Edition 373 | August 13, 2026
The Dyslexic AI Newsletter by LM Lab AI

Reading Time: 10 minutes | Listening Time: 13 minutes

I'm going to say this too bluntly first, because the softer version hides the point.

Studying is cooked.

Not learning. Learning is not cooked. Learning may be more important than it has ever been.

But studying, as the thing education worships, measures, rewards, and morally praises, is cooked.

The future of education is not going to care nearly as much about whether you studied.

It's going to care whether you can show proof of work.

And that's a much bigger change than it sounds like.

For most of my life, education ran on a simple assumption.

If you studied, you learned.

So we built a whole system around the visible signs of studying. Did you read the chapter. Did you take notes. Did you fill out the worksheet. Did you sit in the room. Did you spend the time. Did you underline the right parts. Did you turn something in by Friday.

Those things were never the point. They were proxies.

A proxy is what you measure when the thing you actually care about is too hard to measure directly.

We cared whether a student understood the material, could use it, could think with it, could carry it into a new situation. But that's messy. It takes time. It requires judgment. It doesn't fit neatly into a gradebook.

So we measured the easier thing.

We measured studying, and then we started believing the proxy was the point.

For a long time, that was good enough to keep the machine moving.

AI didn't invent the problem with studying. It just made the problem impossible to ignore.

A student can now summarize the chapter without reading it. Generate the study guide without studying. Produce the five-paragraph essay without wrestling with the idea. Make flashcards, quizzes, outlines, discussion posts, and reflections at a speed that makes the old signals look ridiculous.

This is where people panic and say, "How will we know if they did the work?"

That question is telling on us.

Because it means the thing we were calling education was already built around proving a student performed the ritual.

Did they do the work.

Not did the work change them. Not can they use it. Not can they build with it. Not can they explain it when the situation shifts.

Did they do the work.

AI exposed how much of school was checking for the behavior of learning instead of the evidence of learning.

That's the part that's cooked.

Studying is an input.

Learning is a change in capability.

That sentence feels simple, but sit with it and half of education starts wobbling.

Studying says, "I spent time with the material."

Learning says, "I can now do something I could not do before."

Those are not the same claim.

You can study for three hours and learn almost nothing. You can also learn something important in eight minutes because the right explanation finally hit the right part of your brain.

Every dyslexic person reading this already knows that.

We know what it feels like to spend twice as long with the material and still get measured as if the time itself should have guaranteed the result. We know what it feels like to understand the big idea and lose points anyway, because the proof had to come through spelling, handwriting, or timed reading, some tiny hallway our brain was never built to walk through neatly.

School loved studying because studying looked fair. Everybody gets the same chapter. Same test. Same hour.

But sameness isn't fairness. Sometimes sameness just means everybody has to squeeze their thinking through the same keyhole.

Homework is studying's enforcement arm. Same broken assumption, except now it follows your kid home.

I'll be honest about my bias here. I've never liked homework. Not as a coach, not as an educator, and definitely not as a parent.

My kids are in school for something like six or seven hours. Then they come home, and instead of getting my kids, I get a second shift as an unpaid assistant teacher for a subject I didn't teach, using a method I wasn't shown, on a worksheet I didn't assign.

It's been a battle in my house for years. And the worst part isn't the battle itself. It's what the battle costs.

That time after school is the time I actually have with them. That's the window. That's when the real conversations happen, the ones that come out sideways while you're driving somewhere or making dinner. And instead we're at the table, both frustrated, fighting over a packet that exists mostly to prove somebody did something.

We are trading the highest-quality hours of a family's day for a compliance signal.

Here's my actual position, and it's not complicated.

If a kid is in school for seven hours and can't learn what they need to learn in seven hours, that's not the kid's problem to solve at 8pm. That's a school design problem. Fix the seven hours first. Fix the pacing, the class size, the delivery, the model itself. Then come talk to me about whether we need to borrow from the evening too.

Homework is what a system assigns when it won't examine its own schedule.

And this is exactly why I've been such a proponent of the flipped classroom model for so long. Flip it. Let the passive intake, the lecture, the reading, the explanation, happen on the kid's own time, at their own speed, with pause and rewind and captions and whatever else they need. Then use the seven hours you already have for the part that actually requires other humans in the room: the practice, the questions, the coaching, the arguing, the reps.

That's not radical. That's just putting the expensive resource, teacher attention, on the part that needs it.

And it matters twice as much for kids like mine and kids like me. A dyslexic kid doing homework alone at a kitchen table at 8pm is doing the hardest possible version of the task at the worst possible hour with the least available help. We've engineered that. Then we call the resulting struggle a work-ethic problem.

Proof of work does to homework what it does to studying. It stops asking whether the packet came back and starts asking what the kid can do now.

If a kid can show me what they built, solved, taught, or figured out, I don't need a worksheet to tell me they learned. And if they can't show me any of that, the worksheet was never going to tell me anyway.

So what replaces all of it?

Proof of work.

And no, I don't mean crypto. Nobody panic. Put the blockchain down.

I mean visible evidence that learning changed what you can do.

A thing you made. A problem you solved. A process you improved. A person you taught. A prototype you tested. A decision you can defend. A mistake you can explain. A revision history that shows your thinking got better.

Proof of work isn't just "turn in a final product."

That gets gamed too. AI can make polished final products all day. Sometimes too polished, honestly. Suspiciously polished. Like a seventh grader suddenly writing like a McKinsey consultant who found a thesaurus in a thunderstorm.

Proof of work has to include the trail.

What did you start with? What did you try? Where did it fail? What did you change? What feedback did you get? What did the tool do? What did you decide? What can you do now that you couldn't do before?

That's the new transcript. Not a list of classes completed. A record of capability gained.

This could be very good for dyslexic thinkers.

I want to say "could" loudly, because it's not automatic.

If education hears "proof of work" and turns it into more essays, more slides, more polished documents, we'll just recreate the same measurement problem with better fonts.

But designed right, proof of work opens the room.

Because proof of work can be a voice memo. A diagram. A prototype. A customer interview. A coached explanation. A video walkthrough. A before-and-after. A spreadsheet that actually helped someone make a decision. A messy sketch that shows the pattern before the words catch up.

For a lot of us, the best thinking never showed up cleanly on the old scoreboards. We could see the system, make the leap, connect the strange dots, solve the practical problem, and still get marked down because the written explanation limped across the finish line covered in spelling mistakes.

Proof of work replaces "can you perform school correctly" with "can you make your learning visible in a form that matches the work."

That's a different world.

This doesn't make teachers less important. I think it makes good teachers more important.

If the old job was delivering content and checking compliance, AI is going to eat a lot of that.

But if the new job is designing challenges, judging evidence, coaching revision, spotting shallow work, asking better questions, and helping students turn effort into capability, then teachers matter enormously. Maybe more than before.

The teacher stops being a gatekeeper of information and becomes a coach watching practice.

And coaches know something schools sometimes forget. You don't learn by watching the drill. You learn by doing the rep, getting feedback, adjusting, and doing it again.

The rep is the proof. The adjustment is the learning. The final score was never enough.

If I were rebuilding school from scratch, I'd stop asking "did you study" and start asking:

What did you build with this? Who did you help with this? What problem can you solve now? What changed between your first attempt and your last? Where did you use AI, and what judgment did you bring that it couldn't? What can you explain without the tool in your hand, and what can you do better with it?

That last pair matters most.

We don't need to pretend the tools don't exist. That isn't rigor. That's nostalgia wearing a hall pass.

The future isn't "no AI." It's knowing what the human is responsible for when AI is in the room.

That responsibility is judgment, taste, direction, context, ethics, revision, and the nerve to look at a good-looking output and say, "no, that's not it yet."

You can't prove that with a worksheet. You prove it by working.

And this isn't only a school thing. For years a credential did the talking for you. It said you completed the path, sat in the rooms, passed the tests. That still matters in places, and please don't let your surgeon swap medical school for a vibes-based portfolio.

But in more and more fields the question is shifting to: show me what you've built. Show me how you think. Show me your judgment under constraints. Show me the receipts.

If you're an adult who didn't thrive in school, that should get your attention. You may have spent years believing you were bad at learning when you were actually just bad at being measured by school. Those are not the same thing.

Take one thing you're currently studying, researching, or "trying to learn."

Stop treating understanding as the goal. Understanding is invisible. It can hide forever.

Build the smallest thing that proves the learning changed your capability, then revise it once.

Learning AI? Build a workflow that saves you twenty minutes. Learning marketing? Write one offer and send it to five real people. Learning finance? Explain one number in your business well enough that someone else can make a decision from it. Learning a tool? Solve a real problem with it before you watch the fifth tutorial.

Don't wait until you feel ready. Ready is usually just studying in a nicer jacket.

And if you're a parent: tonight, ask your kid what they can do now that they couldn't do last month. See if either of you can answer. That answer is worth more than the packet.

"I am trying to learn [topic]. I do not want a study plan. I want proof of work. Ask me what I need this knowledge for, what real-world situation I want to handle better, and what constraints I have. Then design three small projects that would prove I am actually gaining capability. For each one, tell me what artifact I should produce, what process trail I should keep, how I should use AI without letting it do all the thinking, and what evidence would show that I learned something."

Run that before you ask for another reading list.

The reading list isn't bad. It's just not the proof.

The proof is what changes after the reading.

Matt "Coach" Ivey

Founder, LM Lab AI | Creator, The Dyslexic AI Newsletter

Dictated, not typed. Obviously.

Topic coming soon. Stay tuned.