The Next Education Shift: What If a Learner Never Had to Start From Zero Again?

There is something I keep thinking about when I look at education today.

We have never had more knowledge, more teachers, more universities, more learning platforms or more technology available to us. We can put a lesson from a great university on a student's screen in seconds. We can translate content into different languages. We can assess students digitally. We can now use AI to explain a concept, generate questions and respond to a learner's answers.

And yet, something very basic is still missing.

Continuity.

A learner spends years moving through education, but the education system often sees that learner in pieces.

One teacher knows one part of the story.

The next teacher starts again.

The school knows another part.

The parent knows something completely different.

The tutor knows where the child struggles.

An assessment system knows the scores.

An online platform knows what the learner completed.

The university eventually receives a transcript.

But who really understands the learner's journey?

That question has been on my mind for some time.

We solved scale. We may now need to solve continuity.

If we look back at older forms of education, one thing stands out.

The teacher often knew the learner over a long period.

The traditional gurukul is one cultural example. The interesting part is not whether we should recreate an ancient education system — we clearly should not. It is the idea of sustained observation.

The familiar story of Dronacharya and Arjuna illustrates a teacher observing a learner's ability, practice, discipline and development over time.

European apprenticeship traditions had a different form, but there was a similar principle. A learner developed alongside someone with deeper experience.

The limitation was obvious.

It was difficult to scale.

Modern education solved that.

We created schools, universities, curricula, examinations, teacher-training systems and eventually digital platforms that could educate millions.

That was an extraordinary achievement.

But there was a trade-off.

The more we scaled education, the harder it became for the system to maintain a deep, continuous understanding of each learner.

We became very good at delivering education.

We became reasonably good at measuring education.

We are still learning how to continuously understand the learner.

And the numbers tell us why this matters.

Since 2015, 110 million more children, adolescents and young people have entered school, while 40 million more young people are completing secondary school. Global tertiary enrolment has increased from 30% in 2010 to 43%. Yet 251 million children and young people remain out of school. [1]

Getting children into education is a huge achievement.

But getting them to actually learn is another challenge.

The World Bank estimated that learning poverty in low- and middle-income countries reached around 70% after the pandemic — meaning roughly seven in ten children could not read and understand a simple age-appropriate text by age 10. [2]

So perhaps the next education challenge is not simply access.

It is understanding.

A child does not experience education as a collection of systems

Imagine a 12-year-old student entering Grade 7.

They have already spent years learning.

They have strengths.

They have weaknesses.

They may love science but dislike mathematics.

They may understand a concept when someone explains it visually but struggle with written questions.

They may have developed confidence in one subject and anxiety in another.

Their parents know some of this.

Their previous teacher knows some of it.

A tutor may know something else.

The school has assessment records.

The LMS has activity records.

But when the learner enters a new classroom, how much of that understanding actually travels with them?

Often, not enough.

And then we wonder why learning gaps become difficult to fix.

The test tells us the student is behind.

But the real question is:

When did the learner begin falling behind?

What caused it?

Was someone able to see it?

Was an intervention attempted?

Did it work?

Did the next teacher know?

That is where longitudinal intelligence becomes important.

Not simply remembering what a learner did.

Understanding how the learner developed.

This is not only a problem in developing countries

It would be easy to look at this as an access problem in developing countries.

It isn't.

The challenge simply looks different in different places.

A rural school in South Africa may struggle to access specialist teachers.

A rural school in India may have an excellent teacher but limited resources and a large classroom.

A school in New York may have highly qualified teachers and excellent technology, but a teacher may still be trying to understand 25 or 30 individual learners.

Europe has its own teacher recruitment and retention challenges.

Latin America faces significant learning and inequality challenges.

In high-performing education systems, students can still struggle with motivation, anxiety and pressure.

PISA 2022 found that across OECD countries, 65% of students worried about getting poor marks in mathematics, 55% felt anxious about failing mathematics, and around 40% reported feeling nervous, helpless or anxious while solving mathematics problems or doing mathematics homework. Mathematics anxiety had also increased in most participating countries and economies between 2012 and 2022. [3]

So even where the infrastructure is strong, something can still go wrong between the learner and learning.

That is why I don't think the answer is simply “more content.”

The teacher is not the problem

This is perhaps the most important point.

Teachers are not failing because they cannot personalize education perfectly for every student.

The system is asking too much of one human being.

UNESCO estimates that the world will need 44 million additional primary and secondary teachers by 2030. Sub-Saharan Africa alone is expected to need around 15 million more teachers. UNESCO also points out that teacher shortages are increasingly affecting higher-income regions through recruitment and retention challenges. [4]

Now imagine being a teacher with 40 students.

You have to teach.

Plan.

Assess.

Manage the classroom.

Communicate with parents.

Handle administration.

And somehow remember the learning journey of every student.

A great teacher can do remarkable things.

But even a great teacher cannot be expected to maintain a detailed, evolving mental model of dozens of learners over many years.

That is not a teacher problem.

It is a scale problem.

And this is where technology becomes interesting.

What if expertise could travel instead of students?

Imagine a mathematics teacher in rural Maharashtra who is exceptional at teaching algebra.

Her classroom may have 40 students.

Her influence may effectively stop at the walls of that classroom.

Now imagine that her teaching approach could be captured, configured and extended through an AI-supported teaching system.

Not a video.

Not a recorded lecture.

Not a generic chatbot.

An interactive AI Teacher that understands her pedagogy, her way of explaining concepts, the curriculum context and the learner's progress.

A student in New York could potentially learn through an experience influenced by that teacher's expertise.

A learner in Nairobi could access the same conceptual strength.

A student in Singapore could encounter it within a different curriculum and cultural context.

The teacher remains in Maharashtra.

But her expertise is no longer geographically trapped there.

This is one of the possibilities that makes AI genuinely interesting to me.

For centuries, education has often moved learners toward scarce expertise.

Perhaps technology can increasingly move expertise toward learners.

That is a very different idea.

And it changes the conversation about educational equity.

Imagine the reverse as well

Imagine a brilliant science teacher in New York.

Why should that teacher's expertise be available only to students who happen to attend that school?

Why couldn't a learner in rural South Africa benefit from an AI-supported learning experience built around that expertise?

The answer cannot simply be “put the teacher on Zoom.”

That doesn't scale.

The teacher cannot be simultaneously present in 500 classrooms.

But what if the teacher's expertise could become part of an intelligent educational ecosystem that could adapt explanations, interact with students, understand responses and involve a human teacher when necessary?

Then geography starts to matter less.

Not because every student receives the same education.

But because more students can gain access to high-quality expertise.

That is the beginning of what I think of as educational equilibrium.

Not equality of outcomes.

Not identical classrooms.

But a world where geography and family wealth have less influence on whether a learner can access the expertise they need.

But AI alone is not enough

This is where I think we need to be careful.

We could build thousands of AI tutors.

One for mathematics.

One for English.

One for science.

One for coding.

One for exam preparation.

One for university research.

It sounds wonderful.

But we could also create a new problem.

Fragmentation.

The mathematics AI knows one part of the learner.

The school knows another.

The assessment platform knows another.

The parent knows another.

The university eventually knows another.

We would have intelligent systems everywhere, but the learner would still be moving between disconnected worlds.

That is why I believe the next important layer in education may not be another application.

It may be an Education Operating System.

Not an LMS replacement.

Not an ERP replacement.

Not a giant database.

Not an AI chatbot.

An architectural layer that allows the different parts of education to work together around the learner.

The school can continue using its systems.

The university can continue using its systems.

Teachers can continue using their tools.

Assessment providers can continue operating.

Parents can continue participating.

AI Teachers can continue evolving.

But there needs to be a way for the right context to move between them.

Safely.

With governance.

With consent.

With purpose.

And without turning education into surveillance.

What would actually change in the classroom?

Let's return to that Grade 7 mathematics student.

The student is working on:

3/4 + 2/3

Instead of simply selecting an answer, the learner writes the calculation on a digital notebook.

They explain their thinking through voice.

The system sees the working.

It recognizes that the learner knows how to add fractions but doesn't understand why a common denominator is required.

Instead of saying “wrong,” the AI Teacher changes the explanation.

Perhaps it asks a question.

Perhaps it draws a visual representation.

Perhaps it gives another example.

The learner tries again.

The system sees improvement.

Another learner in the same classroom may need an entirely different explanation.

The human teacher doesn't need to become 30 different teachers.

The intelligent layer provides additional capacity.

The teacher remains the person who understands the classroom, the child and the human context.

That distinction matters enormously.

The goal should not be AI versus teachers.

It should be AI with teachers.

Then something even more important happens: the system remembers

Six months later, the learner encounters another difficult concept.

The system already knows something important.

The learner struggled with fractions.

A visual explanation worked.

Repeated practice helped.

The learner eventually improved.

Now the learner is struggling again with an abstract mathematical concept.

That history matters.

The next teacher shouldn't have to rediscover it.

The learner shouldn't have to start from zero.

This is what I mean by longitudinal intelligence.

Not simply collecting more student data.

But maintaining a useful understanding of the learner's development over time.

That could eventually change the way we think about assessment.

Instead of asking only:

“What did the student score?”

we can begin asking:

“How did this learner develop?”

That is a much more powerful question.

Parents need understanding, not more dashboards

Most parents don't need another dashboard with 20 graphs.

They want to know:

Is my child progressing?

Where are they struggling?

What is improving?

What can I do?

Imagine receiving meaningful developmental context rather than simply receiving a report card.

That could change the relationship between parents and schools.

The parent becomes a partner in learning rather than someone who discovers problems only after an examination.

Schools need a different kind of intelligence

A school principal has another problem.

They don't need to watch every interaction between a student and an AI Teacher.

They need to understand the school.

Where are learning gaps increasing?

Which subjects are causing persistent difficulty?

Which interventions are working?

Where are teachers overloaded?

Which students need additional support?

Are learning outcomes improving?

The same educational intelligence can therefore serve different people differently.

The learner needs guidance.

The teacher needs context.

The parent needs understanding.

The school needs visibility.

The education expert needs evidence.

The policymaker needs patterns.

The ministry needs system-level insight.

One ecosystem.

Different views.

Different responsibilities.

Different permissions.

That is why governance cannot be something added after the technology is built.

It has to be part of the architecture.

Now imagine what this could mean for researchers

A researcher studying mathematics education may want to understand why one teaching approach works for some learners and not others.

Today, evidence often comes from studies, assessments and aggregated data.

What if, with appropriate governance and privacy, educational systems could create a stronger feedback loop between research and classroom practice?

Research informs pedagogy.

Pedagogy informs teaching.

Teaching produces evidence.

Evidence improves pedagogy.

The loop becomes continuous.

That could make educational innovation much more evidence-driven.

Not because AI knows what good education is.

But because educators and researchers can learn faster about what actually works.

Now move from the classroom to the ministry

A policymaker doesn't need to know that a particular child got question seven wrong.

They need to know that thousands of Grade 7 learners across a region are struggling with the same concept.

They need to know whether the problem is concentrated in rural schools.

Whether teacher availability is a factor.

Whether a particular intervention is helping.

Where additional resources might have the greatest impact.

That is where an Education Operating System could eventually become more than a school technology.

It could become part of the intelligence infrastructure of an education ecosystem.

But this must be done carefully.

The objective should never be to create a giant surveillance system for children.

The individual learner should receive personalized support.

The teacher should receive relevant context.

The school should receive appropriate institutional insight.

Researchers should receive governed, ethical evidence.

Policymakers should receive appropriately aggregated patterns.

The more intelligence we create, the more responsibility we have to govern it.

And then comes the hardest test: can it work everywhere?

UNICEF and ITU estimated that 1.3 billion school-age children — about two-thirds of children aged 3–17 — did not have internet access at home. The gap was enormous across regions: 95% of school-age children in West and Central Africa, 88% in South Asia, 75% in the Middle East and North Africa, 49% in Latin America and the Caribbean, 42% in Eastern Europe and Central Asia, and 32% in East Asia and the Pacific were reported as unconnected at home. [5]

There was also a major urban-rural divide. UNICEF reported that around 42% of urban children had internet access at home compared with only around 23% of rural children. [6]

That means a global education architecture cannot assume that every learner has the same device, bandwidth or environment.

A system designed only for New York, London or Singapore will not solve the global education problem.

It needs to work with different realities.

High-resource school.

Low-resource school.

Urban classroom.

Rural classroom.

Strong connectivity.

Limited connectivity.

Different languages.

Different curricula.

Different cultures.

Different policy environments.

The principle can remain common.

The implementation must remain local.

This is where I see the real opportunity

Imagine a rural school in South Africa where specialist teaching is limited.

The local teacher remains at the centre.

But the teacher has access to intelligent teaching support that can bring additional subject expertise into the classroom.

Imagine a rural school in India where an exceptional mathematics teacher has developed a powerful way of explaining difficult concepts.

Her physical classroom may have 40 students.

But her expertise could potentially be extended to thousands of learners through an AI-supported teaching layer.

Imagine a student in New York learning from an AI Teacher influenced by that educator's approach.

Imagine a student in Nairobi learning from the same underlying expertise, adapted to the local curriculum.

Imagine a student in Singapore experiencing it in another context.

The teacher does not need to leave India.

The learner does not need to leave New York.

The expertise travels.

That is the kind of global education network that technology may finally make possible.

And I believe this is where the word “equilibrium” becomes interesting.

We shouldn't try to make every student identical.

We should try to make high-quality learning opportunity less dependent on geography.

The Education Operating System is really about connecting the ecosystem

At some point, education becomes too complex for another standalone application.

There are students.

Parents.

Teachers.

Tutors.

Schools.

School networks.

Universities.

Researchers.

Assessment providers.

Content providers.

Education boards.

Governments.

Employers.

AI Teachers.

Each has a different responsibility.

Each has different information.

Each has different permissions.

Yet they are all connected by the learner.

That is the architectural gap I believe is worth exploring.

An Education Operating System would not replace the education ecosystem.

It would connect it.

The LMS remains.

The SIS remains.

The university platform remains.

The teacher remains.

The parent remains.

The assessment provider remains.

The AI Teacher remains.

But they become participants in a larger, learner-centred architecture.

Why this matters beyond today's AI cycle

This is also why I don't think the idea should be tied to today's AI models.

The models of 2035 will be very different from the models we use today.

Devices will change.

Interfaces will change.

Assessment will change.

Teaching methods will change.

Universities will change.

New education technologies will emerge that we cannot even predict.

But one thing will remain.

A learner will still move through a journey.

Primary school.

Secondary school.

University.

Postgraduate education.

Professional learning.

Career changes.

Lifelong learning.

If the learner's educational context disappears every time they cross an institutional boundary, we will continue rebuilding their story again and again.

An Education Operating System offers a different possibility.

Let the applications change.

Let the AI models change.

Let the institutions evolve.

Let pedagogy improve.

But preserve the learner's continuity — responsibly, with privacy, consent and governance.

That is what could make an EOS a long-term infrastructure layer rather than another EdTech application.

Perhaps the future of education is less about where you learn

For most of human history, geography mattered enormously.

Where you were born influenced which teachers you could meet.

Which school you could attend.

Which university you could reach.

Which specialists were available.

Technology has already started breaking some of those boundaries.

AI could accelerate that dramatically.

A brilliant teacher in rural Maharashtra could potentially influence learners beyond her village.

A specialist in New York could potentially support learners in Nairobi.

A researcher in Singapore could influence classroom practice in Latin America.

A professor in Europe could contribute to learning experiences thousands of kilometres away.

But the goal should not be to replace the human relationship.

It should be to extend it.

Not to make every learner the same.

To give every learner better access to what they need.

Not to make every school identical.

To give every school access to capabilities that geography once restricted.

Not to replace teachers.

To give great teachers greater reach.

And not to collect more data about learners.

To use the right information, responsibly, to help them develop.

That is the education shift I believe is worth exploring.

A child in rural South Africa should not be limited to the expertise available within a few kilometres of their school.

An exceptional teacher in rural India should not be limited to the students physically sitting in front of her.

A parent should not have to understand a child's development through grades alone.

A teacher should not have to reconstruct a learner's history every year.

A school should not have to discover learning gaps only after they become large.

A researcher should be able to learn more quickly about what actually works.

A policymaker should have better evidence before deciding where scarce resources should go.

And a learner should be able to move through education without losing the context of the journey they have already travelled.

If we can responsibly build that, the real disruption will not be that AI became the teacher.

It will be that education became capable of carrying the learner's journey forward.

Perhaps that is what an Education Operating System should ultimately do.

Not replace education.

Not control education.

Not automate education.

Connect education.

So that the most valuable thing in the entire system — the learner's journey — is no longer fragmented.