“I don’t get it” tells you a learner is stuck.

It doesn’t tell you where.

Here’s a common algebra question:

Which equation represents this relationship?

“Seven times a number, n, is 12 less than the square of that number.”

The choices are:

A. 7 + n = n² − 12

B. 7n = n − 12

C. 7n = 12 − n²

D. 7n = n² + 12

E. 7n = n² − 12

Before I explain anything, I ask the student to tell me what the language means.

What does “seven times a number” mean?

What does “the square of that number” mean?

If 7n is 12 less than n², which quantity is larger? How would you get from the larger quantity to the smaller one?

Problem solving can’t begin reliably until those relationships mean something.

Once they do, I ask the student to translate small to big.

Start with the smallest complete piece of meaning.

“Seven times n” becomes 7n.

That eliminates A.

“The square of n” becomes n².

That eliminates B.

Now check the complete relationship.

If 7n is 12 less than n², begin with n² and subtract 12.

That eliminates C and D.

E remains:

7n = n² − 12.

The student didn’t have to hold the entire sentence and all five choices in mind before doing something with them.

Each piece of comprehension produced an action. Each action reduced what remained.

Students often call the errors they make on questions like this “silly mistakes.”

Often, they tried to translate too much at once. A small misunderstanding then traveled through the rest of the problem.

One piece at a time, each meaning can be checked as the relationship is built.

The same principle applies before any strategy begins.

If a problem uses the word “integer,” I stop there first.

Can an integer be negative?

Is zero an integer?

If the student can’t answer those questions, the problem isn’t yet what to do. It’s what the words mean.

This is Functional Learning in two layers:

Comprehension as translation.

Problem solving as sequencing what to do next.

Problem solving can’t compensate for missing comprehension. It depends on it.

So when a student says, “I don’t get it,” or “I keep making silly mistakes,” I ask:

What is the smallest piece you can explain?

What can you write from that piece?

How does it fit the whole?

What can you eliminate now?

Then we take the next piece.

The answers show me what to teach next.

Great learners are built, not born.

In The Wall Street Journal, John Masko argues that AI “should be severely restricted for undergraduates” but is “vital for researchers.”

He sees the campus debate over AI as two different debates because a university does two different kinds of work. Research depends on the technology. Undergraduate learning may need protection from it.

I think he’s right about the split. I’d draw the line in a different place.

The real difference isn’t between researchers and students. It’s between work someone can already do independently and work they are still learning to do.

When you already know how to do the work, AI can save time.

When you’re still building the skill, the same help can skip the struggle that builds it.

A recent set of randomized trials shows why that distinction matters.

Participants worked on fraction problems and SAT-style reading questions with an AI assistant that could provide complete answers on request. After a brief period, the tool was removed.

While the AI was available, they did better. Once it was removed, they solved fewer problems than people who never had it. In the SAT-style reading study, the AI group also skipped 8 percent of the questions. The control group skipped 1 percent.

A field experiment with nearly 1,000 high school math students at one Turkish school found the same divide between assisted performance and independent learning.

Students who practiced with a custom GPT-4 assistant without learning-specific safeguards improved during practice, then scored 17 percent lower on the unassisted exam than students who never had it.

A second group used a guardrailed AI tutor built around teacher-designed hints rather than answers. On the unassisted exam, they matched the control group rather than outperforming it.

The line runs through every learner, and it depends on how the tool is used.

Expertise is task-specific. The same person may be ready to use AI for one kind of work and still need the struggle in another.

For parents, my practical rule is simple.

Have the student attempt the problem first.

Ask the tool for a hint, not the answer.

Then close it and ask the student to finish or explain the work unaided.

AI can amplify a skill someone has already built. When the skill is still being built, the tool has to support the thinking — not replace it.

Great learners are built, not born.

An explanation isn’t complete just because it sounds clear to the person giving it.

It is complete when the learner can reconstruct it.

In 1990, Stanford graduate student Elizabeth Newton asked people to tap the rhythm of a familiar song for someone else to identify.

The tappers expected listeners to name the song about half the time.

Listeners succeeded less than 3 percent of the time.

The tappers weren’t careless. They could hear the melody in their heads. The listeners could hear only knocking.

This is a vivid example of what researchers call the curse of knowledge: once we understand something, it becomes difficult to imagine what the same explanation sounds like to someone who does not yet understand it.

I see it every week.

A parent explains a math problem three different ways, each one clearer to the parent than the last.

A teacher asks a student to identify the main idea without checking whether they know how to distinguish the central point from the details that support it.

A company director delegates a project without making explicit the priorities, sequence, and standard for success that seem obvious to her.

The usual response is to explain again. More detail. Another example. More patience.

But more explanation is often just more tapping.

So before I explain again, I ask two things:

Tell me what you think you’re being asked to do.

Walk me through your process.

As learners answer, they usually discover a gap between what they thought they understood and what the task requires.

That gap falls within one of the three capacities I teach:

Comprehension: translating what the task says.

Critical thinking: comparing ideas and organizing them by how they relate.

Problem solving: sequencing what to do next.

I still explain. But now I can begin from the learner’s understanding rather than my own.

To borrow an Ignatian principle: enter through their door.

Great learners are built, not born.

“Are teenagers getting dumber?”

The Economist recently asked that question after average reading and mathematics performance across OECD countries fell to the lowest levels PISA has ever recorded.

The data are real, and they are serious.

But “dumber” is the wrong conclusion to draw from them.

I hear the individual version every week.

A student looks at a score and says, “I’m just bad at tests.”

A country looks at PISA and asks, “Are our kids getting dumber?”

Same reflex, different scale: turning a measurement into a verdict about ability.

A verdict ends the conversation. A diagnosis starts the work.

PISA is not an intelligence test. It measures whether fifteen-year-olds can use what they know: build meaning from a passage, work through a multi-step problem, and apply knowledge to real-world challenges.

Those are learned capacities. They can be strengthened through practice.

The scores cannot tell us, by themselves, why those capacities have declined. But they should make us ask what students are being asked to practice.

How much sustained reading are they doing?

How often do they work through a difficult problem without an answer appearing immediately?

How often are they asked to explain why one idea follows from another?

I’ve spent twenty-five years teaching students one-on-one. Many arrive believing that a score has told them what kind of learner they are.

The next step is to make the problem specific.

Can the student reconstruct what they read?

Can they explain how they moved from one step to the next?

Can they identify where their understanding breaks?

Once the breakdown has a location, the student has something to practice.

That is the more useful way to read these results: not as a judgment on this generation, but as evidence of what we need to teach and practice more deliberately.

Great learners are built, not born.

You don’t have to feel calm to take the next step.

That’s what I want students to hear as fall admissions testing approaches.

Working memory can hold only a limited amount of information at once. Thoughts like “I’m nervous,” “Don’t be nervous,” and “You’re running out of time” compete with the question in front of them.

Trying not to feel anxious becomes another task competing for their attention.

Think of a stressed air traffic controller: several planes waiting for clearance, phones ringing, pilots asking for instructions.

The controller cannot respond to everything at once.

Some of the noise has to remain while the next plane is guided in.

One plane at a time.

In that moment, adding more strategies can be like picking up another phone line - one more thing to manage when attention is already stretched.

What students need is a process that simplifies the work and helps them identify their next step.

That includes recognizing gaps in understanding: noticing the word, phrase, or step they don’t understand instead of trying to push past it.

This is one part of what I mean by Functional Learning: giving students a deliberate process they can continue to use even when they feel anxious.

The aim is not to make the anxiety disappear.

It is to let the anxiety be while taking the next action.

Let the phone ring. Land one plane.

Great learners are built, not born.

I’ve spent close to twenty years preparing students for the SAT, so you might expect my position on its potential return to the University of California to be simple.

It isn’t.

I think the test measures something real. But I also think most of what is sold as test prep is a waste of time and money.

In her New York Times essay, “This California Educational Juggernaut Doesn’t Need the SAT,” Miriam Pawel argues against reinstating the test. She is right that the SAT will not repair California’s unequal K-12 system.

Two months earlier, the Times editorial board made the opposite case. More than 3,000 UC faculty have now signed two open letters calling for admissions testing to return. They are right that eliminating a measure of readiness does not eliminate the need for readiness.

Both things can be true.

The UC San Diego data show why that distinction matters. The share of first-year students placed below college-level precalculus rose from under 1 percent before 2021 to 11.8 percent in 2025.

There are several reasons for that increase. The test did not create those gaps. Removing it did not make them disappear. It left them to be discovered after students arrived.

Test-optional admissions can create a different problem, too.

When Dartmouth studied its test-optional years, it found that less-advantaged applicants were withholding scores that could have helped them.

Students saw that a 1400 fell below Dartmouth’s published range and assumed they should hide it.

But Dartmouth reads scores in context. A 1400 from a student whose high school averages 1000 can be a powerful signal of what that student accomplished with the opportunities available.

Hiding the measurement didn’t protect those students.

It hid them.

The deepest problem is not whether we keep the test. It is what we do about the gap the test reveals.

This is where conventional test prep earns its bad reputation.

The standard playbook is familiar: drill the tricks, game the sections, spike the number, forget it all by June.

That version deserves the criticism.

The score is feedback. Readiness is the product.

The SAT is imperfect, but it samples real capacities: reading closely, reasoning under constraints, and managing complexity on a clock.

Prepare a student by building those capacities, and the score moves because the readiness moved. The benefits continue long after the test is forgotten.

Prepare a student with shortcuts, and the score may move while the student does not.

Bring back the SAT, replace it with a better measure, or leave it out. No admissions policy can close a readiness gap after it has formed.

The important question is not only how we identify readiness.

It is how we build it.

Great learners are built, not born.

New York City just restricted generative AI for nearly 600,000 students. Los Angeles has blocked it on district-issued devices.

That may be wise. It won’t, by itself, make anyone think.

New York’s own announcement says children need to “wrestle with tough problems on their own.” Exactly right. A restriction can protect the chance to struggle. Only the assignment can require it.

Nearly two thousand years ago, Plutarch compared the mind to wood that needs kindling, not a bottle that needs filling. The kindling, he wrote, should create “an impulse to think independently and an ardent desire for the truth.”

Filling is receiving and repeating. Kindling is what happens when a learner works through what they’ve received: questioning an explanation, making connections, testing what they think they understand.

Students were reciting explanations they couldn’t explain long before ChatGPT — no screen required. AI just made filling available on demand, patiently, at 9pm when the assignment is due.

So here’s the question the bans can’t answer: does the work we assign make kindling unavoidable?

One assignment shape that does — with or without AI in the room:

The student builds the argument first, alone.

Then something pushes back — an AI where it’s allowed, a classmate, a teacher, a parent.

The student decides which objection actually holds, and why.

Then revises — and explains what changed and what didn’t.

The objection can come from anywhere. The judgment has to come from the student.

A school can enforce a ban perfectly and still assign mostly filling. A school can allow AI and still demand kindling.

Whether your school bans AI or embraces it, the test is the same: when the tool is gone, what does the assignment still require the student to think through?

Great learners are built, not born.

When a student says, “I don’t know what that word means,” most adults answer right away.

I understand the impulse. Your kid is curious, or has a lot of homework. Why make them wait?

Because what happens next teaches the student what to do with a gap in their understanding.

Supplying the definition solves the immediate problem. It also teaches the student that confusion is a cue to wait for someone else.

Ask. Receive. Move on.

There’s research behind the pause. In experiments, people have remembered more when they tried to answer before receiving the correct answer - even when their first attempt was wrong.

So I try not to answer - not yet.

Instead, I ask the student to reread the sentence with the unfamiliar word replaced by a blank.

What work is the missing word doing in the sentence - and in the paragraph around it?

Never mind identifying the correct definition yet. What does the author appear to be trying to accomplish with a word in that blank?

Then we look more closely:

Does any part of the word look familiar - and is that resemblance reliable?

What are two meanings the word might have? Which one fits the passage better?

Only then do we check the dictionary.

The dictionary is not there to reward a good guess. It is there to test the student’s reasoning.

If the inference was wrong, we go back and find where it broke.

This takes longer than supplying a definition. That extra time gives the student practice resolving uncertainty.

Over time, something changes.

Students stop treating “I don’t know” as the end of the process. They begin to use it as a location:

This is where I need to look, compare, and test.

This is one small part of what I mean by Functional Learning: teaching the operation, not supplying the result.

Parents often ask what they can do at home.

Keep reading. Keep answering questions. But don’t answer every question immediately.

The next time your child asks what a word means, give them a moment. Ask them to reread the sentence with a blank in its place. Let them describe what the missing word appears to be doing. Let them form a hypothesis.

Then check it together.

The goal is to help students recognize confusion and know how to resolve it.

Readers aren’t born. They’re built.

In his New York Times essay “There’s Really Only One Way to Stop Students From Cheating,” Adam Grant argues that the answer is to build students’ moral identity: make cheating inconsistent with the person they want to be.

I think he’s right about character.

But I think he dismisses another response too quickly.

Grant notes that many teachers are returning to in-class essays and oral exams. “Creating obstacles is never going to be enough,” he writes.

If their purpose is only to make cheating harder, he’s right.

But an oral exam is not primarily an obstacle.

It is a measurement.

I’ve spent twenty-five years teaching students one-on-one. I often ask a student to put a passage or assignment away and explain it - in their own words. Then I ask how the parts connect. Why does one point follow from another? What evidence would change the conclusion?

The difference between following an explanation and being able to reconstruct one is often striking.

AI makes that distinction more urgent. It did not create it.

A finished assignment has always been indirect evidence of learning: we look at the work and infer that the student did the thinking required to produce it.

AI didn’t make learning impossible to measure. It made that inference impossible to trust.

A study published in Science this year, based on responses from more than 95,000 students, called for universities to reform assessment rather than rely on blanket bans or detection systems.

The problem is not only that students can break the rules more easily. It is that the work we collect may no longer tell us what they understand.

This also changes the incentive to cheat.

If the finished paper is what earns the grade, students will optimize the paper. If they also have to explain, reconstruct, defend, or apply the idea, outsourcing the thinking becomes much less useful.

A five-minute check can be simple:

Close the laptop.

Put the work away.

Tell me what you argued - in your own words.

Why does the second point follow from the first?

What would weaken your argument?

The purpose is not to catch the student. It is to find out whether the student can do the thinking.

That distinction - between completing a task and developing the capacity the task was meant to exercise - is central to my Functional Learning work.

Grant is right that character matters. But oral assessment is more than a way to prevent cheating.

It is one way to find out whether the work produced understanding.

Great readers seldom understand every word they read.

But they always know which ones they don’t.

Recently, a student and I came to this sentence in an article:

“After weeks of campaigning, the once-tireless candidate appeared enervated at the final rally.”

She was confident she understood it. She had the sentence backward.

“Enervated” looks as though it should mean “energized.” Especially given the context. But it doesn’t. It means drained.

Because she misunderstood one word, she took the opposite meaning from the sentence — and that misunderstanding colored everything she read after.

What interests me is that this mistake isn’t random. Again and again, students make the same wrong guess.

A confident wrong guess is worse than “I don’t know.” The student doesn’t merely miss one word. They build the wrong scene, then interpret every sentence that follows through it.

The student who stops and says, “I don’t know what enervated means,” is reading better. They flag the gap, keep it open, and proceed carefully.

The student who guesses seals the gap with the wrong material — and never looks back.

In classrooms, we often reward confidence while leaving too little room for students to discover — and admit — what they don’t understand.

Our priorities are reversed.

Students learn the lesson underneath the lesson: Never reveal a gap.

So by high school, when I ask, “What does that word mean — in your own words?” the last thing a student wants to say is “I don’t know.”

They would rather guess. Guessing at least looks like knowing.

I work to reverse that lesson, one session at a time.

With my students, “I don’t know that word” isn’t a failure. It’s the most useful sentence a student can say — the exact location where comprehension can grow.

Once that change takes hold, students stop performing certainty and start noticing their own gaps — mid-sentence and unprompted.

They become comfortable sitting with ambiguity long enough to resolve it instead of papering over it.

That’s when comprehension starts compounding.

Not because they suddenly know more words, but because they’ve stopped pretending to.

Familiarity isn’t fluency.

Confidence isn’t comprehension.

Readers aren’t born. They’re built.

Rose Horowitch’s Atlantic cover story makes a compelling case that America is entering a postliterate era.

I would extend her logic one step further: even when students do read, we routinely mistake recognition for comprehension.

Her piece, “The End of Reading Is Here,” documents the decline. From 1984 to 2025, the share of 13-year-olds who said they rarely or never read for fun rose from 8 to 29 percent.

I’ve spent twenty-five years teaching students one-on-one, and the difference between finishing a passage and understanding it is hard to miss.

When I teach reading comprehension, I routinely pause and ask: “What does this word, sentence, or paragraph mean - in your own words?”

Even with the text open, the task is difficult for many students. Once I close the book, it often becomes nearly impossible.

I often remind my students:

Familiarity isn’t fluency.

More reading still matters. But students who cannot explain a passage need more than additional pages. They need explicit instruction in how to build meaning from a text.

When students know how to work through a difficult passage, they understand more, persist longer, and return to reading more willingly. Depth and volume begin to reinforce each other.

Schools teach decoding explicitly through phonics and fluency instruction, but often leave the operations of comprehension implicit.

Then they assess comprehension as though students should have discovered the method for themselves.

Comprehension can be taught. I teach it as translation:

1. Close the text.

2. Rebuild the idea in your own words.

3. Explain how the parts connect.

4. Check your explanation against the page.

5. Where your explanation breaks, return and revise.

This is the right kind of friction: the student has to produce the explanation, not merely recognize it.

This practice is one part of Functional Learning, which develops three core literacies: reading to learn, organizing knowledge, and sequencing complexity. In instructional terms: comprehension as translation, critical thinking as structure, and problem solving as sequence.

These are not traits students either possess or lack. They are operations that can be named, practiced, and assessed.

I help students, families, and schools build these capacities - so students read more willingly and carry more understanding away from what they read.

Readers aren’t born. They’re built.