Printable lesson summaries

Big idea in one sentence

Superintelligence is a mind that would be much smarter than the smartest humans at almost everything — nobody has built one yet, and understanding it helps us make good choices.

What the 3D scene shows

A staircase where each step holds a cleverer creature — ant, mouse, dog, chimp, human — then three glowing steps rising into the clouds, marked 'smarter than any human', 'smarter than all humans together' and 'superintelligence'. Spark the robot climbs when you tap, click or press the arrow keys.

Key words

• Narrow AI — a computer that is great at one job only.
• AGI — an AI that can learn anything a person can.
• Alignment — making sure AI wants what we really mean.

Quick quiz

1. What does "superintelligence" mean?

   • ☐ A robot with a very strong body
   • ☐ A mind much smarter than the smartest humans at almost everything
   • ☐ Any computer that can play chess

2. Has anyone built a superintelligence yet?

   • ☐ Yes, in 2020
   • ☐ No — it is still an idea
   • ☐ Only in secret

3. Why do researchers say we should be careful?

   • ☐ Because a very smart machine needs to understand what we really mean
   • ☐ Because robots will have red eyes
   • ☐ Because computers hate people

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

Intelligence is a toolbox of many different skills, not a single score — and different minds (people, animals and computers) have very different toolboxes.

What the 3D scene shows

A pink low-poly brain on a stand, with Spark beside it. Five everyday tasks sit as buttons on the left: solving a puzzle, recognising a friend's face, learning a song, planning a party, catching a ball. Tapping or dragging a task onto the brain lights up a different coloured region and explains which part does that work.

Key words

• Intelligence — the ability to learn, understand and solve new problems.
• Memory — storing things you learned so you can use them later.
• Reasoning — working things out step by step.

Quick quiz

1. Which is the best description of intelligence?

   • ☐ Knowing lots of facts
   • ☐ One number that says how clever you are
   • ☐ A toolbox of many different skills

2. A calculator does sums faster than any human. Is it more intelligent than you?

   • ☐ Yes — speed is intelligence
   • ☐ No — it can only do one thing and can't learn anything new
   • ☐ Only if it is plugged in

3. Which of these uses several different skills at once?

   • ☐ Planning a birthday party
   • ☐ Standing still
   • ☐ Blinking

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

Narrow AI does one thing well, AGI could learn anything a person can, and superintelligence would be far better than all of us at nearly everything — only the first one exists today.

What the 3D scene shows

Three robots on a round stage. A small blue robot labelled Narrow AI, a medium green robot labelled AGI, and a giant golden glowing robot labelled Superintelligence. Tapping a robot, or its name, shows a card listing what it can and can't do.

Key words

• Narrow AI — AI that is great at one job only.
• AGI — artificial general intelligence — learns anything a person can.
• ASI — artificial superintelligence — far smarter than any human.

Quick quiz

1. A chess computer that beats every human but can't do anything else is…

   • ☐ Narrow AI
   • ☐ AGI
   • ☐ Superintelligence

2. What does the G in AGI stand for?

   • ☐ Giant
   • ☐ General
   • ☐ Good

3. Which of these exists today?

   • ☐ Narrow AI
   • ☐ AGI
   • ☐ ASI

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

AI learns from examples: each correction nudges millions of tiny settings in a network, so its guesses slowly get better — which means it is only as good as the examples it is given.

What the 3D scene shows

Spark stands beside a pedestal holding a piece of 3D fruit (apple, banana, orange or grapes) and makes a guess. You tell Spark whether it was right or what the fruit really was. Behind Spark, a network of blue dots connected by lines glows brighter with every correction, and a meter shows Spark's skill rising.

Key words

• Training — teaching an AI with lots of examples.
• Neural network — a web of connected 'dots' that learns patterns.
• Data — the examples an AI learns from.

Quick quiz

1. How does most modern AI learn?

   • ☐ Someone types in all the rules
   • ☐ From lots of examples, adjusting a little each time
   • ☐ It is born knowing everything

2. What happens if you show an AI lots of wrongly labelled examples?

   • ☐ It ignores them
   • ☐ It learns the wrong thing
   • ☐ It gets angry

3. Why do the lines in Spark's network get brighter?

   • ☐ Because the connections that led to right answers are strengthened
   • ☐ Because the computer is getting hot
   • ☐ Because Spark is happy

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

A superintelligence could think far faster than us and run as millions of copies — and if it could improve itself, some experts think it might race ahead in an 'intelligence explosion', while others think progress would stay gradual.

What the 3D scene shows

A zoom-out in four stages. Stage 1: a single pink brain with Spark beside it. Stage 2: shelves of colourful books appear — a library. Stage 3: the camera pulls back to show a city of towers, each topped with a glowing dot for a 'thinker'. Stage 4: the whole city sits on a blue-and-green planet. Buttons let you zoom in and out.

Key words

• Intelligence explosion — AI improving itself faster and faster.
• Recursive self-improvement — a system that upgrades its own design.
• Compute — the computer power AI runs on.

Quick quiz

1. Why might a superintelligence be faster than a human at thinking?

   • ☐ It would have a bigger head
   • ☐ Computer signals travel far faster than signals in a brain
   • ☐ It would drink more coffee

2. What is the 'intelligence explosion' idea?

   • ☐ An AI that literally explodes
   • ☐ An AI smart enough to build a smarter AI, which builds a smarter one, faster each time
   • ☐ Lots of people learning at once

3. Do experts agree the intelligence explosion will happen?

   • ☐ Yes, everyone agrees
   • ☐ No — some think it would race ahead, others think progress would stay gradual
   • ☐ Nobody has thought about it

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

Superintelligence does not exist yet; AI has improved faster than most experts predicted, but their guesses about what comes next range from 'this decade' to 'never', and nobody knows for sure.

What the 3D scene shows

A long road with signposts. Each signpost is an AI milestone, from Alan Turing's 1950 question to today. Spark walks along as you press the arrows. Past the 'Today' post the road enters thick fog, where glowing, see-through signs show different expert predictions, each labelled clearly as a prediction and not a fact.

Key words

• Milestone — an important moment on the way to something.
• Prediction — a guess about the future, not a fact.
• Benchmark — a test used to measure how good an AI is.

Quick quiz

1. Is superintelligence here yet?

   • ☐ Yes, since 2022
   • ☐ No — today's AI is impressive but still makes simple mistakes and isn't better than experts at most things
   • ☐ Yes, but it's hidden

2. Why do experts disagree so much about the timing?

   • ☐ They haven't read the same facts
   • ☐ Nothing like this has happened before, so everyone is guessing from clues
   • ☐ Only one of them is clever

3. What is the best way to treat an AI prediction you read online?

   • ☐ As a fact
   • ☐ As one person's guess, and ask what they might be missing
   • ☐ Ignore it completely

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

The danger of superintelligence is not evil robots but a powerful machine misunderstanding what we really want — so the most important job is alignment: making sure AI's goals truly match ours.

What the 3D scene shows

A lavender genie-robot floats beside four small people, a pile of toys, and a scoreboard. You choose a wish, such as 'Make everyone smile'. The machine grants it far too literally — paper smiles glued on faces, toys thrown out of the window, a room stuffed with pillows, a scoreboard hacked to 999,999. Then you reword the wish so the machine understands what you really meant, and the people bounce happily.

Key words

• Alignment — making sure AI wants what we really mean.
• Goal — what a system is trying to achieve.
• Guardrails — rules and checks that stop AI doing harmful things.

Quick quiz

1. In The Wish Machine, why does 'make everyone smile' go wrong?

   • ☐ The machine is evil
   • ☐ The machine does exactly what the words say, not what you meant
   • ☐ The machine is broken

2. What does 'alignment' mean in AI?

   • ☐ Making AI's goals match what people really want
   • ☐ Lining up robots in a row
   • ☐ Making AI faster

3. Which is the best way to feel about AI safety?

   • ☐ Terrified
   • ☐ Careful and curious, like learning to ride a bike with a helmet on
   • ☐ Bored — it's not a real issue

Talk about it

Write your own "big idea in one sentence" for this lesson:

Big idea in one sentence

A carefully built superintelligence could help cure diseases, protect the planet, explore space and teach every child — and getting those good things depends on being honest about the risks and choosing wisely.

What the 3D scene shows

A turntable with four small dioramas: a lab with a test tube and medicine capsules (curing diseases), a boat collecting plastic from a blue sea (cleaning the oceans), a rocket beside a ringed planet (exploring space), and a classroom with three pupils and a tiny Spark helping (helping teachers). Arrows rotate the turntable and a caption explains each scene.

Key words

• Benefit — a good result.
• Risk — a chance that something goes wrong.
• Alignment — making sure AI wants what we really mean.

Quick quiz

1. How is AI already helping with medicine?

   • ☐ It has cured every disease
   • ☐ It helps scientists understand proteins and design new medicines to test
   • ☐ It isn't used in medicine

2. Why is 'helping teachers' on the list instead of 'replacing teachers'?

   • ☐ Because teachers do human things a machine can't, and a patient helper gives them more time for those
   • ☐ Because robots can't read
   • ☐ Because schools don't have computers

3. What is the most important thing that decides whether superintelligence does good or harm?

   • ☐ How big the computers are
   • ☐ Whether it truly wants what people want, and who decides how it is used
   • ☐ Whether it has a friendly voice

Talk about it

Write your own "big idea in one sentence" for this lesson: