Imagine the best case
Every earlier lesson has been careful to say “nobody knows”. That is still true. But it is worth spending one lesson on the question: if we did build a superintelligence, and we built it well, what could it do?
Here are four answers. Each starts with something AI is already doing today, so you can see that this isn’t just dreaming.
1. Curing diseases
Today: AI already predicts the shapes of proteins — the tiny machines inside your cells — which used to take scientists years per protein. It helps design new medicines, spots illnesses in scans that human eyes miss, and finds patterns in medical records that point to causes nobody had noticed.
With superintelligence: Imagine a scientist who has read every medical paper ever written, remembers all of them perfectly, can test millions of ideas in simulation before breakfast, and never gets tired. Diseases that have puzzled doctors for centuries — some cancers, dementia, illnesses so rare that nobody studies them — might finally be understood.
The catch: Being clever is not the same as being right. New medicines would still need real-world testing on real people, which takes time. A superintelligence could speed up the thinking, not skip the checking.
2. Cleaning the oceans and protecting the planet
Today: AI helps track illegal fishing boats from satellites, predicts where plastic gathers in the sea, designs more efficient wind turbines and helps power grids use more clean energy.
With superintelligence: Cleaning the oceans isn’t one problem; it’s a million small ones — where the plastic is, which robots to send, how to stop more arriving, how to help coral recover. A mind that could hold the whole planet’s data in view and coordinate thousands of small helpers at once could make real progress on problems that are too big and tangled for any one team.
The catch: The planet belongs to everyone. Decisions about oceans, rivers and energy are also decisions about people’s jobs and homes. Those choices should be made by people, with AI as a very good adviser — not the other way round.
3. Exploring space
Today: Rovers on Mars already use AI to pick their own routes across rocks, because a radio signal takes up to twenty minutes to reach them from Earth. Telescopes use AI to sift millions of images for interesting objects.
With superintelligence: Space is enormous and slow. A trip to the nearest star would take longer than a human lifetime. Clever probes that could plan, repair themselves and decide what to study could go where people can’t, and report back what they find. Closer to home, a superintelligence might design spacecraft, materials and fuels that we haven’t thought of.
The catch: Space is exciting, but it’s also expensive, and it’s fair to ask whether it should come before problems here on Earth. A superintelligence could help with that decision — but it is our decision.
4. Helping teachers
Today: AI tutors can already explain a maths problem a second way if the first way didn’t click, practise a language with you at midnight, and help teachers with marking and planning.
With superintelligence: Imagine every child having a patient helper that knows exactly what they understand, what they’re confused about and what kind of example would help — and never runs out of time. Teachers would then have more time for the parts of teaching that are truly human: noticing a child is upset, running a debate, sparking excitement.
The catch: Notice the word helping. A screen is not a teacher. Children learn from people — from being seen, encouraged and challenged by someone who cares. The best future is not one where machines replace teachers, but one where teachers have a brilliant assistant.
Things to be careful about
A balanced page has to say this part clearly. The same power that could do the good things above could cause harm if it were built carelessly or used badly. Here are the main worries, without the scary music.
Misunderstanding what we want. The Wish Machine problem. A superintelligence that pursued a badly written goal very effectively could cause a mess far faster than we could fix it. This is why alignment research matters so much.
Who controls it. A tool this powerful in the hands of one company, one government or one person would give them enormous influence over everyone else. Many people argue that the benefits should be shared and the decisions made openly.
Jobs and fairness. If machines can do most kinds of work, how do people earn a living, and how do we make sure the rewards don’t all go to a few? These are questions for governments and citizens — including, one day, you.
Trust and truth. Machines that can write, speak and make pictures as well as any human make it harder to know what is real. We will need new habits, and probably new tools, for knowing who and what to believe.
Moving too fast. Almost every worry on this list gets worse if things happen faster than people can understand and adjust. That is why many experts argue for building carefully, step by step, with checks along the way.
The balance
So: could superintelligence cure diseases, clean the oceans, explore space and help every child learn? Possibly, yes. Could it also cause serious problems? Also yes.
Which one we get is not decided by the machine. It is decided by the people who build it, the rules we set, and how carefully we make sure it wants what we really want. That is a hopeful thought, because it means the outcome is up to us — and the more people who understand it, the better the choices will be.
You have now finished the learning path. If you want to go deeper, try the FAQ, look up any word in the glossary, or — if you are a parent or teacher — see how to use these lessons at home or in class.