The communist regime in Poland made exactly one educational mistake: they forgot to censor mathematics.
Everything else they controlled. History was rewritten. Literature was filtered. Art served the party. But geometry proofs, logical reasoning, algebra — these carried no ideological connotation, so they left them alone. For decades, an entire population behind the Iron Curtain was taught, for free, how to think.
The regime didn't see the problem coming. When millions of people can follow an argument, an argument made from slogans stops working.
I know this because I was one of those students. I grew up in communist Poland, spent thirty-one years teaching mathematics to Canadian teenagers, and I watched the same pattern play out in reverse on this side of the Atlantic — logic removed from curricula, proofs eliminated, deep thinking quietly traded for surface-level output. Meanwhile, back in Poland, the habit of rigorous reasoning had helped hollow out a system that billions of dollars and military force couldn't crack.
The pattern matters because AI is now walking into classrooms at scale — and the same question is sitting on the table: will it teach people to think, or will it do the thinking for them?
What's Actually Happening in 2026
By 2026, roughly 83% of K-12 educators use generative AI in some form. The global AI-in-education market has surpassed $9.5 billion. Tools like Khanmigo and NotebookLM are no longer experimental — they're in the daily workflow of teachers from Toronto to São Paulo.
Khanmigo is worth paying attention to here. It does something the easy AI tools don't: instead of handing students answers, it uses Socratic questioning. Guided hints. Stepwise feedback. The machine is deliberately withholding the shortcut. That's not a limitation — that's a philosophy. The institutions seeing the strongest outcomes are the ones using AI to give teachers more time for mentorship and critical-thinking conversations, not less.
And yet — 70% of teachers worry that AI is weakening critical thinking and research skills in their students. Both things are true simultaneously. AI done right builds thinkers. AI done lazily produces people who can generate a paragraph but can't evaluate one.
The difference between those two outcomes is not the tool. It's the teacher.
The Bridge
Here's what I keep coming back to: the communist system accidentally ran a 40-year experiment proving that thinking is a teachable skill — and that a population trained to think deeply will eventually see through any system that relies on them not to.
The math teacher standing in front of a blackboard in Warsaw in 1975, walking students through a proof, wasn't trying to undermine the regime. She was just teaching math. But the cognitive habit she was building — follow the logic, question the premise, demand evidence — had consequences that outlasted the wall.
That is the irreplaceable thing a teacher does. Not deliver content. Not manage a classroom. But model how a mind works on a problem it hasn't seen before.
AI can personalize the pace. It can flag where a student is stuck. It can generate ten versions of an explanation until one lands. What it cannot do is sit across from a confused seventeen-year-old and say: I've seen this exact moment of confusion in a hundred students before you. Here's what's actually going on. That requires pattern recognition of a different order — not data, but experience. Not algorithm, but judgment.
Top 7 Ways Teachers Can Harness AI Without Losing the Thinking
**1. Use AI to free up time, then spend that time on thinking conversations.**
The strongest outcomes in 2026 come from schools where AI handles admin and planning so teachers can run Socratic discussions. The ratio matters.
**2. Treat AI-generated answers as first drafts, not final answers.**
Train students to critique, revise, and push back on AI output. If they accept it uncritically, you've built a dependency, not a skill.
**3. Choose tools that ask questions, not just answer them.**
Khanmigo's Socratic model exists for a reason. When evaluating AI tools for your classroom, ask: does this tool make students think harder, or less?
**4. Teach the pattern behind the problem, not just the solution.**
This is mathematics at its core. AI can solve the equation; it can't teach why the structure of the equation matters. That's your job.
**5. Use AI assessment data to identify gaps — then intervene yourself.**
AI is excellent at flagging where a student is struggling. The follow-through — the conversation, the reframe, the encouragement — still requires a human who knows that student.
**6. Make the process visible.**
Show students how you think through a problem. Narrate your reasoning out loud. AI can produce polished output; teachers model messy, honest thinking in real time.
**7. Protect the hard parts.**
Don't let AI eliminate the cognitive struggle that builds long-term capability. The friction is the point. A student who works through confusion builds a mental muscle that shortcuts destroy.
The Regime Didn't See It Coming. Don't Make the Same Mistake.
The Iron Curtain eventually cracked not because of military pressure or economic collapse alone, but because decades of rigorous education had produced a population capable of reasoning its way to the truth. The regime handed them the tool — mathematical logic — and assumed it was harmless.
In 2026, we are handing students and schools a tool far more powerful than anything the communist planners imagined. The question is the same one that was always on the table in my classroom: are we building thinkers, or are we building people who can produce output on command?
The answer is up to the teacher in the room. It always was.
Bogdan Marzewski, Author of 'From Communist Paradise to Freedom'