changing shape. If the ground bulges, the AI alerts the scientists that magma might be rising. Let's see if we can spot the crater from space ourselves."

When it clicks

The concept has landed when a pupil suggests that the computer can see things "all at once" across the whole world.

Say this to your class

"I am going to move my hands like tectonic plates. Sometimes they slide smoothly, but sometimes they jerk and stick. You are the 'Earthquake AI'. You need to learn the pattern of my hands. When you see me shake my hands three times, that's the warning pattern. You must clap once to set off the alarm. Let's see if you can learn the pattern."

When it clicks

Pupils are successful when they wait for the full triple-shake instead of clapping at any movement.

Say this to your class

"Let's think about the town of Vulcania. We have a computer saying one thing and a human expert saying another."

When it clicks

The concept has landed when pupils recognise that both the AI and the human can be wrong.

Say this to your class

"When you look at your drawing of a volcano, you know exactly what the labels mean. But a computer just sees lines and colours. To teach an AI to recognise a volcano, we have to show it thousands of drawings and tell it: 'This is a crater' and 'This is an ash cloud'. You are going to be the teacher for your partner's AI brain."

When it clicks

The concept has landed when a pupil explains that the "AI" needs clear labels to learn what a feature looks like.

Say this to your class

"You have been using multiplication to make equivalent fractions. Now you will act out one small part of how an AI system can learn from examples. First, use the labelled examples to suggest a pattern. Then test that pattern on new fraction pairs and check each answer with the same multiplier."

When it clicks

Pupils explain that a matching multiplier on the numerator and denominator is stronger evidence than simply seeing larger numbers.

Say this to your class

"You have seen how a pattern can help with a fraction question. Read what the simulated system says about a pupil's idea. Your job is to check the maths in the answer, not to guess whether a real AI wrote it."

When it clicks

Pupils refer to the fractions and the change to both parts, rather than saying the answer is wrong only because a computer produced it.

Say this to your class

"A fraction answer is more useful when you can see how it was checked. Design a paper checker that helps someone compare two fractions. It must show a maths check and leave room for a person to decide whether the result makes sense."

When it clicks

Pupils design a checker that displays the matching multiplier and asks its user to verify the result.

Say this to your class

"This sheet shows what a made-up fraction system did on three questions. Look for a pattern in the inputs and outputs. Then use your equivalent-fraction knowledge to decide whether that pattern answers a different kind of question too."

When it clicks

Pupils distinguish between the output being mathematically equivalent and the output failing to meet the requested denominator.

Say this to your class

"Watch how quickly a player must read the court before making a pass. We will now turn that decision into a short set of instructions, like an algorithm. Follow the rule exactly first, even when you think another choice would be better. Then we will change the situation and decide where human judgement is needed."

When it clicks

The concept has landed when a pupil says that a rule can guide a pass but cannot notice every changing feature of the game.

Say this to your class

"Keep your eyes on the shared screen and your prediction ready on your mini-whiteboard. I will draw a basketball-related picture in Google QuickDraw, but I will not tell you what the system is trying to recognise until we have made our prediction. We will compare the system's guess with the picture and ask which visual clues may have helped or confused it. Remember, a quick guess is not the same as a guaranteed answer."

When it clicks

The concept has landed when a pupil explains that the output came from matching patterns rather than from the system understanding basketball.

Say this to your class

"You have heard what happened at the school in the story. Take a quiet minute to decide which evidence each person in the hearing would use. Then speak from that person's role, not just from your own first reaction. We are not deciding whether technology is always good or always bad. We are deciding what checks are needed before this report influences a team decision."

When it clicks

The concept has landed when pupils separate useful information from a final decision and name a human check before action.

Say this to your class

"You are going to design a paper version of an assistant for a basketball player. It may notice space, defenders or passing options, but it cannot play the pass or choose the team. Your design must show the information it uses, the advice it gives and the moment when the player checks the advice. A clear design is one that another pupil can understand without you explaining every part."

When it clicks

The concept has landed when a pupil designs an advice screen with a clear check or override rather than an automatic pass button.

Say this to your class

"We are going to see if we can teach a computer to recognise the features of the upper and middle course. We will show it photos of waterfalls and meanders. If we only show it perfect, clear photos, will it still recognise a muddy river or a small waterfall? Let's find out how the computer 'learns' the shapes we are studying today."

When it clicks

The concept has landed when a student suggests that the AI needs "more different types of photos" to get better.

Say this to your class

"We know rivers change from the upper to the lower course. If I told you the land was getting flatter and wider, you'd predict a meander. Computers do this too, but they only know what we've told them. Let's see what happens when the river does something the 'Computer' doesn't expect."

When it clicks

The moment to watch for is when the "AI" student feels confident about the fifth card before it is read, showing they have identified a pattern.

Say this to your class

"Listen to this dilemma about a town just like the ones we've seen in our river diagrams."

When it clicks

Look for students suggesting that the AI and the human expert should work together rather than picking just one.

Say this to your class

"If we want an AI to protect our rivers, we need to give it 'senses'. On your river diagrams, you've labelled waterfalls and meanders. Now, I want you to draw where you would put 'Smart Sensors' and what they should measure to tell the AI which part of the river it is looking at."

When it clicks

Success is when a student chooses a sensor type that matches the feature (for example, a speed sensor for a waterfall or a depth sensor for a floodplain).

Say this to your class

"We are going to listen to two different rhythms. One of these was played by a person sitting at a drum kit, and the other was made by an AI that doesn't have a heartbeat. Listen closely to the pulse. Is it perfectly, robotically steady, or does it 'breathe' a little bit? In your pairs, decide which is which and why the robot might sound different to the human."

When it clicks

The concept has landed when a student uses the word "robotic" or "perfect" to describe the AI rhythm as a negative trait.

Say this to your class

"We've been practising our pulses, but now we're going to see if a computer can keep up with us. We'll use a tool that tries to guess the 'BPM' or pulse of our tapping. If we tap slightly off-beat, notice how the computer's guess jumps around. It's trying to find a pattern in your data (your taps). Let's see if you can be steady enough to keep the computer's guess from changing."

When it clicks

The concept has landed when students notice the lag in the AI's pulse detection and relate it to the computer needing more information.

Say this to your class

"We've seen that AI is great at keeping a perfect pulse, but music is about more than just being a clock. I want us to think about what we lose when we choose a machine's 'perfect' rhythm over a human's 'feeling' pulse. We will look at this from different points of view."

When it clicks

The concept has landed when a student identifies that a "perfect" pulse might be less musical than a human one.

Say this to your class

"If we wanted to build a robot drummer, we'd have to give it a set of very clear rules. It doesn't know what a 'cool' rhythm is; it only knows the instructions we write. In your pairs, you are going to be the architects of a drumming algorithm. You need to write three rules that will make sure your robot keeps a steady pulse but also plays an interesting rhythm. Then, you'll give your rules to another pair to see if they can 'run' your code using body percussion."

When it clicks

The concept has landed when a student realises their rules were too vague and tries to make them more "mathematical" for the "robot".

Say this to your class

"We've been working on keeping the ball on our right-hand side while dribbling. I've set up a tablet with a tool that tries to 'see' humans. As you slalom through the cones, we're going to see if the AI can keep track of you or if it loses you when you move too fast or turn your back. Does it think your hockey stick is part of your arm?"

When it clicks

The concept has landed when a student intentionally hides their stick or changes their pose to "trick" the camera.

Say this to your class

"During this mini-game, one person on the sideline is the 'AI Umpire'. You have a very strict rule: you only signal a foul if the stick goes above knee height. You don't care if it was an accident, you don't care if the player was pushed. You are a machine. Let's see how that changes the game compared to a human umpire who understands 'fair play'."

When it clicks

The concept has landed when players start debating the logic of the AI's decision rather than just complaining about the call.

Say this to your class

"Gather round for a minute. We're going to talk about a big decision in a fictional hockey final."

When it clicks

The concept has landed when students stop talking about the hockey and start talking about accountability (who is to blame).

Say this to your class

"We've been using cones and bibs today. But in 20 years, your hockey pitch might be alive with sensors. In your pairs, I want you to sketch a 'Smart Pitch'. Where would you put the AI sensors? In the ball? In the turf? In the players' shoes? What helpful information would the AI shout out to you during the game?"

When it clicks

The concept has landed when students start considering privacy (what the AI should not see) as well as features.

Say this to your class

"You have already learnt that temperature and pH affect enzyme activity. Now you will test whether a classifier can sort enzyme statements accurately. We will give it examples, test new statements and compare its output with the biology you already know. Treat every output as a claim to check, not as an answer to copy."

When it clicks

The concept has landed when students say that an output depends on the examples and must be checked against enzyme evidence.

Say this to your class

"Look at the catalase results as a scientist, not just as a list of numbers. You will change the evidence available to a hypothetical predictive model and see whether its recommendation changes. Your task is to spot when a neat pattern is weaker than it first appears."

When it clicks

The concept has landed when students explain that the same biological process can produce a different prediction when the available data changes.

Say this to your class

"You know how an enzyme investigation needs controlled conditions and evidence. We are now applying that reasoning to a workplace decision. Read your character's evidence carefully, then argue for accept, revise or reject. Every contribution must name one enzyme fact and one limitation of the AI recommendation."

When it clicks

The concept has landed when students distinguish a useful recommendation from a decision that still requires a named human checker.

Say this to your class

"You have examined how enzyme evidence can be incomplete and how a recommendation can still sound convincing. Now design the paper version of a safer adviser for a catalase investigation. Your design must show what data enters, what the system predicts, what warning it gives and where a scientist takes control."

When it clicks

The concept has landed when students place a real checking action between the AI recommendation and any change to the enzyme investigation.

Say this to your class

"Let's see how a 'Travel Bot' handles holiday planning. I want you to ask the AI to plan a '100% sustainable' holiday to a specific city. But here's the catch: your job is to be the carbon detectives. Did the AI suggest a flight? Did it mention the emissions of that cruise? Let's see if its 'sustainable' plan actually matches our criteria."

When it clicks

The concept has landed when a student questions why the AI didn't mention the carbon footprint of the activities it suggested.

Say this to your class

"I have two adverts for a beach holiday in Spain. One was written by a local travel guide who lives there, and the other was generated by an AI in three seconds. Read both. Which one feels more 'real'? Which one actually gives you specific details about the local people, and which one just uses 'sustainable' buzzwords?"

When it clicks

Students start using terms like "generic" or "template-like" to describe the AI text.

Say this to your class

"The village of 'Port-Eco' is introducing a new AI app. It will automatically direct tourists to 'sustainable' spots to prevent overcrowding. But there's a problem. The owner of a local cafe says the AI never sends people to her because she's not on 'Top 10' lists. The mayor loves it because it stops traffic jams. We are going to hold a hearing: is this AI helping or hurting our local community?"

When it clicks

The activity is successful when students recognise that the AI's efficiency can cause inequality for local people.

Say this to your class

"We've seen how AI can be biased or generic. Now, for your extension task, you are going to design the perfect 'Sustainable Travel Assistant' for the year 2040. It shouldn't just find holidays; it should solve the problems we found in our reading. How does it protect the coral? How does it make sure local people get paid fairly? Draw or map out how your AI makes these choices."

When it clicks

Students start thinking about the inputs (data) needed to make outputs (recommendations) more ethical.