Why Schools Still Test Memory More Than Judgement

By Ilan Gross

Think about the last time you solved a real problem in adult life.

You looked things up. You checked a document. You searched for the thing you half-remembered. You asked someone. You kept notes so you would not have to hold everything in your head.

That is how adults solve problems. It is how professionals solve problems. It is how teachers solve problems outside the exam hall.

Now think about what we ask a student to do in that hall.

Sit alone. No phone. No internet. No notes. No one to ask. Hold a body of facts in your head and reproduce them on demand, unaided, against the clock.

We grade that. We build league tables on it. We treat it as proof of learning.

Then the student walks out of the room into a world where almost no one works that way again.

That is the quiet absurdity at the centre of the school model.

It still tests sealed-memory recall as if the world has not changed.

The skill made sense once

This is not a cheap shot at teachers, and it is not the claim that school was always pointless.

Memorising used to be valuable.

For most of history, if you did not carry knowledge around inside your head, you often did not have it. The book was in a library you might not reach. The expert was somewhere else. The fact you needed at the moment you needed it was unavailable unless you had already stored it.

In that world, the person who could hold a great deal in memory and produce it accurately had a genuine, scarce and valuable skill.

Grading it made sense.

That world is gone.

Not fading.

Gone.

The fact you need is now in the phone in your pocket. It can be found in seconds. Increasingly, AI can also explain it, compare it, summarise it, challenge it and apply it.

The scarcity that made memory so valuable has changed.

What has not changed enough is the exam.

We are still grading the storage skill long after the storage problem was solved.

“But you cannot think about things you know nothing about”

Here is the serious objection, and it deserves a fair hearing.

It is true.

You cannot think well about a subject you know nothing about.

Judgement is not free-floating. It runs on knowledge that has actually been internalised.

A historian can interrogate a source because they carry the period in their bones, not because they can simply look up a date. A doctor reads a symptom against years of anatomy and experience, not by starting from zero each time. A student who knows nothing cannot magically become wise because they have access to Google or AI.

Strip out all internalised knowledge and you get a different problem.

You get students who can retrieve any fact and understand none of it. They can sound fluent but think weakly. They can accept a confident answer because they have no internal reference to check it against.

That is not learning.

So yes, some knowledge has to live in the student’s head.

Memory is part of the scaffolding that judgement is built on.

You cannot remove the scaffolding and pretend the building will stand.

But that defends the foundation, not the bulk

The objection is right about foundational knowledge.

The problem is that it gets used to defend too much.

There is a real difference between knowledge a student needs in their head in order to think, and knowledge they only had to memorise because there was once no faster way to access it.

The multiplication concept is foundational. Understanding what multiplication means, when to use it, and what a reasonable answer looks like matters.

But manually grinding through every calculation is not the same kind of skill.

That is why calculators did not destroy maths education. They forced education to separate understanding from mechanical execution.

The same distinction now has to be made with memory.

Some knowledge must be internalised because judgement depends on it. That should be taught seriously.

But a lot of what schools still test is not foundational. It is retrievable information dressed up as rigour because it is easy to mark.

A date. A definition. A list. A formula copied from memory. A fact the student will look up the one time they ever need it.

That kind of recall produces a clean mark. It sorts students into a rank order. It fits the exam system.

But administrative convenience is not the same as education.

The honest position is not “memorising is dead.”

It is also not “keep drilling everything because some of it matters.”

The honest position is harder:

Separate the two.

Decide what genuinely has to be internalised because thinking depends on it. Teach that harder than ever.

For the rest, stop pretending that reproducing information from sealed memory is the point.

It never was.

The point was always what knowledge was for: understanding, questioning, checking, connecting, explaining and applying.

AI makes the old model harder to defend

Google changed access to information.

AI changes what students can do with information.

A search engine gives a student places to look. AI can give the student a complete answer, a summary, an explanation, a study plan, a draft essay, a counterargument or a worked example.

That is why schools are nervous.

The fear is understandable. A student can now produce work that looks competent without doing the thinking. They can submit fluent answers that hide weak understanding.

That is a real problem.

But banning the tool does not solve it.

It only protects the old assessment model for a little longer.

If an assignment can be completed by asking AI for a generic answer, the question is not only whether the student cheated.

The deeper question is whether the task was measuring the right thing in the first place.

A better model would not ask only, “Did the student use AI?”

It would ask:

Can the student explain the answer? Can they defend it? Can they challenge it? Can they check whether it is true? Can they show what they changed and why? Can they spot when AI is confident but wrong? Can they use the answer to think, not just submit?

That is a better test of learning.

Why the model defends recall anyway

It is worth being honest about why the school model holds on.

It is not stupidity. It is not bad faith. The people inside the system usually believe they are protecting learning.

The deeper issue is that the school model can measure recall.

It has a century of machinery built to do exactly that: standardised tests, exam halls, mark schemes, grades, league tables and performance comparisons.

It does not have the same reliable machinery to measure judgement, originality, the quality of a question, or whether a student can tell a true answer from a confident one.

So when the world changes, the system defends the part it knows how to grade.

Then it calls that defence a principle.

That is the real reason the exam hall still looks the way it does.

Not because anyone has made a serious case that reproducing facts unaided is the main skill students need in a world of search, AI, calculators, documents and digital tools.

But because recall is easy to test.

Judgement is harder.

What education should measure instead

The answer is not to remove knowledge from education.

It is to be more honest about which knowledge matters.

A modern education system should still teach foundational knowledge. But it should also test whether students can use knowledge well.

That means measuring things like:

• whether a student can explain an idea in their own words
• whether they can apply it to a new situation
• whether they can compare two possible answers
• whether they can find the weak assumption
• whether they can check a claim against evidence
• whether they can ask a better question
• whether they can recognise when an answer is incomplete
• whether they can use AI without surrendering judgement
• whether they can show their reasoning, not just their result

That is not a softer version of education.

It is a more demanding one.

It is harder to teach. It is harder to mark. It is harder to standardise.

But it is closer to real learning.

The real test

The test for any exam is simple:

Does it measure something a person actually does outside this room?

If the honest answer is yes, keep it.

If the honest answer is no, question it.

If the exam only measures whether a student can hold facts in a sealed memory that every real situation would let them open, then it is not protecting rigour. It is protecting habit.

AI, search engines, calculators and digital tools have changed the value of memory. They have not removed the need for knowledge. They have exposed the difference between knowledge that supports thinking and information that can simply be retrieved.

That distinction now matters.

If a fact must live in the student’s head because judgement depends on it, teach it properly.

If a fact is only being memorised because the exam system knows how to mark it, stop pretending that is rigour.

The real educational task is not to make students compete with machines at storing information.

It is to teach them how to think when information is everywhere.