GM. This is The Great Unlock, your weekly cheat sheet for the AI revolution. We filter out the noise, red pen fake news, and give out stickers to those deserving.

Today's deep dive: a Fordham Institute piece just diagnosed a "teaching recession" behind the learning recession we already knew about. The diagnosis is right. The prescription is half right. Let's unlock it.

Let’s unlock it.

The Mechanism

Mike Schmoker's argument, laid out this week in the Fordham Institute's Flypaper, is simple and depressing: teachers have mostly stopped teaching. Not in the sense of showing up, grading papers, running a classroom. In the specific sense he means: "explicit, sustained, whole-class lessons." Instead, most classrooms have drifted toward facilitation — kids on screens, kids on worksheets, groups moving at their own pace while the adult in the room circulates and nudges.

Schmoker has receipts. A Connecticut superintendent and assistant superintendent audited their own new hires, found they'd come from seventeen different teacher-prep programs, and discovered that not one had received practical training in "checking for understanding" — supposedly one of the vital elements of instruction. Two Minnesota teachers, writing in this same publication back in 2020, said it plainer: their ed schools trained them in everything except how to actually teach, and warned them against being caught at the front of the room during observations.

None of this is controversial among people who've spent time in classrooms. Direct, structured, whole-class instruction has a real track record. Bloom's two-sigma research is decades old at this point. Studies of urban districts have found explicit instruction produces outsized gains precisely because it's rare. The evidence isn't new. The willingness of teacher prep to ignore it is what's remarkable.

Ed schools didn't forget how to teach teaching. They decided it wasn't their job.

The Structural Problem

Here's where we want to push past Schmoker's frame instead of just nodding along with it.

His argument — and the broader "explicit instruction" movement's argument — treats teaching as synonymous with delivery. The teacher stands up, presents content in a calibrated sequence, checks for understanding, reteaches what didn't land. That's the job. When it's done well, it works. When it isn't done at all, kids fall further behind. Fine. All true.

But go back further than the last twenty years of ed-school failure, and you find something worth sitting with. The word "teaching" doesn't come from a root that means "deliver information." It comes from the late Old English tecunge — "act of providing guidance" — and the Middle English tēchen, from Old English tæcan: to show, to point out, to declare, to demonstrate. The etymological center of gravity isn't a lecture. It's a guide pointing at the trail.

That's a genuinely different job than the one Schmoker is describing. "Sage on the stage" — the calibrated deliverer of content, checker of comprehension, re-teacher of missed material — is a real and valuable function. But it is not what the word originally meant, and it is not, I'd argue, the part of the job that's actually irreplaceable anymore.

Because here's the uncomfortable second half of the argument Schmoker doesn't make: AI is already better than the median classroom at exactly the thing he wants ed schools to train for. A well-built AI tutor delivers explicit, sequenced content without fatigue, checks for understanding after every single problem instead of every single unit, and reteaches instantly and infinitely, at whatever pace the individual kid needs — not the pace of thirty kids averaged together. The Connecticut superintendent couldn't find a teacher trained to check for understanding. An AI system checks for understanding by default, every time, for every student, and never skips it because it's Friday and the fire drill ran long.

The delivery-and-check-for-understanding loop that ed schools failed to teach humans is the exact loop machines already do better.

The Great Unlock Thesis

So here's the reframe. Stop trying to win back "explicit instruction" as the core of teaching. Let it go — hand delivery and mastery-checking to the systems that are already better at both, and free the actual humans in the building to do the part of the original word that AI can't touch: guide, show, point out, demonstrate what matters and why it's worth learning in the first place.

This isn't a concession to EdTech hype. It's a correction of a category error the field has been making for decades: treating "content delivery" and "teaching" as the same noun. They aren't. Content delivery is a subroutine. Teaching, at its root, is the relationship in which a more experienced person shows a less experienced one where to look and why to care.

And this is where we part ways with Schmoker's "vital element." He locates the missing piece of instruction in checking for understanding — a mechanical, procedural skill that (he's right) most new teachers were never trained to run. I’d argue the actually vital, actually neglected element sits one layer upstream of that: motivation. You cannot check for understanding in a kid who has already decided the material isn't worth understanding. Checking for understanding without motivation is just more precisely measuring apathy.

Teacher colleges don't just fail to train "checking for understanding" — as the article documents. They fail, arguably worse, to train anyone in how to make a fourteen-year-old actually want to learn quadratic equations, or believe they're capable of learning them, or care whether they do. That's not a footnote to instruction. Bloom's own two-sigma work found that tutoring's biggest gains came bundled with something beyond pacing and feedback — engagement, formative correction, a relationship that made effort feel worthwhile. Strip the motivation out and even perfect delivery underperforms. Add motivation to imperfect delivery and outcomes climb anyway. That ordering matters.

If AI is going to own delivery and mastery-checking — and it increasingly can, at a fidelity most human classrooms have never achieved — then the highest-leverage thing a school can do with its newly freed-up teachers isn't to retrain them harder on the delivery loop. It's to finally build what ed schools never built: a real curriculum for how a teacher earns a student's motivation to learn.

Four questions for anyone running a school or system right now:

1. If an AI system delivered core content and checked for mastery at a pace matched to each student, what would you actually want your teachers doing with the time that frees up?

2. Do you have anything resembling a training program — not a workshop, a real mastery-based training program — in how teachers build student motivation? If Schmoker's Connecticut district couldn't find "checking for understanding" training across seventeen ed schools, what are the odds you'll find motivation training?

3. Are you measuring "engagement" as a compliance metric (seat time, completion rate) or as an actual signal of whether a kid wants to be there?

4. What would it cost you to keep treating content delivery as the core of the teaching job, once the tools that do it better than your best teacher cost less than your worst substitute?

Not sage on the stage. Guide on the side.

Sources: Mike Schmoker, "The learning recession might be caused by a teaching recession," Thomas B. Fordham Institute, 7.16.2026. Etymology per Online Etymology Dictionary, entries for "teach" and "teaching." Bloom, B.S., "The 2 Sigma Problem," Educational Researcher, 1984.

Old English tæcan — to show, point out, declare, demonstrate — became Middle English tēchen, became late Old English tecunge, "act of providing guidance." Somewhere between there and a 2026 teacher-prep syllabus, "guidance" got swapped out for "delivery," and nobody ever voted on the change.

Not deliver of content. The motivator.

This week's target: "Checking for understanding is the vital element of instruction."

This is the Fordham piece's actual claim, sourced from a Connecticut district's audit of its own new hires: teachers weren't trained in the "most vital elements of instruction — like checking for understanding." It's a reasonable diagnosis of what's missing from teacher prep. It is not, we'd argue, a correct diagnosis of what's vital.

Checking for understanding is a measurement tool. It tells you whether content landed. It says nothing about whether the student wanted it to land in the first place. Treating it as the vital element mistakes the thermometer for the fever. You can build the most rigorously calibrated checks-for-understanding protocol in the country and it will tell you, with great precision, that a demotivated kid still doesn't understand — over and over, unit after unit, with excellent data quality and zero improvement in outcomes.

Motivation is upstream of comprehension, which is upstream of any check for it. Skip it, and "checking for understanding" becomes an expensive way to confirm what you already suspected: the kid checked out before the lesson started.

Ask any working teacher which is harder to teach a new colleague — how to run a formative quiz, or how to make a bored quiet kid decide, on his own, that he wants to try. There's no contest. And there's no seventeen-district audit finding that ed schools taught the second one either, because almost none of them try.

Not checking for understanding. Checking for why they'd want to.

Ms. Alvarez teaches eighth-grade algebra in a district piloting an AI tutoring platform this year. For the first month, she ran her class the way she always had: warm-up, direct instruction, guided practice, exit ticket. Textbook Schmoker. It worked about as well as it ever had — good for the kids who were already motivated, a grind for the ones who weren't.

In month two, she handed the delivery loop over. The AI tutor took first pass at explicit instruction on new content, walked kids through worked examples at their own pace, and flagged mastery gaps in real time — the exact checking-for-understanding function the Connecticut district couldn't find trained anywhere in seventeen programs. Ms. Alvarez stopped standing at the front of the room.

What she started doing instead: sitting down, one-on-one and in small clusters, with the kids the dashboard flagged as disengaged rather than just behind. Not reteaching the content — the system had that covered and was doing it more precisely than she could, one student at a time. Asking why a kid had stopped trying. Connecting a lesson on exponential growth to a kid's actual interest in how his gaming clan's subscriber count worked. Sitting with a girl who'd quietly decided math "wasn't for her" and spending twenty minutes on nothing but why that story wasn't true.

Her mastery-check data didn't change much between month one and month two — the AI was already handling that better than she had. What changed was who showed up and stayed engaged long enough for the checks to mean anything. Attendance in her afternoon block went up. So did the number of kids voluntarily doing extra practice sets at home, unprompted, because for the first time someone had made the case for why it mattered to them personally rather than just checking whether they'd absorbed it.

Ms. Alvarez didn't get better at delivering content. She stopped delivering it and became, for the first time since she was hired, an actual teacher, in the true sense of the word.

Not a content deliverer. A mentor.

 ________________________________________________________

That's all the unlock for today. Tune in next week.

Stay awesome, you unlockers!

Want more? Order The Great Unlock here.