O Levels and A Levels Were Built for a World That No Longer Exists

For decades, the path looked reassuringly simple. At fourteen or fifteen, choose your subjects. Memorise the syllabus. Practise past papers. Learn the marking scheme. Sit your O Levels. Choose three or four A Levels. Repeat. Get the grades. Enter university. Find a respectable career.

For millions of families, particularly across South Asia, the Middle East and other parts of the Commonwealth, this became more than an education system. It became a philosophy of childhood. A good child studied hard. A successful parent produced good grades. A prestigious school produced A*s. And an A* was treated almost as proof that the future would work out.

There was always something questionable about reducing a young person to letters on a certificate. But in the age of artificial intelligence, the problem has become much harder to ignore.

O Levels and A Levels are not useless qualifications. They remain internationally recognised, universities understand them, and many of their courses teach rigorous knowledge and genuine intellectual skills. The problem is bigger than whether the courses are good. The question is whether organising adolescence around them is still a sensible way to educate a human being. Increasingly, the answer is no.

We are preparing children for a world that disappeared while they were studying for it

The traditional examination system makes sense in a particular kind of economy: information is difficult to obtain, expertise is scarce, workers specialise early, institutions decide which knowledge matters, and employers need credentials to determine who knows what. Schools therefore transmit established bodies of information, examinations test whether students have mastered them, and universities select students using the resulting grades.

Artificial intelligence changes the equation. A teenager with an internet connection can now ask an AI system to explain calculus at six levels of difficulty, generate Python code, critique an essay, translate a document, analyse a spreadsheet, simulate a debate, produce illustrations, brainstorm business models or tutor them interactively.

This does not mean knowledge is obsolete. Quite the opposite. A person who knows nothing will struggle to tell whether AI is producing brilliance or nonsense. But it changes the valuable skill from 'Can you reproduce the answer?' toward 'Can you formulate the problem, interrogate the answer, verify it, improve it and do something useful with it?'

UNESCO's guidance on generative AI calls for a human-centred approach to AI in education and explicitly discusses redesigning teaching, learning, research and curriculum around these technologies rather than pretending they do not exist. That is a profound shift. Our examination culture has barely begun absorbing it.

The past-paper economy is an educational warning sign

Spend time around serious O-Level and A-Level students and an odd vocabulary quickly appears: 'Is this in the syllabus?' 'How many marks is this worth?' 'What does the examiner want?' 'Has this question appeared before?' 'What does the mark scheme say?'

These are rational questions. That is precisely the problem. When success depends on a bounded examination, intelligent students optimise for the examination.

Transparent assessment is better than mysterious assessment. But it inevitably creates an enormous parallel industry devoted not simply to learning mathematics, literature, physics or history, but to learning how the examination rewards particular responses.

A child can become extraordinarily good at school while remaining surprisingly inexperienced at dealing with problems for which there is no syllabus, no mark scheme, no past paper, no examiner — and sometimes not yet a correct answer.

Unfortunately, that is increasingly what adult life looks like.

AI makes the distinction between knowledge and intelligence impossible to ignore

Consider two seventeen-year-olds studying economics.

Student A memorises definitions, practises hundreds of examination questions, learns essay structures and receives an A*.

Student B understands the same foundations but also spends a year analysing local food prices, interviewing shopkeepers, learning basic statistics, building an inflation dashboard, using AI to analyse datasets, finding errors in the AI's conclusions, reading economists who disagree with one another, publishing her methodology and presenting her findings publicly.

Who is better educated?

The examination system has an excellent mechanism for measuring Student A. It has a much weaker mechanism for recognising Student B. Yet Student B is practising something much closer to the intellectual environment she will encounter as an adult: knowledge, judgment, technology, communication, uncertainty and initiative working together.

The World Economic Forum's Future of Jobs Report 2025 places analytical thinking at the top of its core-skills ranking, alongside resilience, flexibility and agility, leadership and social influence, creative thinking, motivation and self-awareness, technological literacy, empathy and active listening, and curiosity and lifelong learning. AI and big data also rank among important skills.

That list does not describe an examination machine. It describes a capable human being.

The biggest mistake is confusing academic rigour with educational completeness

Defenders of O and A Levels correctly point out that the curriculum is not simply rote memorisation. Cambridge O Level says its curriculum develops creative thinking, enquiry, problem-solving, practical skills and theoretical knowledge. Individual subjects can be intellectually impressive.

So the serious criticism cannot be 'O Levels teach nothing useful.' That is false.

The criticism is that the qualifications measure a relatively narrow portion of what a flourishing twenty-first-century human being needs, yet families often allow them to dominate an enormous portion of adolescence.

A system can contain excellent mathematics and still be educationally incomplete. A child can understand organic chemistry and know almost nothing about money. Write elegant essays and never cook dinner. Calculate differentiation while being unable to negotiate a disagreement. Recite the causes of a historical revolution but never organise a real project. Study computer science without ever shipping something another human uses. Earn eight A*s while having no idea what they enjoy doing when nobody is grading them.

Those are not failures of intelligence. They are failures of educational design.

Childhood has an opportunity cost

Every hour spent doing one thing is an hour not spent doing another. This sounds obvious until we apply it to childhood.

Imagine a fifteen-year-old spending three hours every evening preparing for exams. Over 200 school days, that is roughly 600 hours. Across several examination-intensive years, thousands of hours of adolescence can be directed toward optimisation for assessment.

Now consider what else 1,000 hours can do: a child could become a competent musician, build dozens of software projects, read one hundred books, become an excellent cook, learn filmmaking, start a small business, volunteer extensively, learn another language, spend hundreds of afternoons outdoors, train seriously in a sport, or study whatever strange subject happens to seize their imagination.

Or simply spend more time talking to grandparents, playing with siblings and figuring out who they are.

The question is not whether studying for examinations produces benefits. Of course it does. The meaningful question is whether those benefits are worth everything we give up to obtain them.

We specialise children remarkably early

At an age when many children barely know what professions exist, we ask them to begin closing doors. Choose your subjects. Science or humanities? Economics or biology? Computer science or art? Then A Levels narrow the funnel further.

Three or four subjects can become the intellectual universe of a sixteen-year-old. The implicit message is: decide what sort of person you are, then proceed down the appropriate educational corridor.

Real life rarely behaves like this. A founder may need psychology, accounting, coding, design and storytelling before lunch. A climate scientist may need statistics, politics and communication. A filmmaker may need history, technology and business. A doctor may work with AI systems. An artist may become an entrepreneur. An engineer may become a journalist.

New professions will emerge that today's fourteen-year-olds cannot select because they do not yet have names. The future favours intellectual mobility. Our educational instincts still favour intellectual lanes.

The deeper casualty may be curiosity

Young children are relentless researchers. Why is the moon following us? Where does money come from? Can fish sleep? Why did that empire disappear? What happens if I mix these? Can I build one? Why? Why? Why?

Then something strange happens. Over time, education can train children to replace 'What do I want to understand?' with 'Will this be on the exam?'

That may be one of the most expensive exchanges civilisation makes. Curiosity is not a decorative personality trait in the AI age. When existing knowledge becomes extraordinarily cheap to retrieve, advantage increasingly belongs to people who know what to investigate.

The student who asks unusual questions has somewhere to go. The student trained principally to provide expected answers may discover that machines are exceptionally good competitors.

Even mathematics exposes the problem

Mathematical knowledge matters enormously in an AI-powered world. But there is a difference between understanding mathematics and becoming anxious about performing mathematics on command.

PISA data show substantial mathematics anxiety among teenagers. Across OECD countries, many students report worrying about poor maths marks and failing, while large shares report nervousness or helplessness around mathematics tasks.

This is an extraordinary educational contradiction. We can spend thousands of hours teaching a subject that explains patterns throughout reality and leave many students frightened of it.

Something has gone wrong when learning becomes detached from the joy of being able to understand more of the world.

Happiness is not the opposite of achievement

The phrase 'happy children' sometimes makes ambitious parents uncomfortable. It can sound like an argument for lowering standards: let children play all day, avoid difficult things, never push them.

That would be another mistake.

A meaningful childhood should contain difficulty. Children should occasionally struggle. They should learn discipline. Finish things they want to abandon. Experience losing. Practise when they would rather stop. Read demanding books. Solve difficult equations. Take criticism. Discover that improvement usually requires repetition.

But difficulty needs a purpose beyond producing a letter on a transcript. There is a difference between productive struggle and permanent performance anxiety.

The goal should not be to raise children who are happy because nothing difficult is ever demanded of them. It should be to raise children whose competence makes them confident enough to confront difficult things.

What should replace the O-Level/A-Level mindset?

Not chaos. Not abandoning mathematics and hoping curiosity takes care of everything. Not putting an eleven-year-old in front of ChatGPT and calling it education. And certainly not eliminating foundational knowledge.

A serious alternative would be more rigorous in some ways than conventional schooling. But its unit of measurement would change.

Instead of asking primarily, 'What subjects has this child passed?' we would ask, 'What can this young person understand, create, communicate and contribute?'

1. Intellectual foundations

Every child still needs mathematics, science, history, literature, writing, geography, statistics, basic economics and civics. The foundations matter because AI cannot compensate for a person who lacks enough knowledge to interrogate what it produces. But subjects should become foundations for exploration rather than cages.

2. AI literacy

Children should understand that AI can be extraordinarily capable and confidently wrong. They should know how to use AI tools, construct useful prompts, verify claims, identify hallucinations, compare sources, recognise synthetic media, protect private information, understand algorithmic bias, automate repetitive work and decide when not to use AI.

The valuable student is no longer the one who can compete with a machine at information reproduction. It is the one who can supervise machines intelligently.

3. Making things

Every child should repeatedly turn ideas into reality: build a website, grow vegetables, produce a documentary, design furniture, write a book, create a game, make a robot, run an experiment, organise an event, start a tiny business, publish a magazine, repair something broken.

Projects expose children to the messy parts of competence that examinations struggle to reproduce. Things go wrong. Materials run out. People disagree. Budgets change. The first version fails. Instructions are unclear. Nobody knows the answer. Perfect. That is education.

4. Communication

AI makes mediocre communication cheap. That makes excellent human communication more valuable. Children should learn to speak publicly, write clearly, tell stories, listen carefully, persuade ethically, interview people, debate respectfully, teach something they understand and change their minds when evidence demands it.

5. Human relationships

Children need practice being human among humans: friendship, conflict, negotiation, teamwork, empathy, leadership, caring for younger children, speaking to elderly people, working with people unlike themselves, reading emotions, apologising and setting boundaries.

Perhaps the great irony of AI will be that machines make deeply human abilities more economically valuable, not less.

6. Physical competence

Children are not brains being transported between examinations. They have bodies. Education should include strength, coordination, sport, sleep, nutrition, outdoor exploration and practical physical capability. Climb. Swim. Run. Plant. Cook. Build. Walk through a forest. Fix a bicycle.

A childhood spent almost entirely sitting at desks in preparation for adulthood spent almost entirely sitting at screens is a peculiar definition of progress.

7. Financial and entrepreneurial literacy

A teenager can explain photosynthesis but graduate without understanding compound interest; know Shakespeare but not a payslip; calculate simultaneous equations but not compare a loan. That should be indefensible.

Children should encounter saving, investing, taxes, insurance, interest, contracts, pricing, sales, profit, debt, risk, negotiation and entrepreneurship. Not because every child should become an entrepreneur. Because every child will live inside an economy.

8. Self-knowledge

One of education's central jobs should be helping a child discover: What am I good at? What energises me? What drains me? Do I like working alone? Do I enjoy people? Do I like abstract ideas or physical objects? Do I want stability or adventure? What kind of problems am I willing to spend ten years solving?

Those questions are difficult to answer from an examination desk. They require experiences.

And yes, children should sometimes be bored

A holistic childhood does not require every minute to be an enrichment activity. One of its great advantages may be restoring unstructured time: time to wander, draw, build ridiculous things, read something with no academic value, argue with a sibling, kick a football, watch ants, invent games and think.

Being permanently scheduled is not the same thing as being well educated. Creativity often begins in the gaps adults forgot to organise.

So should families literally abandon O and A Levels?

For some children and in some countries, immediately abandoning recognised qualifications would create unnecessary barriers to university or professional pathways. Credentials still matter. Cambridge O Levels and A Levels remain widely recognised, and A-Level routes remain built explicitly around progression into higher education.

The sensible argument is not 'Burn the textbooks tomorrow.' It is: stop treating the qualification as the education.

If a teenager needs O Levels or A Levels as a passport into university, treat them as a passport. Pass them efficiently. Do not allow the passport application to consume the journey.

A student might spend part of the week mastering the academic material required for credentials and the rest building a much broader education around projects, AI, reading, sport, nature, entrepreneurship, art, community and relationships.

Where families have viable alternative routes — portfolio admissions, alternative qualifications, vocational pathways, dual enrolment, apprenticeships or universities with flexible admissions — the case for constructing something more ambitious becomes stronger still.

The portfolio may eventually matter more than the transcript

Imagine two eighteen-year-olds.

The first says: 'I received A*s in eight O Levels and three A Levels.' Excellent.

The second says: 'I studied mathematics, physics and history. I built three apps. One failed and one has 4,000 users. I interviewed twenty small businesses for a research project on automation. I speak conversational Arabic. I worked six months at an animal shelter. I can use Python and several AI systems. I produced a short documentary. I play tennis. I manage a small investment portfolio. Here is everything I built, the mistakes I made, the books I read and what I want to investigate next.'

Which young person sounds prepared for an uncertain world?

A transcript tells us what examinations someone passed. A portfolio begins to tell us who they have become.

Education should produce agency

This may be the word that matters most: agency.

The belief: I can encounter something I do not understand and learn it. I can have an idea and build it. I can discover that I am wrong and change direction. I can learn without waiting for a teacher. I can use technology without being controlled by it. I can work with other people. I can tolerate uncertainty. I can create value. I can construct a meaningful life.

AI makes this capacity enormously important because nobody can confidently describe the labour market a twelve-year-old will enter in ten years.

The rational strategy is not to prepare that child for one predicted future. It is to make them adaptable enough for many futures.

The goal was never supposed to be the exam

Perhaps the strangest thing about modern education is how easily the measurement system became the purpose of the system. We created examinations to measure learning. Then gradually shaped learning around examinations. We built universities to develop human capability. Then shaped childhood around university admission. We created credentials to indicate competence. Then began mistaking credentials for competence itself.

AI offers us an uncomfortable opportunity to reconsider all of this.

A generation ago, parents could reasonably believe that exceptional grades represented security. Today there is no certificate capable of guaranteeing that.

The safest child may instead be the one who is curious enough to keep learning, confident enough to begin, sceptical enough to question, resilient enough to fail, technically capable enough to use powerful tools, socially capable enough to work with humans and grounded enough to know that a career is only one part of a life.

That child should know mathematics. They should read great books. They should understand science and history. They should work hard. But they should also make things. Earn things. Break things. Repair things. Plant things. Question things. Travel when possible. Play. Compete. Help. Talk to adults. Care for younger children. Spend time outside. Fall in love with obscure subjects. Change their mind. Discover something they are terrible at. Discover something they cannot stop doing.

And have enough unscheduled childhood left to become a person rather than merely a candidate.

The real competition in the AI age will not be between the child who earned an A and the child who earned a B. It will be between people who were trained to follow pathways and people who learned how to create them.

O Levels and A Levels may remain useful credentials for some time. But as the organising philosophy of adolescence, their moment is passing.

The education system of the future should ask a far more ambitious question than 'What grade did this child get?'

What kind of human being are we helping this child become?