I landed at Mirabel Airport with two suitcases and three hundred US dollars between my wife and me. One suitcase was full of cans of Polish meat, because when you leave a country where food is rationed, you do not arrive somewhere new with an empty stomach if you can help it. The border officer opened it, saw the cans, and his face did something I had not expected. Not suspicion. Not amusement. A kind of quiet recognition. He understood, without my having to explain it, where I had come from and what those cans meant.

He let me through.

I did not cry. I was too tired to cry. I was also too alert. In my country you learn very young that arriving somewhere new is not the end of anything; it is the beginning of the next round of figuring out who watches you and what rules you are already breaking without knowing it.

I have been thinking about that sentence for two weeks, because a whole generation is standing in an arrivals hall right now, and nobody has handed them the rules either.

What the people in the room just said

In late June, Microsoft published its annual AI in Education report, run by PSB Insights across 3,345 respondents in K-12 and higher education in six countries.

Two numbers belong side by side. 92% of students and education leaders, and 88% of educators, have already used AI for school-related purposes. But 77% of students, and 53% of educators, say they have not received formal AI training.

Near-universal use. Most of them never taught how.

And this is not a story about teenagers sneaking around. 87% of educators and leaders, and 79% of students, agree that using AI responsibly matters for students' futures. 66% of educators and 52% of students want their institution to provide training monthly or quarterly. They are asking for it.

That is not a discipline problem. That is a room full of people standing inside a system whose rules nobody wrote down, asking someone to please write them down.

Two weeks ago, one state finally did

On July 1, Ohio's deadline landed. Under House Bill 96, which added Section 3301.24 to the Ohio Revised Code, every public school district, community school and STEM school in the state had to adopt a formal AI policy. Ohio is the first state to turn AI guidance from a suggestion into a statutory duty.

Notice what the law does not do. It requires a policy. It does not prescribe what that policy says, does not mandate AI instruction, and does not require anyone to use AI at all.

A written rule beats a shrug, and Ohio moved while other states were still forming committees. But adopting a model policy makes a district compliant, not safe. The gaps lawyers are already pointing at, special education and AI detection tools used for discipline and the contracts governing student data, are exactly where a student gets hurt. Compliance is a signature. Safety is a practice.

The arrivals hall

When I walked out of that airport, Canada did not hate me. It simply did not explain itself. The rules existed, about work, about money, about how you speak to the person behind the counter, and every one of them was known by everybody except me. Nobody was cruel. Nobody sat me down either.

English is my fourth language. I never had time for ESL classes because I was too busy working and building a life. I did not have the luxury of a classroom. I learned to speak on the job and in the streets. I made it. I could also tell you exactly what it cost, and how much of it was luck.

The first thing Canada taught me was that freedom is not a destination. Freedom is a workload.

That is precisely what we have handed students. AI is enormous freedom: any question, any hour, no judgment, no queue. And freedom without the rules is not liberation. It is exposure. The 77% are learning it on the job and in the streets, finding out which rules they were already breaking at the worst possible moment. An integrity hearing. A failed exam. A teacher who simply stops believing them.

The difference is that I knew I had just arrived somewhere. They do not.

Seven rules worth teaching before the policy arrives

  • Draw the line between help and handover. Asking a machine to explain a step is study; asking it to produce the finished thing is not.
  • Make explaining it out loud the price of admission. If a student cannot say back what they submitted, with no screen in front of them, they did not learn it.
  • Teach that it is confidently wrong. Brookings reported in January that a typical chatbot runs around 70% accurate, and confidence is not accuracy.
  • Use it after the attempt, not instead of it. The struggle is not the obstacle to the learning; it is the mechanism.
  • Say plainly what the tool is not. Not a friend, not a counsellor, not private.
  • Ask it for the question, not just the answer. A machine will answer anything you ask it, but it will never tell you what is worth asking.
  • Write the rule down, even if it is one sentence. An unwritten rule is exactly the rule a student breaks without knowing it, and the student always pays.

What it comes down to

Be fair to the tool: AI genuinely helps when it is built to. In a randomised trial published in PNAS in 2025, roughly 1,000 high school math students got either a plain chatbot or the same model with guardrails that asked questions instead of giving answers. Sitting an exam alone, the plain-chatbot group scored 17% worse than students who had never touched AI. In the guardrailed group, that harm was essentially erased. Same tool. The design decided.

So the tool was never the question. The question is who does the explaining, and right now, for 77% of students, the answer is nobody.

Ohio wrote a rule, and I am glad it did. But a rule is a sign in the arrivals hall. It is not the person who stops, turns around, and tells you what you are about to get wrong.

So what if the real scandal is not that students are using AI badly? What if it is that we handed nearly eight in ten of them the most powerful learning tool of their lives, told them not one rule for using it, and then called what followed cheating?

I got out of a country that governed by unwritten rules everybody was still expected to obey. I did not expect to watch a generation handed the same arrangement, with better hardware.

Who is doing the explaining where you are?

Bogdan Marzewski, Author of 'From Communist Paradise to Freedom'

Bogdan Marzewski, Author of 'From Communist Paradise to Freedom'
Previous Posts
July 1, 2026 · AI & Learning

Teachers Say AI Is Bigger Than the Internet. They Also Say It's Eroding Trust.

A new poll of 545 teachers says AI will reshape school more than the internet did — while eroding critical thinking and trust. A thirty-year teacher on what that erosion looks like.

Read more →
June 17, 2026 · AI & Learning

The Best AI Tutors Refuse to Give the Answer

Khan Academy's AI tutor won't hand kids the answer — and the research says that's exactly why it works. A thirty-year math teacher on struggle, proofs, and what our schools quietly threw away.

Read more →
June 3, 2026 · AI & Learning

Mastering the Thinking Pattern: Why AI Needs a Teacher's Wisdom

AI is now in 83% of classrooms — but the teachers seeing real results aren't the ones handing it the keys. They're the ones who know what AI can't replace: the discipline of deep thinking. A retired…

Read more →
May 20, 2026 · AI & Learning

Resilience in Learning: From 1989 Scarcity to 2026 AI Abundance

In 1989 Poland, the communist regime accidentally taught an entire generation how to think — and that skill brought down the wall. In 2026, AI is doing something similar. Here is what a retired math…

Read more →