Thinkism and the Teacher's Dilemma: Why Understanding Follows Doing
I read Daniel Lemire’s recent essay “We see something that works, and then we understand it” and something cracked open in me. Lemire argues that real progress rarely follows the “understand first, then apply” model we were taught in school. Instead, we stumble upon something that works, and only then do we build the understanding around it. He calls the opposite belief (that pure thinking can solve any problem) thinkism, borrowing the term from Kevin Kelly.
I want to write about what happens when a teacher who was failed by thinkism tries to teach inside a system built on it. And why I believe the answer might involve AI as a tool for the curation problem that no one talks about.
The School That Failed Me
I was a student who hated school. Not because I was lazy or disengaged…the opposite was true. I was hungry to learn, but what I was given felt like eating raw ingredients instead of a meal. Every subject arrived as a set of abstractions, disconnected from any context where they mattered. I was told that “this is how it works” before I ever had a reason to care whether it worked at all.
Lemire puts it perfectly: the linear model of learning says Isaac Newton sat down, derived his three laws, and then the world changed. But that’s backwards. The pendulum clock came first in 1656. Hooke and Newton came after, trying to explain what they were already seeing work. Understanding followed practice, not the other way around.
I didn’t learn by being handed formulas. I learned (when I did learn) by being thrown into situations where something needed to happen, and figuring out how to make it happen. The moments that shaped me were never in the classroom. They were in the spaces between classes, in side projects, in conversations that weren’t part of any curriculum.
Thinkism in the Classroom
Lemire notes that thinkism works in school, and that’s the problem. The teacher gives you the concepts, then gives you a problem designed to use exactly those concepts. It feels like learning. It feels like understanding. But it’s a closed loop. The student learns to pattern-match: “which concept does this problem require?” Not “how does the world actually work?”
As a teacher myself now, I see this every day. Students are anxious when I put something on an exam that wasn’t explicitly covered. They’re not angry, they’re frightened. Because they’ve been trained to believe that knowledge is a finite set of answers delivered by authority, not a living thing you build by encountering the unknown.
And here’s the part that breaks me: I was once that student. I know what it feels like to sit in a lecture that made no sense, to memorize without comprehension, to pass the test and forget everything. So when I design a lesson, I try to do the opposite. I try to give students the problem first, the mess, the real thing, and then help them build the framework around it.
The 10x Curating Problem
But here’s what nobody tells you about teaching this way: it takes an absurd amount of effort.
I estimate that for every hour of effective, connection-driven instruction, I spend roughly ten hours curating, finding the right example, the right analogy, the right moment to intervene. A single lesson that lands might require:
- Scanning dozens of sources for a specific case study that resonates
- Knowing which student is struggling with which misconception
- Reading the room in real time and abandoning the plan when it’s not working
- Building the bridge between where they are and where they need to be (a bridge that is different for every single student)
This is the human connection part of teaching. It’s not scalable, not reproducible. A great teacher in front of thirty students is doing thirty different things at once, adjusting in real time, improvising based on micro-expressions and hesitation and the specific energy of a room on a Tuesday afternoon.
No lesson plan survives contact with the classroom. The magic is in the adaptation, and that adaptation demands deep knowledge of both the subject and the student.
Where AI Fits
I’ve been thinking about this: what if AI’s real gift to education isn’t as a tutor, not as a content generator, but as a curation engine?
The bottleneck for human teachers isn’t knowledge. We can explain things. The bottleneck is time and context. We don’t have time to find the perfect example for every student. We don’t have time to read every article, watch every video, and distill every book down to the one insight that a specific student needs right now.
AI could change that. Imagine a system that:
- Reads widely across sources (articles, research, books, projects, tutorials) and maps relationships between ideas
- Understands a student’s current level, interests, and misconceptions
- Surfaces exactly the right piece of content at exactly the right moment
- Helps the teacher prepare by doing the first 90% of the research, so the teacher can spend their energy on the last 10% (the human part)
This isn’t replacing the teacher. It’s giving the teacher superpowers. The curation is the unglamorous, essential work that makes the real teaching possible. AI can handle the scale of information; the human handles the scale of meaning.
Nobody Knows Anything
Lemire ends his essay with a line I keep coming back to: “nobody knows anything.” The world is so complex that even the smartest individual knows only a fraction of what there is to know, and much of what they think they know is slightly wrong.
This is terrifying for a student trained in thinkism. If knowledge is a tower you build from the ground up, and the foundation is shaky, what hope do you have? But if knowledge is something you build by doing (by seeing what works, by failing, by iterating), then not knowing everything is the starting condition, not the failure state.
As a teacher, my job isn’t to deliver complete knowledge. That’s impossible for anyone. My job is to create the conditions where a student can encounter a real problem, engage with it honestly, and build understanding from the inside out. And if AI can help me prepare for that (by curating, by connecting, by surfacing what matters), then I’ll take all the help I can get.
The real understanding still has to be earned. But maybe the path to it doesn’t have to be so lonely.
This article was written in response to Daniel Lemire’s “We see something that works, and then we understand it”.
Crepi il lupo! 🐺