Education for the Next 30 Years: Schooling Is the Installer, Education Is the Spine
How do we prepare a child for artificial intelligence, an ageing world, growing global debt and jobs that may not yet exist?
AI, ageing populations and economic uncertainty will reshape the next 30 years. Discover why schooling is only the installer—and education must become a lifelong adaptive spine.
The Future Is Becoming Harder to See
A child entering Primary 1 today may still be working in the 2070s.
That is an extraordinary thought.
We are trying to prepare young people for a working life that may stretch across several technological eras, multiple economic cycles and occupations that do not yet have names.
Artificial intelligence is moving from a specialised tool into an everyday layer of work, communication, research and decision-making. The International Labour Organization estimates that approximately one in four jobs worldwide has some degree of exposure to transformation by generative AI. This does not mean that one in four jobs will simply disappear. It means that the tasks inside many jobs are likely to be reorganised, automated or redesigned.
At the same time, the world is carrying higher levels of public debt. The International Monetary Fund reported that global public debt rose to just under 94 per cent of global GDP in 2025 and projected that it could reach 100 per cent by 2029. Governments may have to manage rising interest costs alongside expenditure on healthcare, pensions, security, infrastructure and social support.
Populations are also ageing.
In Singapore, citizens aged 65 and above formed 20.7 per cent of the citizen population in 2025. By 2030, the figure is projected to reach approximately 23.9 per cent—close to one in every four citizens.
Globally, even younger and faster-growing societies are expected to age over the coming decades. Longer lives are an achievement, but they also change the balance between workers, dependants, caregivers, taxpayers and public expenditure.
These are question marks on the horizon:
- What happens when machines can perform more cognitive work?
- What happens when fewer working adults support larger elderly populations?
- What happens when governments have less financial room to absorb every disruption?
- What happens when students study for qualifications that no longer connect cleanly to future work?
- What happens when the knowledge taught at 15 becomes outdated by 25?
This is the negative slice of the future.
But it is not the whole future.
In fact, this uncertainty brings us back to the real purpose of education.
This Is What Education Is For
Education was never supposed to eliminate uncertainty.
It was supposed to prepare human beings to enter uncertainty without becoming helpless.
We cannot give children a map of every road they will travel over the next 30 years. Many of those roads have not been built.
What we can give them is an internal navigation system.
We can teach them how to read changing conditions, acquire new knowledge, recognise weak assumptions, make decisions, correct mistakes, work with others and rebuild themselves when an old route closes.
That is the deeper promise of education.
It does not promise that nothing will change.
It develops a person who can change without losing direction.
This is why we need to separate three words that are often treated as though they mean the same thing:
studying, schooling and education.
They are connected.
But they are not identical.
Studying, Schooling and Education Are Different Things
Studying Is an Activity
Studying is something a student does.
A student reads a chapter, memorises vocabulary, practises algebra, revises a Science topic or prepares for an examination.
Studying may last for twenty minutes, two hours or several years.
It is an activity directed towards acquiring or improving a particular piece of knowledge or skill.
Studying matters.
But studying alone is not yet education.
A student can study intensely, score well and still remain unable to transfer knowledge into a new situation. Another may memorise many answers without learning how to recognise when an answer is unreliable.
Studying describes the immediate work.
Education describes what that work gradually builds inside the person.
Schooling Is a Structured Environment
Schooling is the organised system around the student.
It provides teachers, timetables, classmates, subjects, curricula, examinations, routines, expectations, activities and social experiences.
School gives the child a place in which learning is structured rather than left entirely to chance.
This structure is important because young people do not automatically know:
- what they need to learn;
- what sequence they should learn it in;
- how much practice is required;
- how to distinguish understanding from familiarity;
- or how to persist when learning becomes uncomfortable.
Schooling provides the architecture through which these abilities can begin to form.
It is the installer.
But the installer is not the finished system.
A child does not attend school merely to become good at attending school.
A student does not learn Mathematics only to complete Mathematics worksheets. The student is learning to represent problems, recognise structures, follow logic, manage precision and remain calm when the answer is not immediately visible.
A student does not learn English only to produce an examination composition. The student is learning to interpret meaning, detect ambiguity, organise thought, communicate clearly and understand perspectives beyond his or her own.
Schooling is the training ground.
Education is what should remain when the school bell no longer rings.
Education Is the Portable System
Education is the part that travels with the person.
It is the internal structure that remains useful when:
- the textbook changes;
- the examination ends;
- the job disappears;
- a new technology arrives;
- the person enters a different culture;
- the economy weakens;
- or life presents a problem that was never included in the syllabus.
Education includes knowledge, but it is larger than stored knowledge.
It includes the ability to obtain new knowledge.
It includes the ability to question knowledge.
It includes judgment about when and how knowledge should be used.
It includes character, because capability without responsibility can become dangerous.
It includes agency, because a person who knows many things but cannot initiate action remains dependent on someone else to provide instructions.
It includes adaptability, because the future will not ask our children for permission before it changes.
This is the distinction we need:
Studying is the immediate activity. Schooling is the installation environment. Education is the lifelong system that must continue running.
Education Must Become the Invariant
An invariant is something that remains stable even while other parts of a system change.
Over the next 30 years, particular technologies will change.
Job titles will change.
Industries will rise and fall.
The software used in school today may look primitive to students in 2040.
Some facts will be updated. Some methods will be automated. Some professional boundaries will shift.
Therefore, education cannot depend entirely on teaching children one permanent collection of answers.
The content matters, but content alone cannot be the invariant.
The invariant must be the child’s growing capacity to:
learn, understand, verify, connect, apply, communicate, adapt and act responsibly.
That capacity becomes the education spine.
A healthy spine does not tell the body where to walk.
It allows the body to remain upright, balanced and capable of movement across many environments.
In the same way, education should not lock a child into one predicted future.
It should give the child enough structure to move through many possible futures.
The Subjects Are Training Grounds
Parents sometimes ask a reasonable question:
“Will my child actually use this topic in adult life?”
The honest answer is that the child may not directly use every topic.
Most adults will not reproduce every chemical equation, geometric proof or historical date they once learned.
But this does not make the learning pointless.
The deeper question is:
What human capability was the subject training?
| School experience | The education being installed |
|---|---|
| Mathematics | Logical structure, modelling, precision, pattern recognition and persistence |
| English | Interpretation, communication, inference, persuasion and clarity of thought |
| Science | Evidence, causation, testing, observation and disciplined scepticism |
| Humanities | Context, systems, culture, power, consequences and multiple perspectives |
| Art and music | Perception, expression, composition, taste and creative iteration |
| Physical education | Coordination, health, discipline, resilience and bodily awareness |
| Group projects | Cooperation, negotiation, leadership and responsibility |
| Co-curricular activities | Commitment, identity, teamwork, recovery and belonging |
| Examinations | Recall, execution, time management and performance under constraints |
| Corrections | Humility, diagnosis, adaptation and error repair |
This does not mean every worksheet is perfectly designed or every examination measures everything that matters.
It means the visible task and the deeper educational purpose should not be confused.
The worksheet is the equipment.
The capability is the result we are trying to build.
The Greatest Risk Is Not That AI Becomes Intelligent
The greatest educational risk is not simply that artificial intelligence becomes more capable.
It is that human beings become less capable because intelligence is available on demand.
A calculator can produce an answer without building number sense.
A search engine can locate information without developing understanding.
A generative AI system can produce a fluent paragraph without requiring the user to form a clear thought.
This creates a new educational problem.
For the first time, students may be able to produce the appearance of competence much more easily than competence itself.
A student can submit polished work while remaining uncertain about what the work means.
The danger is not merely cheating.
The deeper danger is cognitive outsourcing before cognitive development.
A young person who has already developed strong foundations can use AI as a multiplier.
That student can ask better questions, compare alternatives, challenge assumptions, test explanations and improve a piece of work.
But a young person without foundations may not know when the AI is wrong.
The student may accept an elegant error because it sounds confident.
The difference will not be access to AI. Access is likely to become increasingly common.
The difference will be the quality of the human mind directing it.
Singapore’s Ministry of Education is already integrating AI through an approach that places pedagogy first, keeps students at the centre and introduces age-appropriate safeguards. MOE’s broader direction also emphasises adaptive thinking, digital literacy and the development of future-ready learners rather than technology for its own sake.
This is the right starting direction.
AI should not replace the education spine.
It should connect to a spine that is already growing.
In an AI World, Foundations Become More Important
There is a temptation to believe that because AI can perform a task, human beings no longer need to understand that task.
Usually, the opposite is true.
When an answer can be generated instantly, the ability to evaluate the answer becomes more valuable.
When thousands of explanations are available, the ability to select a sound explanation becomes more important.
When content becomes abundant, attention becomes scarce.
When machines can imitate certainty, human judgment becomes essential.
Students will still need strong literacy because they must understand instructions, detect implication, interpret evidence and recognise manipulation.
They will still need Mathematics because the world will remain full of quantities, probabilities, trade-offs, models and risk.
They will still need Science because claims about health, technology, energy and the environment must be assessed through evidence rather than appearance.
They will still need history and the humanities because technology does not remove culture, power, identity, conflict or human motivation.
AI does not make foundational education obsolete.
It increases the cost of weak foundations.
The Coming Burden on the Young
We should discuss ageing populations carefully.
Older people are not a burden simply because they are older.
They carry experience, memory, relationships, skills and contributions that cannot be reduced to a fiscal calculation. Healthier ageing can also support longer working lives and continued participation in society.
But demographic structure still matters.
When the proportion of working-age adults becomes smaller relative to the population requiring healthcare, care work and retirement support, greater pressure can fall on the productive part of society.
Tomorrow’s young adults may need to:
- sustain higher productivity;
- care for parents and grandparents;
- fund their own longer retirements;
- navigate more expensive healthcare systems;
- adapt repeatedly to changing work;
- and contribute to societies facing tighter public budgets.
We should not tell children this to frighten them.
We should understand it so we can prepare them properly.
The answer is not to make childhood feel like an economic emergency.
The answer is to build more capable, healthy, adaptable and cooperative adults.
A frightened child is not automatically a prepared child.
More worksheets do not automatically create more resilience.
Pressure without architecture can simply create exhaustion.
The response must be better education, not merely more academic activity.
The Qualification-to-Future Disconnect
For much of modern educational history, the route appeared reasonably clear:
Study a recognised course.
Obtain a qualification.
Enter a related profession.
Develop experience.
Continue in that profession.
That route still exists, but it may become less stable.
A qualification will remain valuable, especially when it represents deep knowledge, disciplined practice and a recognised standard.
But qualifications may no longer guarantee a direct or permanent connection to one occupation.
Parts of a job may be automated.
New tools may redraw professional boundaries.
An industry may need fewer entry-level workers but more experienced workers capable of supervising automated systems.
A graduate may need to combine knowledge from several fields rather than remain inside one narrow discipline.
The International Labour Organization’s research on generative AI stresses job transformation more than a simple universal replacement of human work. That distinction matters. The future may not divide neatly into “jobs that survive” and “jobs that disappear”. Many jobs are likely to be reassembled around new combinations of human and machine capability.
This means schooling cannot be designed only as a conveyor belt towards a first job.
It must install the ability to re-enter learning.
A student’s first qualification should not represent the end of education.
It should demonstrate that the student has developed enough structure to continue educating himself or herself.
A Proposed MOE V3.0
The phrase MOE V3.0 is used here as an eduKateSG conceptual model. It is not the official name of an existing Ministry of Education policy.
It describes the next mental model we may need for education.
MOE V1.0: Access and Foundations
The first major purpose of mass education was to build widespread literacy, numeracy, civic participation and a shared base of knowledge.
Children needed access to schools.
Societies needed teachers, curricula and institutions.
This was education as national infrastructure.
MOE V2.0: Excellence, Pathways and Capability
The next stage expanded pathways, improved teaching quality, used technology, recognised different strengths and developed broader competencies.
Singapore’s current 21st Century Competencies framework includes adaptive and inventive thinking, communication, collaboration, civic literacy and social-emotional competencies.
MOE’s “Learn for Life” direction also reflects the recognition that education must be holistic and continue beyond narrow academic performance.
MOE V3.0: The Lifelong Adaptive Spine
MOE V3.0 would take the next conceptual step.
School would still teach subjects.
Students would still sit assessments.
Knowledge would still matter.
But the system would be understood more clearly as the installation stage of a lifelong education architecture.
The outcome would not only be a student who has completed school.
It would be a person with a functioning education spine.
That spine would contain seven connected structures.
1. Strong Foundations
Students need literacy, numeracy, scientific reasoning, cultural knowledge and the ability to communicate.
Foundations must be sufficiently deep that the student can operate without constant technological support.
AI can extend these foundations.
It must not become a substitute for never having built them.
2. A Learning Engine
Students need to understand how learning works inside themselves.
They should gradually recognise:
- what they understand;
- what they only recognise;
- what they have forgotten;
- where an error began;
- what kind of practice is required;
- and when they need assistance.
This is metacognition: the ability to observe and regulate one’s own thinking.
A student with a functioning learning engine does not merely say, “I am bad at Mathematics.”
The student can say:
“I understand the concept, but I lose accuracy during algebraic manipulation.”
That diagnosis creates a repair path.
3. Judgment
The future will provide young people with more information than they can possibly process.
Therefore, the scarce capability will not be information collection.
It will be judgment.
Students must learn to ask:
- Is this claim supported?
- What evidence would change my mind?
- What assumptions are hidden?
- Is the source reliable?
- What has been omitted?
- Who benefits from this interpretation?
- What are the second-order consequences?
- Am I seeing the problem clearly, or merely reacting quickly?
Judgment is what prevents intelligence from becoming easily manipulated.
4. Agency
Education should produce people who can initiate useful action.
Agency means a student does not remain permanently dependent on a teacher, parent, employer or machine to provide the next instruction.
The student can define a problem, decide on a first step, seek resources, monitor progress and revise the plan.
This does not mean children should be left alone.
Agency is developed through guided responsibility.
First, the adult models.
Then, the adult and child act together.
Eventually, the child becomes able to act independently.
5. AI Partnership
Students should learn how to work with intelligent systems without surrendering intellectual control.
They should know when AI is useful for:
- generating possibilities;
- explaining unfamiliar concepts;
- organising information;
- comparing approaches;
- rehearsing questions;
- providing feedback;
- and accelerating routine work.
They should also know when not to trust it.
Every student should learn that fluency is not proof, confidence is not accuracy and convenience is not understanding.
The future-ready student is not the one who refuses AI.
It is the one who can use AI while remaining intellectually present.
6. Human Cooperation and Responsibility
Many future problems will not be solved by isolated intelligence.
Ageing, public health, climate resilience, technological governance, social trust and economic inequality are coordination problems.
They require people who can listen, negotiate, contribute, disagree constructively and accept responsibility for shared outcomes.
Education therefore cannot be reduced to individual competition.
A society containing many high-scoring individuals can still fail if they cannot cooperate, trust one another or act beyond immediate self-interest.
The education spine must connect personal capability to social responsibility.
7. Renewal Across Life
Perhaps the most important future capability is the ability to begin again.
A person may need to learn a new field at 32.
Rebuild a career at 45.
Adopt unfamiliar technology at 57.
Care for family while retraining.
Move between employment, entrepreneurship and community responsibilities.
Education must therefore include the confidence to become a beginner repeatedly.
This is not a failure of the first education.
It is the purpose of the first education.
The schooling phase succeeds when it creates a person who can continue learning after schooling ends.
Education Is Not the Prediction of a Job
Parents naturally want security for their children.
They ask:
“What should my child become?”
It is a loving question.
But over the next 30 years, it may be too narrow.
A stronger set of questions would be:
- What can my child understand deeply?
- What can my child learn independently?
- Can my child communicate clearly?
- Can my child recognise weak evidence?
- Can my child work with both people and machines?
- Can my child recover from failure?
- Can my child enter an unfamiliar situation without becoming paralysed?
- Can my child carry responsibility?
- Can my child become useful in more than one environment?
A career is one expression of these capabilities.
It is not the whole architecture.
We should not prepare a child only to occupy one chair in the future.
We should prepare the child to remain capable even when the chairs are rearranged.
What Parents Can Do Now
Parents do not need to predict the technology of 2056.
They need to protect the development of the child’s education spine.
Do Not Confuse Completion with Understanding
A completed worksheet is not always a learned concept.
Ask the child to explain the reasoning, not merely show the answer.
A simple question such as “How did you know what to do?” can reveal whether the student has built a reusable method.
Allow Some Productive Difficulty
Children should receive support.
But support should not remove every moment of uncertainty.
If an adult or AI immediately supplies every answer, the child does not practise searching, testing and recovering.
Help the child remain inside the problem long enough to develop capability.
Treat Mistakes as Information
A mistake is not only a mark lost.
It shows where the student’s internal model diverged from the task.
Was the concept misunderstood?
Was a keyword missed?
Was the method unstable?
Did the student rush?
Did working memory become overloaded?
Correction is where education often becomes visible.
Protect Attention
In an environment of constant stimulation, sustained attention will become a serious competitive and personal advantage.
Reading, writing, solving and thinking without interruption may appear ordinary.
They are becoming less ordinary.
Let Children See Adults Learn
Parents do not need to pretend to know everything.
A child benefits from seeing an adult say:
“I do not know yet. Let us find out.”
That sentence models intellectual security.
It shows that not knowing is not shameful and that learning is not restricted to children.
What Students Can Do Now
Students may look at school and see chapters, tests and deadlines.
Try to see the training beneath them.
When you learn algebra, you are practising how to keep relationships balanced while transforming a problem.
When you write an essay, you are learning how to turn thoughts into a structure another mind can follow.
When you correct a Science answer, you are learning the difference between what you intended to say and what your evidence actually supports.
When you work with classmates, you are learning that other people do not automatically interpret the world in the same way you do.
When you prepare for an examination, you are learning how to perform when time and information are limited.
Not every lesson will feel inspiring.
Not every task will be perfectly designed.
But you can still ask:
What capability can I extract from this?
That question turns the student from a passenger into a participant.
Use school.
Do not merely pass through it.
Are Examinations Still Necessary?
Examinations will continue to have a role.
They provide common standards, reveal whether knowledge can be retrieved independently and test performance under constraints.
But an examination is a measurement window.
It is not the entire person.
The problem begins when the measurement becomes the purpose of education.
A student may achieve an excellent result and still need to develop initiative, judgment, communication or resilience.
Another may possess strong long-term potential that is not fully captured by one assessment format.
The intelligent response is not to dismiss examinations or worship them.
It is to use them accurately.
Prepare students seriously.
Interpret the results carefully.
Repair what the results reveal.
Then continue building the person.
Tuition Must Also Change
Tuition cannot respond to the next 30 years by becoming an even larger pile of worksheets.
That would increase academic activity without necessarily strengthening education.
Good tuition should identify what is preventing the student from moving forward.
Sometimes the problem is missing knowledge.
Sometimes it is weak language.
Sometimes it is an unstable method.
Sometimes it is poor pacing, low confidence, accumulated misconceptions or an inability to study independently.
The tutor’s role is not merely to supply answers.
It is to help the student install a more reliable process.
At eduKateSG, this means using English, Mathematics and Science as training grounds for clearer thinking.
We teach the subject because the subject matters.
But we also look beneath the subject:
- Can the student recognise the structure?
- Can the student explain the method?
- Can the student detect an error?
- Can the student connect one idea to another?
- Can the student remain calm when the question changes?
- Can the student eventually perform without the tutor?
The best outcome is not permanent dependency on tuition.
It is a student whose internal system has become stronger.
The Future Needs More Than Efficient Workers
There is one final danger in discussions about education and the future.
We may begin to speak about children only as future economic units.
We may ask how productive they will be, how much tax they will contribute, how well they will compete and whether they can support an ageing society.
These questions matter.
But they are incomplete.
Education must also prepare human beings to live meaningful lives.
A highly productive society can still be lonely.
A technologically advanced society can still be unwise.
An efficient person can still lack purpose.
The future needs capable workers, but it also needs good parents, neighbours, citizens, friends, creators, caregivers and leaders.
Education must preserve what is human while expanding what humans can do.
That is why values, relationships, culture, beauty, responsibility and belonging cannot be treated as decorative additions to the “real” curriculum.
They are part of the education spine.
They help determine what our intelligence is used for.
The Real Promise of MOE V3.0
The next version of education should not promise parents that it can predict the future.
It should promise something more honest and more valuable.
It should build children who can meet the future.
Schooling is the installer.
Subjects are the training grounds.
Teachers provide sequence, challenge, explanation and care.
Examinations provide partial measurements.
Parents provide security, values and continuity.
Communities provide culture and belonging.
Artificial intelligence becomes a tool and multiplier.
But education is the invariant running through all of them.
It is the spine that allows the person to remain upright while the environment moves.
Over the next 30 years, the world may become older, more indebted, more automated and more difficult to predict.
Our children may inherit problems that our generation could not completely solve.
But they will also inherit tools, knowledge and possibilities that no previous generation possessed.
We should not prepare them only to endure that world.
We should prepare them to understand it, improve it and create within it.
That is education.
Not merely studying.
Not merely schooling.
Not merely collecting certificates before life begins.
Education is the lifelong capacity to keep becoming capable.
And that may be the most important system we can install in any child.
What Is the Difference Between Education and Studying?
Studying and education are connected, but they are not the same thing.
Studying is the activity. Education is what the activity is supposed to build inside the person.
A student studies when he reads a chapter, memorises vocabulary, practises Mathematics, revises Science or prepares for an examination.
These are visible actions.
Education is less visible.
Education is what remains after the worksheet is completed, the test is over and the student no longer has a teacher standing beside him.
It is the ability to understand, reason, communicate, make judgments, correct mistakes, learn independently and adapt when the situation changes.
A student may study a great deal without becoming deeply educated.
He may memorise model answers without understanding the ideas.
He may complete hundreds of questions without knowing how to recognise a new version of the problem.
He may score well while still depending completely on instructions.
This means that more studying does not automatically produce more education.
The quality of the studying matters.
The purpose behind the studying matters.
And most importantly, what the student is becoming through the process matters.
Studying Is Temporary. Education Must Travel With the Student.
Studying usually has a short destination.
Finish the homework.
Prepare for the test.
Complete the syllabus.
Pass the examination.
Education has a longer destination.
It should travel with the student into Secondary School, post-secondary education, work, relationships, citizenship and adulthood.
When the exact question changes, education helps the student adapt.
When the technology changes, education helps the student learn again.
When instructions are incomplete, education helps the student decide what to do next.
When the student is wrong, education helps him examine the mistake rather than hide from it.
This is why education cannot be reduced to the amount of content covered.
Content is necessary, but content is not the final product.
The final product is a more capable human being.
Schooling Is the Installation Period
Schooling is the organised environment where education is first installed.
The timetable, subjects, teachers, examinations, classmates and routines form the training ground.
Mathematics is not only about obtaining the correct answer. It can train structure, precision, logic and persistence.
English is not only about completing comprehension passages. It can train interpretation, communication, inference and clarity of thought.
Science is not only about remembering keywords. It can train observation, evidence, causation and disciplined questioning.
Group work can train cooperation.
Corrections can train humility and repair.
Examinations can train recall, preparation and performance under limits.
The visible lesson is the subject.
The deeper lesson is the capability being built through it.
The problem begins when schooling becomes so focused on completing visible tasks that it forgets the invisible system it is supposed to install.
Why We Need to Rethink Teaching
Traditional teaching often begins with a simple model:
The teacher knows.
The student does not know.
The teacher delivers information.
The student remembers it.
The examination checks whether the information can be reproduced.
This model is useful for some forms of learning. Students still need direct instruction, explanation, demonstration, memory and practice.
But it is no longer enough.
The modern student is growing up in a world where information is abundant, answers can be generated instantly and artificial intelligence can produce work that looks intelligent.
The challenge is no longer only whether students can obtain an answer.
The challenge is whether they can understand, verify and use it.
Teaching must therefore move beyond the transfer of answers.
It must begin building the systems that allow students to operate when answers are uncertain.
From Answer Delivery to Capability Building
Instead of asking only:
“Can the student produce the correct answer?”
We should also ask:
“Can the student explain why it is correct?”
“Can the student recognise when the method should be used?”
“Can the student detect a wrong answer?”
“Can the student transfer the idea into a new situation?”
“Can the student continue without constant help?”
This changes the teacher’s role.
The teacher is no longer only a distributor of knowledge.
The teacher becomes a designer of learning, a model of thinking and a guide who gradually transfers responsibility to the student.
The purpose is not to make the teacher unnecessary.
The purpose is to ensure that the student becomes increasingly capable.
From Completing Work to Understanding Work
A student can complete ten pages without learning much.
Another student may complete three questions and develop a much stronger understanding because he examined his method carefully.
Teaching should therefore pay more attention to the quality of thought behind the work.
Why did the student choose this method?
Where did the mistake begin?
Was the problem conceptual, procedural or caused by rushing?
Can the student describe the difference between what he knows and what he is still guessing?
These questions convert mistakes into information.
They help students develop metacognition: the ability to observe and regulate their own thinking.
This may become one of the most important forms of education in the future.
A student who can diagnose his own learning can continue improving long after formal schooling ends.
From Dependency to Guided Independence
Good teaching should support students without creating permanent dependency.
At the beginning, the teacher may demonstrate every step.
Later, the teacher asks the student to complete part of the process.
Eventually, the student should be able to identify the problem, choose an approach, monitor his work and correct himself.
This progression matters.
Too little support creates confusion.
Too much support prevents independence.
The goal is not to remove difficulty.
The goal is to make difficulty productive.
Students need enough challenge to grow, but enough structure to remain engaged.
Teaching should not carry the student forever.
It should gradually strengthen the student’s own legs.
From Memorising Information to Building Judgment
In the past, access to information was limited.
Today, information is everywhere.
The modern problem is deciding what to trust.
Students must learn to ask:
Is this source reliable?
What evidence supports the claim?
What assumptions are hidden?
Does the answer make sense?
What information is missing?
Could there be another explanation?
This becomes especially important when students use artificial intelligence.
AI can produce confident and polished responses, but confidence is not proof.
A student with weak foundations may accept a convincing error.
A student with strong foundations can question, compare, test and improve the output.
This means teaching must develop intellectual resistance.
Students should not become suspicious of everything.
They should become capable of examining what they are given.
From Teaching for One Examination to Teaching for Repeated Renewal
Examinations still matter.
They provide standards, deadlines and evidence of performance.
But teaching cannot stop at examination readiness.
A student may change jobs several times, enter industries that do not yet exist and need to learn unfamiliar technologies throughout adulthood.
The most important outcome of school may therefore be the ability to return to learning.
Can the student become a beginner again?
Can he manage frustration?
Can he break a large problem into smaller parts?
Can he identify what he needs to learn next?
Can he build competence without waiting for a formal classroom?
Teaching should prepare students not only to graduate, but to renew themselves repeatedly.
What Teaching Should Begin to Look Like
The future of teaching is not about abandoning knowledge, teachers or examinations.
It is about placing them inside a larger purpose.
Students still need strong foundations.
They still need explicit teaching.
They still need practice, correction and disciplined effort.
But every subject should also help install a deeper system.
Teaching should develop:
- knowledge that can be used, not merely recalled;
- understanding that can transfer into new situations;
- attention strong enough to resist distraction;
- judgment strong enough to question weak answers;
- language strong enough to express thought clearly;
- resilience strong enough to remain inside a difficult problem;
- and independence strong enough to continue learning without constant supervision.
This is the shift.
We should stop thinking of education as the delivery of a finished package.
Education is the construction of a person who can keep developing.
A Better Question for Every Lesson
Instead of asking only:
“What content are we covering today?”
Teachers, tutors and parents can also ask:
“What capability are we building through this lesson?”
The content may be fractions.
The capability may be proportional reasoning.
The content may be a comprehension passage.
The capability may be inference.
The content may be an experiment.
The capability may be testing a claim against evidence.
The content may be a correction.
The capability may be learning how to recover from error.
Once we see this, teaching becomes more than syllabus coverage.
Every lesson becomes part of the student’s internal architecture.
Education Is What Remains
Studying helps the student meet the next task.
Education helps the student meet the next stage of life.
Studying may produce an answer.
Education produces the person who knows what to do when the answer is not obvious.
That is why we need to rethink teaching.
The future will not reward only those who can remember what they were taught.
It will increasingly reward those who can learn, unlearn, relearn, judge, communicate and adapt.
School should therefore be more than a place where students collect information.
It should be the place where they begin installing the system that will carry them through the rest of their lives.
Studying is part of education.
But education is the larger process.
It is not merely what a student knows today.
It is the growing ability to become capable again tomorrow.
The Next Part of Education: Career and the Extraction of Capital in the Future Economy
Finding work when artificial intelligence, robotics and demogrle.
Career is where that capability meets the economy.
This is the next part of the education question.
It is not enough to ask:
What should our children study?
We must also ask:
How will their knowledge, abilities and judgment connect to economic value in the world they eventually enter?
For much of the previous century, the connection appeared relatively straightforward.
A student studied.
The student obtained a qualification.
The qualification opened the door to a recognised occupation.
The employer provided the equipment, customers, systems and capital.
The worker provided time and expertise.
In return, the worker received a salary.
That arrangement will not disappear.
But it may no longer be the only—or even the dominant—way that many people organise their working lives.
Artificial intelligence can now perform parts of cognitive work.
Robotics can perform a growing range of physical work.
Digital platforms can connect individuals directly to customers.
A small team can use technology that was once available only to a large company.
At the same time, many countries are ageing, birth rates are falling and the proportion of working-age people is declining.
These changes may meet at one important junction.
We can call it the demographic–automation nexus.
It is the point at which societies need more productive capacity, but have fewer available human workers—and machines become increasingly capable of supplying part of that missing capacity.
That meeting point may reshape what a job is, what a career is and how people earn a share of the wealth produced by the economy.
The World Is Not Simply Losing Population
We should begin with an important distinction.
The entire global population is not currently falling.
The United Nations projects that the world population will continue growing for several more decades, reaching a peak of approximately 10.3 billion in the mid-2080s before gradually declining. However, many countries are already experiencing—or are approaching—population peaks, declining birth rates, ageing populations and shrinking shares of working-age people. because economies are not powered by population numbers alone.
They are affected by the relationship between:
- children;
- working-age adults;
- older people;
- productive capacity;
- care requirements;
- consumption;
- taxation;
- and public expenditure.
Singapore illustrates the shift clearly.
In 2025, 59.8 per cent of Singapore citizens were aged between 20 and 64, down from 64.5 per cent in 2015. Citizens aged 65 and above formed 20.7 per cent of the citizen population, and that proportion is projected to reach approximately 23.9 per cent by 2030—close to one in four citizens. mean older people cease to contribute.
Many can and will remain economically, socially and intellectually active for longer. Better health, more flexible work and redesigned workplaces can extend productive working lives. IMF research suggests that improved labour-market outcomes among people aged 50 and above could meaningfully offset some of the economic pressure created by ageing. ll structural challenge remains.
There may be fewer young workers entering some economies precisely when the need for healthcare, eldercare, infrastructure maintenance and public support is increasing.
This creates pressure to produce more with each available worker.
And that is where AI and robotics arrive.
AI and Robotics Are Coming Online at the Same Time
Artificial intelligence affects cognitive tasks.
It can summarise, translate, classify, generate, compare, calculate, search, analyse and assist with decisions.
Robotics affects physical tasks.
Robots can move materials, inspect products, transport goods, assist with medical procedures, clean environments and perform repetitive or hazardous work.
These technologies are not advancing at exactly the same speed, and their effects will vary considerably across industries.
But both are becoming part of the productive infrastructure of the economy.
The International Labour Organization estimates that one in four workers worldwide is in an occupation with some exposure to generative AI. However, its research concludes that job transformation is more likely than the complete replacement of most occupations. Tasks within jobs are more likely to be reorganised than every exposed job simply disappearing. lso expanding.
The International Federation of Robotics reported that 542,000 industrial robots were installed globally in 2024, more than twice the annual number installed ten years earlier. Robotics is increasingly being presented not only as a method of reducing costs, but also as a response to labour shortages, dangerous work and the need for greater operational resilience. inistry of Manpower has similarly described AI as a potential response to structural constraints such as an ageing workforce and persistent labour tightness, although adoption remains uneven across companies. ore not looking at one isolated change.
We are looking at several forces approaching one another:
Fewer available workers in ageing economies.
More demand for care, services and productive output.
Greater cognitive capability from AI.
Greater physical capability from robotics.
More capital required to purchase, train and operate these systems.
Where these forces meet, the economic structure may begin to change.
The Demographic–Automation Nexus
In the past, automation was often described mainly as a threat to workers.
A company purchased a machine because it wanted to reduce labour costs.
That still happens.
But in an ageing economy, the purpose of automation can change.
A company may automate because it cannot find enough workers.
A hospital may introduce technology because care demand is growing faster than the available workforce.
A logistics company may use autonomous systems because the volume of goods is increasing while the pool of drivers and warehouse workers is tightening.
A manufacturer may use robots because continuity, safety and precision cannot depend entirely on an increasingly scarce labour supply.
This creates an unusual economic possibility.
The same technology can be:
- a substitute for some human tasks;
- a complement to other human tasks;
- a response to labour shortages;
- a source of higher productivity;
- and a source of greater inequality.
Whether workers benefit depends partly on where they stand in relation to the technology.
Do they compete against it?
Do they operate it?
Do they improve it?
Do they own part of it?
Do they use it to multiply their own output?
Or do they remain outside the productive system entirely?
This is why the future of career cannot be understood only as a search for a job title.
It must also be understood as a search for a position within the productive system.
A Job Is Not the Same as a Career
A job is an agreement.
A person performs a defined role for an organisation and receives compensation.
A career is larger.
A career is the evolving system through which a person’s capabilities meet changing economic demand over time.
A job may last two years.
A career may last 40 or 50 years.
The job may disappear while the underlying capabilities remain valuable.
The company may close while the person’s relationships, reputation and knowledge continue.
The technology may change while the problem being solved remains.
Therefore, students should not be educated only to fit one job description.
They should learn to recognise the deeper economic structure beneath a job:
- What problem is being solved?
- Who needs the solution?
- Why are they willing to pay for it?
- What human capabilities are required?
- Which tasks can machines perform?
- Which tasks still require human judgment?
- Who owns the tools?
- Who controls access to the customers?
- Who receives the value created?
These questions reveal the real shape of a career.
What Does “Extraction of Capital” Mean?
The phrase extraction of capital can sound as though a person is taking something away from someone else.
That is not the intended meaning here.
We are talking about how a person secures a legitimate economic claim on the value that he or she helps create.
A clearer term may be value capture.
A person creates value by solving a problem, satisfying a need, reducing risk, improving a system or producing something useful.
Value capture is how part of that created value returns to the person.
For example:
- An employee captures value through a salary.
- A specialist may capture value through professional fees.
- A salesperson may receive commission.
- An author may receive royalties.
- A business owner may receive profit.
- A shareholder may receive dividends or capital appreciation.
- A creator may license intellectual property.
- A software developer may own a product that can be sold repeatedly.
- A teacher may develop a curriculum or learning system that can serve many students.
- A technician may combine human expertise with machines to perform work that neither could accomplish as effectively alone.
This is not merely about becoming rich.
It is about understanding how capability becomes livelihood.
It is also about whether a person remains economically dependent on selling each hour once—or gradually develops assets, systems and ownership that can continue producing value.
The Old Model: Sell Time Into an Existing Machine
The traditional employment model works like this:
- The company owns the capital.
- The company owns or rents the premises.
- The company purchases the equipment.
- The company develops the customer relationships.
- The company organises the productive system.
- The worker enters the system and contributes labour.
- The worker receives wages.
This arrangement can be highly effective.
The employee does not need to build the entire organisation before earning an income.
The company absorbs risk, provides structure and coordinates many people.
There is nothing inherently wrong with employment.
But employees should understand the exchange.
They are placing their human capital—time, knowledge, judgment, energy and relationships—inside a system predominantly owned by someone else.
Their income depends on how valuable their contribution is, how scarce their capabilities are, how easily they can be replaced and how much bargaining power they possess.
AI and robotics may alter all four conditions.
The New Productive Stack
Future economic output may increasingly be produced by a connected stack:
Human capability
plus
Artificial intelligence
plus
Robotics and physical systems
plus
Data
plus
Energy and computing infrastructure
plus
Financial capital
plus
distribution and customer access
plus
trust, regulation and social legitimacy.
The person who understands only one small task inside this stack may be vulnerable if that task becomes automated.
The person who can connect several layers may become more valuable.
The person who owns or controls part of the stack may capture a larger share of the economic value it creates.
This does not mean every child must become a founder or technology entrepreneur.
It means every child should eventually understand where value flows.
Education should help students see the system they are entering rather than preparing them to stand blindly inside one small part of it.
The Future May Separate Work From Income More Sharply
In the industrial economy, workers and machines were often easy to distinguish.
The worker performed the work.
The machine was a tool.
In the AI economy, the boundary may become less clear.
A person may ask an AI agent to research, write, code, analyse and communicate.
A designer may produce in one day what once required a small team.
A small company may coordinate hundreds of automated processes.
A technician may supervise a fleet of machines rather than operate one machine directly.
A professional may spend less time producing first drafts and more time defining problems, checking outputs and accepting responsibility.
The economic question then becomes:
When a human and a machine jointly produce the output, who receives the income?
The answer depends on ownership, contracts, market power and scarcity.
IMF research warns that AI could increase wealth inequality when productivity gains raise returns to capital, particularly if ownership of AI systems is concentrated. AI may affect wage inequality differently from wealth inequality: some workers may benefit from higher productivity, while the owners of productive capital may capture a substantial share of the gains. areer education must go beyond employability.
Young people need to understand both labour and capital.
Labour and Capital
Labour is what a person does.
Capital is what a person or organisation owns or controls that helps produce further value.
Capital can include:
- money;
- equipment;
- property;
- software;
- intellectual property;
- business ownership;
- data;
- a recognised brand;
- distribution systems;
- trusted customer relationships;
- and productive technology.
Education itself creates a form of capital.
Knowledge, judgment, communication ability and professional competence form human capital.
But human capital is tied to the person.
It usually has to be activated through work.
A skilled surgeon cannot perform unlimited operations simultaneously.
A skilled teacher cannot personally teach an unlimited number of students at the same time.
Technology can introduce leverage.
A surgeon may use robotic assistance.
A teacher may develop digital resources, diagnostic systems or curricula.
A designer may use AI to generate and test more possibilities.
A small company may automate administration and serve more customers.
Leverage allows the same unit of human capability to influence a larger amount of output.
In the next economy, career resilience may depend partly on whether a person can combine:
human capital + machine leverage + some form of ownership or bargaining power.
Five Ways People May Earn in the Next Economy
Most people will probably use a combination of several models.
1. Selling Human Time
The person works for an hourly wage, monthly salary or professional fee.
This remains important because many tasks require presence, responsibility, judgment and coordination.
But time is finite.
Income generally stops when the work stops.
2. Selling Scarce Expertise
A person with rare, difficult or trusted capabilities can command higher compensation.
This may include specialised technical knowledge, complex decision-making, leadership, negotiation, diagnosis or responsibility for high-stakes outcomes.
The value lies not merely in knowing information, but in being trusted to apply it correctly.
3. Operating Productive Technology
A person may use AI, robotics or automated systems to produce more than an unaided worker could.
The worker is not replaced by the machine.
The worker becomes the operator, integrator, verifier or orchestrator of the machine.
4. Building Reusable Assets
The person creates something that can generate value more than once.
This may include:
- software;
- designs;
- patents;
- educational systems;
- media;
- research;
- processes;
- brands;
- databases;
- or intellectual property.
The work is performed once or periodically, but the asset may continue producing value.
5. Owning Part of the Productive System
The person may own a business, shares, equipment, intellectual property or part of a platform.
This introduces risk.
Ownership does not guarantee success.
But it can provide a claim on the productivity of a system beyond the direct sale of personal hours.
These are not mutually exclusive.
A future engineer may receive a salary, develop rare expertise, operate AI systems, own intellectual property and invest part of the income earned.
A teacher may teach classes, build a curriculum, create diagnostic tools and own part of the educational organisation.
A nurse may provide direct care, supervise care technology, train others and contribute to the design of better care systems.
Career may become a portfolio of value-producing positions rather than one fixed occupational identity.
Where Might Human Work Remain Valuable?
No one can produce a perfectly reliable list of “safe jobs”.
A job that looks safe today may change.
A job that appears highly exposed to AI may become more productive rather than disappear.
The ILO’s analysis emphasises that task transformation is generally more likely than complete occupational replacement. approach is to identify the conditions under which human contribution remains important.
When the Physical World Resists Simplification
Buildings, electrical systems, machines, bodies, roads and natural environments are complicated.
They contain irregularities that do not always appear in clean digital data.
Work involving installation, maintenance, inspection, repair and physical adaptation may remain important even as robotics grows.
The role may change from performing every task manually to directing, maintaining or working alongside machines.
When Responsibility Cannot Be Easily Delegated
A machine can make a recommendation.
But society may still require a human being to take responsibility for the decision.
Medicine, law, engineering, finance, public administration and safety-critical industries may use more AI while retaining human accountability.
When Trust Matters
People do not always want only the technically correct answer.
They want to know whether the person advising them understands their circumstances and will remain responsible for the outcome.
Trust matters in education, healthcare, leadership, counselling, negotiation, management and care.
When the Problem Is Not Clearly Defined
AI works well when given a clear task and sufficient information.
But many important problems begin in confusion.
Someone must determine:
- what the real problem is;
- which goals matter;
- whose interests are affected;
- what constraints exist;
- and what a good outcome would look like.
Problem definition may become more valuable as answer generation becomes cheaper.
When Human Preference Creates the Value
People care about stories, identity, taste, belonging and culture.
A technically perfect product may still fail if it does not understand the people it is meant to serve.
Creative and cultural work will also use AI, but human intention and social meaning may remain central.
When Care Is the Product
An ageing population may increase demand for healthcare, eldercare, rehabilitation, community support and redesigned living environments.
SkillsFuture Singapore’s 2025 future-economy work identifies the Care, Digital and Green economies as important areas of changing job and skills demand. more technology.
But care is not only the completion of a task.
It also involves dignity, reassurance, observation, trust and human presence.
How to Find a Job in the Next Economy
Students are often told to follow their passion or choose a promising industry.
Both can be useful.
Neither is enough.
A stronger approach is to work through five questions.
1. Where Is the Persistent Human Need?
Do not begin with the job title.
Begin with the need.
People will continue to need:
- food;
- energy;
- shelter;
- health;
- care;
- education;
- transport;
- safety;
- communication;
- financial coordination;
- trustworthy information;
- entertainment;
- belonging;
- and functioning institutions.
The tools used to meet these needs will change.
The underlying needs are more persistent.
2. Which Parts Can Machines Perform?
Break the occupation into tasks.
Which tasks involve:
- routine information processing;
- standardised physical movement;
- predictable communication;
- pattern recognition;
- calculation;
- documentation;
- scheduling;
- or basic content generation?
These are more likely to be automated or accelerated.
3. Which Parts Become More Valuable After Automation?
When routine output becomes cheaper, other capabilities may become more valuable:
- judgment;
- problem definition;
- verification;
- accountability;
- relationship-building;
- coordination;
- physical adaptation;
- creativity;
- and strategic direction.
The goal is not to hide from technology.
It is to move towards the parts of the system where human contribution remains scarce and consequential.
4. How Can Technology Multiply the Person?
A student should eventually learn to ask:
What can I accomplish with AI and machines that I could not accomplish alone?
The value may no longer lie in competing against the machine at the task it performs best.
It may lie in directing the machine towards a useful human outcome.
5. How Will I Capture Part of the Value?
This is the question traditional career education often neglects.
Will the person receive:
- a salary;
- a fee;
- commission;
- royalties;
- profit;
- equity;
- ownership;
- licensing income;
- or some combination?
The person does not need to maximise every dollar.
But he or she should understand the economic agreement.
Creating value and capturing value are related, but they are not identical.
A person can create enormous value while receiving very little of it.
Another can control the customer relationship, platform or productive asset and receive a much larger share.
Economic literacy helps young people see the difference.
The New Career Formula
A possible formula for the next economy is:
**Persistent problem
- strong human capability
- AI or robotic leverage
- trust and responsibility
- a legitimate claim on the value created
= career resilience**
No element guarantees success.
But together they provide a stronger position than simply memorising the requirements of one current job description.
What Education Must Now Install
If this economic change is coming, education must respond before students enter the labour market.
Teach Students to See Jobs as Systems
Students should learn to look beneath occupational titles.
A doctor is part of a healthcare system.
A teacher is part of an education system.
An engineer is part of a productive and infrastructural system.
A designer is part of a system connecting human needs, production and communication.
Seeing the system allows students to understand where technology enters, where value is created and where responsibility remains.
Teach Task Decomposition
Students should be able to take a complex activity and divide it into parts.
What requires memory?
What requires calculation?
What requires judgment?
What can be automated?
What must be checked?
What requires a human relationship?
This is useful for both learning and career planning.
Teach AI Collaboration Without Intellectual Surrender
Students must become capable of using AI to extend their work.
But they must retain enough knowledge to evaluate the result.
The student should be able to say:
- what the machine was asked to do;
- why the output may be useful;
- what could be wrong;
- what must be verified;
- and who remains responsible.
Teach Economic and Capital Literacy
Students should understand:
- wages;
- revenue;
- cost;
- profit;
- risk;
- debt;
- equity;
- ownership;
- intellectual property;
- taxation;
- compounding;
- and opportunity cost.
This is not about turning every child into a trader.
It is about helping young people understand the system in which their work will be exchanged.
Teach Students to Build
Students should create things that have to work outside the classroom.
A product.
A service.
A research project.
A piece of software.
An event.
A publication.
A community initiative.
A lesson.
A prototype.
Building teaches the connection between ideas, execution, users, constraints and consequences.
Teach Career Renewal
The first qualification will not be the final education.
Students need to know how to re-enter learning, transfer skills and become beginners again.
Research on ageing and AI suggests that the transferability of skills between occupations will become increasingly important as workers move through changing labour markets. spine must therefore support repeated career reconstruction.
What Parents Should Begin Asking
Parents naturally ask:
“What job should my child choose?”
That question can now be expanded.
Ask:
- What important problems can my child learn to solve?
- Which capabilities are becoming more valuable?
- Can my child work effectively with AI?
- Can my child verify machine-generated output?
- Can my child communicate with other people?
- Can my child build trust?
- Can my child understand how an organisation creates value?
- Can my child develop something reusable?
- Can my child adapt when one task becomes automated?
- Can my child recognise the difference between income and ownership?
- Can my child continue learning without returning to zero every time?
These questions do not replace academic achievement.
They explain what academic achievement must eventually support.
What Students Should Begin Seeing
Students often see school as a sequence of subjects.
But beneath those subjects, they are building parts of their future economic capability.
English helps a person explain, persuade, negotiate and understand.
Mathematics helps a person model, measure, compare and reason.
Science helps a person test claims against the physical world.
Humanities help a person understand institutions, societies, cultures and consequences.
Art helps a person develop perception, taste and expression.
Group work develops coordination.
Examinations develop execution under constraints.
Corrections develop error detection.
These capabilities can later be combined in ways that no single school subject can predict.
A student does not need to know the exact job that will exist in 2045.
But the student should be building the ability to recognise where useful work is emerging.
Do Not Prepare Children Only to Compete With Machines
A child who is trained only to perform routine tasks faster may eventually meet a machine that is faster still.
That is a poor long-term strategy.
The goal is not to make children into inferior versions of computers.
It is to develop the capabilities needed to direct powerful systems towards worthwhile outcomes.
Humans still decide:
- what matters;
- which problems deserve attention;
- what trade-offs are acceptable;
- who should benefit;
- what risks society will tolerate;
- and what kind of future should be built.
Machines may increase productive capacity.
They do not automatically provide a moral direction for that capacity.
That remains an educational task.
The Nexus Is Both a Risk and an Opportunity
The demographic–automation nexus contains a serious risk.
If AI and robotics produce more wealth while ownership remains concentrated, the returns may flow disproportionately towards the owners of capital.
Some workers may become more productive.
Others may lose bargaining power.
The economy may become more efficient while the distribution of security becomes less equal. IMF research has repeatedly highlighted the possibility that automation and AI could increase returns to capital and widen wealth inequality. nexus contains an opportunity.
AI and robotics may help societies:
- maintain output with fewer workers;
- reduce dangerous and exhausting work;
- support older workers;
- extend healthcare capacity;
- create smaller and more capable companies;
- improve access to expertise;
- and allow individuals to build productive systems that once required large organisations.
The outcome is not predetermined.
It depends on education, ownership, institutions, regulation and the choices societies make.
The Next Purpose of Career Education
Career education used to help a student choose an occupation.
The next version must do more.
It must help the student understand:
- how human needs create economic demand;
- how organisations convert resources into value;
- how AI and robotics change the task structure;
- how capabilities move between occupations;
- how income is negotiated;
- how ownership affects the distribution of gains;
- how productive assets are built;
- and how a person can remain useful through repeated technological change.
The student should not leave school knowing only how to apply for a job.
The student should understand what a job is.
A job is one interface between a person and the economy.
It is not the whole economy.
It is not the whole career.
And it is no longer enough to prepare a child only to be selected by someone else.
Education should also help the child become capable of creating, organising and owning something useful.
Education, Career and Capital
Education builds capability.
Career connects that capability to human need.
Technology multiplies the capability.
Capital organises the productive system.
Ownership determines who receives part of the resulting value.
These are connected.
A child may one day work inside a company.
Lead a team.
Operate machines.
Build an automated service.
Care for an ageing population.
Create intellectual property.
Start an organisation.
Move between several occupations.
Or combine all these roles across a long working life.
We cannot predict the precise route.
But we can prepare the architecture.
The future economy may have fewer young workers and more intelligent machines.
That does not automatically make human beings unnecessary.
It changes the question.
The important question is no longer simply:
Where can my child find a job?
It becomes:
Where can my child apply human capability, use technological leverage, solve an important problem and secure a fair claim on the value created?
That is the next part of education.
Not merely studying for a qualification.
Not merely entering a profession.
But understanding how to remain useful, productive, adaptive and economically secure inside a changing system.
The future career may not be one ladder.
It may be a moving network of problems, capabilities, machines, relationships and ownership.
Education must give young people the spine to move through that network without losing their balance.
And the intelligence to recognise where the next real value is being created.
The Race: When the Price of Intelligence Falls
Why human beings must stop converging with machines—and begin developing the differences that machines make more valuable
For most of human history, intelligence was expensive.
Not because intelligent people were rare, but because intelligence was difficult to preserve, distribute and apply at scale.
Knowledge had to travel through a human being.
An experienced craftsperson taught an apprentice.
A scholar copied a manuscript.
A teacher stood before a small group of students.
A physician carried knowledge accumulated through years of training and experience.
A ruler depended on a limited number of literate advisers.
Intelligence existed, but access to it was constrained by distance, time, wealth, language and social position.
A book illustrates this history.
Before mechanical printing, books had to be copied by hand. They were costly objects, available mainly to religious institutions, wealthy patrons and small scholarly communities.
The printing press changed the economics of knowledge.
Books became easier to reproduce. Production increased, and the cost of books relative to income fell sharply. Knowledge that had once been locked inside a few manuscripts could travel across cities, countries and generations.
The printed book lowered the price of stored intelligence.
Mass schooling lowered the price of access to organised intelligence.
The internet lowered the price of finding and transmitting information.
Artificial intelligence is beginning to lower the price of performing parts of cognitive work.
That is a different kind of change.
From Expensive Knowledge to Cheap Cognition
A book can preserve an explanation.
The internet can locate an explanation.
AI can increasingly produce, adapt, compare and apply explanations.
It can summarise a report.
Translate a document.
Generate a first draft.
Analyse patterns.
Write computer code.
Compare alternatives.
Construct a lesson.
Answer questions.
Simulate a conversation.
Organise a plan.
This does not mean AI possesses every form of human intelligence or that its answers are always reliable.
It means that some useful cognitive operations can now be purchased at rapidly falling cost.
Stanford’s 2025 AI Index reported that the inference cost of obtaining performance around the level of GPT-3.5 fell more than 280-fold between November 2022 and October 2024. Smaller models also became substantially more capable.
The exact numbers will continue to change.
The larger direction is more important:
A unit of usable machine intelligence is becoming cheaper.
What once required a trained person, a large organisation or many hours of work may increasingly be available through software in seconds.
This is the next price collapse.
The printing press made copies of knowledge cheaper.
The internet made distribution cheaper.
AI makes some forms of cognitive production cheaper.
That changes the race.
The Four Great Reductions in the Price of Intelligence
We can see the development in four broad stages.
Stage One: Intelligence Embodied in a Person
Knowledge travelled mainly through direct human contact.
To access intelligence, a person had to meet the teacher, master, scholar, healer or adviser who possessed it.
Capability was local, scarce and difficult to reproduce.
Stage Two: Intelligence Preserved in Print
The printing press allowed ideas to survive their original author and travel without requiring the author’s physical presence.
Knowledge became reproducible.
A person could learn from a mind that lived in another country or another century.
Stage Three: Intelligence Connected Through the Internet
The internet reduced the cost of searching, copying and transmitting information.
The problem was no longer simply scarcity.
It became abundance.
People gained access to more information than they could evaluate or absorb.
Stage Four: Intelligence Performed by Machines
AI does not merely retrieve existing material.
It can transform information into an output designed for the immediate user.
The user can ask a question, request a comparison, change the difficulty, challenge the response and generate another version.
The machine begins to participate in the cognitive process.
This fourth stage is still developing.
But it already asks education a serious question:
What happens when the ability to produce a competent-looking answer is no longer scarce?
Information Is Not Intelligence
We should be careful with our language.
Information and intelligence are not identical.
Information consists of facts, signals, descriptions and recorded knowledge.
Intelligence involves the ability to interpret information, detect relationships, solve problems, form judgments and act towards an objective.
The internet made information abundant.
AI is making certain applications of intelligence more abundant.
But neither automatically produces wisdom.
A person may have access to unlimited information and remain confused.
A machine may produce a convincing answer that is incorrect.
A society may become more computationally capable without becoming more responsible.
Therefore, the falling price of machine intelligence does not remove the need for human education.
It changes what human education must accomplish.
Intelligence Will Become More Measurable
The capabilities of AI are increasingly being measured against defined tasks and aspects of human ability.
Benchmarks already test performance in language, mathematics, coding, scientific reasoning, perception and other domains.
The OECD has developed beta AI Capability Indicators intended to compare AI systems with human abilities across areas including language, social interaction, problem-solving and physical manipulation. The OECD also acknowledges that present measurements are incomplete and that many advanced capabilities remain difficult to benchmark properly.
This distinction matters.
We should not imagine that all intelligence will soon be reduced to one perfect number.
Human intelligence is multidimensional.
It changes across contexts.
It includes learned knowledge, reasoning, creativity, physical capability, social understanding, emotional regulation and practical judgment.
Even apparently simple abilities can be difficult to define and measure.
But more cognitive performance will become observable, comparable and priced.
An organisation may be able to estimate:
- how much reasoning performance it receives per dollar;
- how accurately a system completes a defined task;
- how quickly it processes a body of information;
- how many customers it can assist simultaneously;
- how often human review is required;
- and whether a machine or person performs the task more reliably.
Once capability can be measured and purchased, it becomes an economic input.
And once the price of that input falls, any human work that is nearly identical to it may face pressure.
The Convergence Problem
For many years, education has encouraged students to converge towards a common answer.
The same syllabus.
The same model response.
The same examination method.
The same approved structure.
The same definition of success.
Some convergence is necessary.
Children need shared foundations.
Mathematics requires common principles.
Language requires conventions.
Science requires standards of evidence.
Society requires enough common knowledge for people to communicate and cooperate.
But excessive convergence creates a new danger.
If every student is trained to perform the same standardised cognitive procedure, and AI becomes extremely good at that same procedure, then the student is entering direct competition with a rapidly improving and rapidly cheapening system.
The student may spend years learning to generate work that a machine can produce in seconds.
This does not make the education worthless.
The foundational understanding may still be essential for verifying and directing the machine.
But it means reproduction cannot remain the final destination.
Education must move beyond:
“Can the student produce the standard answer?”
It must ask:
“What can this student notice, decide, create, connect or take responsibility for that changes the value of the answer?”
The Race Is Not Human Against Machine
It is tempting to imagine a straightforward race.
Humans on one side.
Artificial intelligence on the other.
Both attempt the same tasks.
The faster and cheaper competitor wins.
That is the wrong race.
A person should not try to compete with a computer at being a computer.
A human being will not win through greater processing speed, perfect recall, continuous operation or the ability to examine millions of possibilities simultaneously.
AI systems can perform some forms of computation, search and pattern processing at a scale no individual human mind can match.
Humans, however, exist inside the world that the intelligence is being used to change.
We experience consequences.
We form relationships.
We live inside families and communities.
We possess bodies that can be harmed.
We inherit histories.
We create institutions.
We decide what we are willing to accept.
We remain answerable to one another.
The important division is therefore not:
Which side is more intelligent?
It is:
Which form of intelligence is appropriate for which part of the problem?
AI should do things that machines can do extraordinarily well.
Humans should develop and protect the capabilities through which people establish purposes, interpret consequences, create legitimacy and accept responsibility.
Then the two can be combined.
From Convergence to Divergence
At present, AI is converging towards human capability.
It is learning to write, reason, speak, see, create, diagnose, plan and operate in ways that resemble human work.
At the same time, humans are converging towards machines.
We are learning to follow templates.
Optimise for measurable output.
Respond rapidly.
Process greater volumes of information.
Produce standardised answers.
Work through digital interfaces.
Reduce complex judgment to repeatable procedures.
The two lines are moving towards one another.
This creates the most dangerous zone: the area in which human work becomes machine-like at the same moment that machines become more human-like.
That is where substitution becomes easiest.
We therefore need a new educational movement:
AI may continue converging towards human tasks. Human development must begin deliberately diverging towards capabilities, combinations and responsibilities that become more valuable beside AI.
Divergence does not mean becoming irrational, inefficient or anti-technological.
It does not mean refusing to use AI.
It means refusing to educate every person into the same narrow, automatable shape.
The Boundary Will Keep Moving
We should not construct education around a permanent list titled:
Things AI Cannot Do
Such a list may age badly.
Capabilities that appear uniquely human today may become partly automatable tomorrow.
The OECD’s work on AI and future skills exists precisely because the boundary between human and machine capability is changing and must be monitored continually.
The better question is not:
What can AI never do?
It is:
Where should human capability be positioned as AI develops?
That requires three forms of movement.
Move Away From Direct Substitution
Do not build a career entirely around a task whose output can be generated more cheaply by a machine.
Move Towards Complementarity
Develop capabilities that help define, direct, verify, contextualise or apply machine output.
Move Towards Human Responsibility
Occupy the parts of the system where a human being must remain accountable for the goal, decision, relationship or consequence.
The aim is not to locate one safe island and stay there forever.
The aim is to develop enough intelligence to keep moving as the boundary moves.
Human Intelligence Must Become More Human
For centuries, education often treated intellectual development as the accumulation of knowledge.
Knowledge remains essential.
But when machines can provide enormous amounts of knowledge on demand, human development must become deeper.
Human intelligence must include the ability to decide:
- what deserves attention;
- which problem is worth solving;
- what information is missing;
- which values are in conflict;
- what consequences are acceptable;
- who may be harmed;
- when a technically correct answer is socially wrong;
- when efficiency should not be the highest objective;
- and who will accept responsibility if the decision fails.
These questions cannot be settled through computation alone.
A machine can assist with them.
It can present options, reveal patterns and simulate consequences.
But a civilisation still needs human beings and institutions to determine the ends towards which machine capability is directed.
Intelligence can optimise a route.
It does not independently establish why the destination should matter to us.
The Human Side of the Divergence
The human edge will not come from one magical skill.
It will come from combinations.
Purpose
AI can assist in reaching an objective.
Human beings must still argue about which objectives are worth pursuing.
Problem Selection
When answers become cheap, selecting the right question becomes more valuable.
A perfectly optimised answer to the wrong problem remains wasteful.
Context
A machine may recognise broad patterns while missing local history, unspoken expectations, personal circumstances or cultural meaning.
Human beings live inside these contexts.
Judgment
Many decisions involve incomplete evidence, competing interests and irreversible consequences.
Judgment is the disciplined act of choosing despite uncertainty.
Responsibility
A machine may contribute to a decision.
But a human professional, leader or institution may still need to explain it, defend it and carry its consequences.
Relationships
Trust is built through continuity, conduct, shared risk and the belief that another person genuinely recognises one’s circumstances.
Machines may support relationships.
They do not remove the social need for accountable human connection.
Embodied Capability
The physical world remains irregular.
Bodies, homes, machinery, cities and natural environments do not always behave like clean digital models.
Human beings who can combine thought with physical perception and action may remain highly valuable.
Meaning
People do not live by efficiency alone.
They care about dignity, identity, belonging, beauty, memory and purpose.
A civilisation must decide not merely what it can produce, but what kind of life its production is meant to support.
These are not reasons to reject AI.
They are the human structures required to use AI intelligently.
AI Must Also Diverge From Humans
There is another side to the argument.
We should not force AI to imitate human beings in every respect.
AI becomes most valuable when it performs work that human beings cannot perform well.
It can examine quantities of data too large for one person.
It can operate continuously.
It can model many scenarios.
It can search large spaces of possible solutions.
It can preserve and retrieve enormous amounts of information.
It can coordinate complex systems at machine speed.
It can enter dangerous, distant or inhospitable environments through robotics.
The purpose of AI should not merely be to create a cheaper imitation of an office worker.
Its larger promise is to expand the space of what civilisation can understand and accomplish.
This gives us a double divergence:
Humans become more deeply human.
Machines become more powerfully machine-like.
The value lies at the interface.
Human beings provide direction, context, values and responsibility.
Machines provide scale, speed, simulation and computational reach.
Each should strengthen the other without erasing the other.
Diversity Becomes Civilisation’s Advantage
This brings us to one of the largest educational changes ahead.
Diversity will become more valuable.
Not diversity as a decorative slogan.
Not diversity as a box to tick.
Diversity as productive architecture.
When intelligence was scarce, civilisation benefited from standardising knowledge and spreading it widely.
When machine intelligence becomes abundant and standard answers become cheap, civilisation benefits from producing different questions, perspectives, combinations and routes.
If everyone uses the same systems, follows the same recommendations and accepts the same optimisation, society may converge towards a cognitive monoculture.
The outputs may be efficient.
They may also share the same blind spots.
A resilient civilisation needs people who notice different things.
People trained in different disciplines.
People shaped by different communities.
People with different temperaments.
People who preserve local knowledge.
People who challenge assumptions that others cannot see.
People who combine ideas that normally remain separated.
Research suggests that diversity can expand the range of ideas and perspectives available to a group, but the benefit is not automatic. Communication structures, inclusion, trust and the design of collaboration affect whether diversity improves or harms collective problem-solving.
That is the important lesson.
Diversity is not simply the presence of difference.
It is the ability to turn difference into collective intelligence.
Diversity Is a Search Strategy
Imagine civilisation trying to find a route through an unknown landscape.
If everyone walks in the same direction, the group moves quickly.
But if that direction is wrong, everyone becomes lost together.
If different groups explore different routes, progress may initially appear less efficient.
Some routes will fail.
Others may discover water, shelter or a path that no central planner anticipated.
Diversity expands the search.
This applies to:
- scientific hypotheses;
- business models;
- cultural expression;
- educational methods;
- political ideas;
- technological designs;
- and ways of organising society.
AI can generate many variations, but the sources from which it learns and the objectives through which it is directed still matter.
If the underlying culture becomes too uniform, machine-generated variety may be broad in appearance but narrow in origin.
Human diversity protects the source material of civilisation.
It keeps multiple histories, questions, values and forms of knowledge alive.
Diversity Is Also Risk Management
A financial portfolio containing one asset may perform well—until that asset fails.
An ecosystem containing one crop may be efficient—until a disease attacks that crop.
A society dependent on one form of intelligence may be powerful—until it meets a problem that intelligence cannot see.
Cognitive, cultural and disciplinary diversity act as a form of risk distribution.
Different people will make different mistakes.
More importantly, they may detect one another’s mistakes.
One person sees the number.
Another sees the person hidden by the number.
One sees the engineering possibility.
Another sees the social consequence.
One sees the immediate gain.
Another recognises the historical pattern.
One preserves stability.
Another recognises when stability has become stagnation.
Civilisation becomes more intelligent not when every person thinks identically, but when sufficiently different minds can cooperate without losing the ability to disagree.
Education Cannot Manufacture Identical Excellence
The previous educational model often treated fairness as producing the same pathway for everyone.
A common curriculum created shared foundations and national capability.
That remains valuable.
But the future requires a more sophisticated balance:
Common foundations, followed by intelligent divergence.
Every child needs literacy.
But not every child must use language in the same way.
Every child needs numeracy.
But not every child must apply Mathematics to the same problems.
Every child needs scientific reasoning.
But some will become experimentalists, some engineers, some clinicians, some environmental thinkers and some informed citizens.
Every child needs digital and AI literacy.
But they should not all emerge as interchangeable operators of identical systems.
The purpose of a strong foundation is not to make every building look the same.
It is to allow many different buildings to stand.
What This Changes in the Classroom
Teaching must gradually move from standardised reproduction towards differentiated capability.
Teach the Common Core Properly
Divergence without foundations becomes fragmentation.
Students still need language, Mathematics, scientific understanding, historical context and digital literacy.
They cannot challenge a system they do not understand.
They cannot direct AI if they cannot evaluate its output.
Ask Students to Produce More Than Answers
Students should explain:
- why they chose the problem;
- which assumptions they made;
- what evidence they rejected;
- how AI was used;
- what the AI missed;
- what alternative route was considered;
- and who might experience the result differently.
Reward Useful Difference
A student should not receive credit merely for being different.
Difference must remain disciplined by evidence, logic and consequences.
But when several sound approaches exist, education should not punish students for failing to imitate the model response exactly.
Create Cross-Disciplinary Work
The future may reward people who connect fields.
Biology with computation.
Engineering with ethics.
Healthcare with design.
Mathematics with public policy.
History with technology.
Education with cognitive science.
Many important discoveries emerge at boundaries because the assumptions of one discipline become visible from another.
Preserve Human Contact
When machine tutoring becomes common, schools should not conclude that human relationships matter less.
They may matter more.
Teachers can observe hesitation, identity, effort, belonging and emotional change across time.
The teacher’s future role may shift away from delivering every piece of information and towards diagnosing development, designing challenge, building culture and guiding judgment.
Let Students Develop a Personal Edge
Students should gradually discover:
- what they notice;
- what they care about;
- how they think;
- what kinds of problems hold their attention;
- which environments bring out their best work;
- and what combination of capabilities makes them unusually useful.
This is not an excuse to avoid weak areas.
It is the beginning of purposeful divergence.
The New Educational Question
The old question was:
How intelligent is this student?
The next question may be:
What form of intelligence is this student developing, and how does it combine with the intelligence now available from machines and other people?
A student may not be the fastest at routine calculation.
But the student may be exceptional at recognising which variables matter.
Another may not generate the most fluent first draft.
But the student may possess unusual moral clarity or cultural understanding.
Another may struggle with standard classroom performance but excel at physical systems, spatial problems or real-world repair.
Another may connect ideas across fields that specialists keep separate.
When standard cognitive output is expensive, systems tend to reward those who can produce it.
When standard cognitive output becomes cheap, unusual combinations become more valuable.
A New Formula for Intelligence
We may need to stop treating intelligence as an isolated possession.
In the future, effective intelligence may look more like this:
**Human foundations
- personal difference
- collective diversity
- machine capability
- sound judgment
- responsibility
= civilisation intelligence**
The individual does not need to know everything.
The individual needs enough foundation to contribute something reliable, enough difference to add something that is not already present, and enough humility to connect with other forms of intelligence.
AI becomes part of this system.
It is not the whole system.
What Parents Should Understand
Parents may fear that AI will make their child less competitive.
The instinctive response is to push the child to learn more, move faster and accumulate more qualifications.
Some of that preparation may help.
But acceleration alone is not a strategy when the competitor is a machine.
Parents should protect the foundations while helping the child develop a recognisable human edge.
Ask:
- What does my child notice that others overlook?
- Can my child explain why something matters?
- Can my child form an independent judgment?
- Can my child work with people who think differently?
- Can my child use AI without surrendering thought?
- Can my child build something from an idea?
- Can my child take responsibility when an outcome is uncertain?
- Can my child connect knowledge across subjects?
- Can my child remain curious after the examination is over?
- Is my child becoming more capable—or merely more compliant?
A child still needs discipline.
But discipline should protect development, not erase individuality.
What Students Should Understand
You do not need to defeat AI.
You need to learn how to stand beside it without disappearing into it.
Use AI to expand what you can see.
Let it explain difficult concepts.
Ask it to compare methods.
Use it to test ideas.
Let it handle parts of the work that do not require your full human attention.
But do not allow it to take over the exact struggle through which your intelligence is supposed to grow.
Before accepting an answer, ask:
- Do I understand it?
- Is it true?
- What assumptions does it contain?
- What would happen if it were applied?
- What is missing?
- What can I add that comes from my own observation, experience and judgment?
Your future value will not come from pretending AI does not exist.
It will come from becoming the kind of person who makes AI more useful, safer, wiser and more connected to reality.
Intelligence Becomes the Floor, Not the Ceiling
As machine intelligence becomes cheaper, more people may gain access to abilities that once belonged only to specialists or large organisations.
This could widen human opportunity.
It could allow a small business, student, teacher, researcher or community group to use extraordinary cognitive tools.
AI may raise the floor.
But raising the floor does not determine the ceiling.
The ceiling will depend on what human beings decide to build with the capability.
If everyone asks the same questions, accepts the same answers and pursues the same objectives, cheaper intelligence may produce greater sameness.
If people combine strong foundations with different perspectives, disciplines, values and lived experience, cheaper intelligence may produce an explosion of useful possibility.
That is why diversity matters.
Not because difference is automatically correct.
Not because every idea has equal value.
But because civilisation cannot explore an unknown future through one mind, one model or one route.
The Race We Should Actually Run
The race is not to preserve the high price of intelligence.
That price is falling.
The race is not to prevent machines from becoming capable.
They will continue to improve.
The race is to improve human education quickly enough that cheap machine intelligence expands human civilisation rather than flattening it.
We need children who possess strong foundations but do not become identical.
Children who can use AI but are not intellectually absorbed by it.
Children who can cooperate without surrendering independent judgment.
Children who can inherit knowledge without becoming trapped inside old answers.
Children who can create new directions when the existing path becomes automated.
The future will contain convergence.
Human and machine capabilities will overlap.
But overlap should not become erasure.
We must build the divergence deliberately.
Let machines become extraordinarily good at what machines can do.
Let human beings become more thoughtful, responsible, diverse, relational, courageous and alive to consequence.
Then connect the two.
The next civilisation will not be built by human intelligence alone.
Nor should it be handed entirely to artificial intelligence.
It will be built at the interface between different forms of intelligence.
That interface becomes powerful only when the two sides bring different strengths.
The price of intelligence is falling.
The value of difference is rising.
Education must now prepare children for both.
Frequently Asked Questions
What is the difference between schooling and education?
Schooling is the structured period in which teachers, curricula, routines and assessments organise learning. Education is the knowledge, judgment, character and learning capacity that remain with the person and continue developing throughout life.
Will artificial intelligence make studying unnecessary?
No. AI may reduce the need to perform some routine tasks manually, but it increases the importance of foundations, judgment, verification and clear thinking. Students need enough understanding to direct AI, evaluate its outputs and recognise errors.
What should parents prioritise for the next 30 years?
Parents should protect strong academic foundations while also developing attention, independence, communication, adaptability, responsibility and the ability to learn from mistakes.
Will qualifications still matter?
Yes. Qualifications will continue to signal knowledge, discipline and recognised achievement. However, one qualification may no longer be sufficient for an entire working life. The ability to renew and extend one’s capabilities will become increasingly important.
Is MOE V3.0 an official Ministry of Education programme?
No. In this article, MOE V3.0 is an eduKateSG conceptual framework for thinking about the next stage of education: schooling as the installation period for a lifelong, adaptive education spine.
What is the education spine?
The education spine is the durable internal system that helps a person learn, reason, verify information, communicate, cooperate, adapt and act responsibly across changing situations.
How should students use AI for schoolwork?
Students should use AI to support explanation, comparison, brainstorming, practice and feedback. They should not use it to bypass the thinking and foundational practice required to become independently capable.
What is the ultimate purpose of school?
School should provide more than subject content and examination preparation. Its deeper purpose is to install the foundations, habits, judgment, relationships and learning capacity that allow a person to continue developing after formal schooling ends.
