Why I Built Still In Charge

I spent two years as a para-educator working with high school students the system had mostly given up on. Students who had been told, in a hundred small ways, that the material was not for them. Not in those words. The message arrived in how the content was delivered, the speed it moved, the vocabulary it assumed they already had. By the time I met them, most of them had agreed with the verdict. They were not bad at the subject. They had been handed it in a form built for someone else, and then quietly counted as the ones who could not keep up.

I think about those students every time I read another headline about AI in schools.

Right now the conversation about young people and AI has settled into one question: how much should we restrict it. How early, how often, whether it belongs in the classroom at all. States are passing rules. Districts are writing policies. Teachers unions are calling for limits. Every one of those is a real response to a real worry, and I am not here to tell anyone their worry is wrong.

But the restriction conversation is answering a different question than the one I care about. The variable everyone is debating is exposure: how much of the tool a kid touches. The variable I keep watching is the one almost no one is naming: what the kid has been shown the tool is for.

Those are not the same thing, and the difference is the whole reason Still In Charge exists.

Here is what I mean. For most of history, serious knowledge has been locked behind three doors. You needed the credential to get into the field. You needed the vocabulary to read it. And you needed the time and money to absorb it from the inside. Those barriers were never about whether ordinary people were smart enough. They were about access. The knowledge was on the shelf; most people simply could not reach it.

This is the first tool in history that opens all three doors at once. A student who never thought of herself as a science person can now learn the actual material of a field they were told they have no aptitude for, in language they already speak, at a pace they can hold, asking the same question five different ways until one of them lands. The barrier was never their capacity. It was the framing. And for the first time, the framing can adapt to the child.

That is an extraordinary thing. It is also fragile, because it depends entirely on one variable: whether anyone shows them that this is what the tool is for.

Because the same tool, encountered as something smaller, stays a novelty. A kid who only ever meets AI as a way to generate an avatar, finish a worksheet without reading it, or get an answer they never check has not been given a smaller version of the same gift; they have been given a different tool entirely, one that does the thinking so they do not have to. Same product. Opposite outcome. The only difference is the introduction.

And the introduction is not being handled. The platforms that reach kids first have no obligation to introduce a learning tool as a learning tool, and they do not. The schools are busy deciding whether to ban it. The parents are mostly learning it at the same time their kids are, from the same sources. So the default introduction, the one that arrives when no one chooses otherwise, is whatever the platform was optimized to show. For most kids, that is not the version that opens the three doors. It is the version that keeps them scrolling.

This is the part that keeps me up. The students who get shown the real thing will mostly be the ones whose parents already use it that way, or who land in a school that built understanding instead of a ban, or who happen to find the right corner of the internet at the right age. The students who do not get shown will inherit the smaller tool, not because anyone decided they should, but because no one decided they should not. The gap that produces is not a gap in access. Every one of them has the tool in their pocket. It is a gap in what they were taught it was for. And that gap will follow them into adulthood and into work, and it will look, from the outside, exactly like the old verdict: some kids just could not keep up.

I have seen that verdict be wrong too many times to let it get re-installed by accident.

So Still In Charge is not an attempt to restrict AI, and it is not an attempt to cheerlead it. It is an attempt to do the one thing the restriction debate skips: give a young person a way to stay the author of their own decisions while using a tool that is very good at making decisions for them.

That is why it is five principles and not forty pages. Use AI to sharpen your thinking, not replace it. Protect what is private. Verify what it gives you before you act on it. Follow the rules of the room you are in. And when something feels wrong, pause and ask. They are deliberately small enough that a fifteen-year-old can actually carry them into the moment that matters, when it is eleven at night, an assignment is due and there is no adult in the room. They are written as decisions the student owns, not rules imposed on them, because a rule a person does not understand only works while someone is watching, and no one is watching at eleven at night.

Underneath all five is one idea: AI can assist, but it cannot own the responsibility. That stays with the student. Not because a policy says so, but because it is simply true. The tool does not carry the consequence of the choice. The person does. Which means the choice was always theirs, whether or not anyone ever told them that.

A note to the three people this actually depends on.

To teachers. AI cannot replace you, and I want to be precise about why, because it is not a courtesy. The part of teaching that matters most is not the delivery of facts. It is knowing which student is quietly lost and why, deciding what a particular kid needs on a particular day, noticing the one who is struggling for reasons that have nothing to do with the material. A computer model does none of those things. It has no stake in the student and no judgment about them. What it can do is the one thing you have never had enough hours for. The best teacher in the building still has thirty or forty students in a room and several more rooms before the day is done, and no one, however gifted, can rebuild the same lesson forty different ways, at forty different paces, in the language each student actually thinks in. That was never a failure of effort. It is a limit of arithmetic. So you give the class the facts and the standard, and the tool can sit with one student and re-explain what you taught until it lands, or show them how to research it themselves. It does not replace the teaching. It extends your reach to the student you could not get to today. The judgment about what they need stays yours, because that is the part no tool has.

To parents. Start with what this tool can actually do for your kid, because the headlines rarely lead with it. In the right hands it is the most patient tutor they will ever have, one that will explain a concept they are stuck on, over and over again, without getting frustrated, without resentment, without somewhere else it needs to be. It can take a subject your child struggles with and meet them exactly where they are, and let them chase a real curiosity as far as it goes. That is genuinely new, and it is worth wanting for them.

The dangers are real too, and I am not going to wave them away. AI can produce confident, wrong answers a kid will accept without checking. It can absorb private information a teenager does not think twice about typing in. It can quietly do the thinking that their developing mind needs to do for itself. Those are not hypothetical. They are happening now.

But here is the part the screen-time debate keeps missing. Your kid is already using this. The question was never whether to allow it. And no parent, however vigilant, actually controls what a teenager with a phone and an internet connection does when the door is closed. You can set limits. You cannot be in the room. So the lever that actually works is not the one most of the energy is going toward.

The lever that works is the oldest one in parenting. You teach values. You teach responsibility. You teach that choices have consequences and that the consequence belongs to the person who made the choice. You do not need to understand AI, or even use it, to teach those things. They are exactly the skills a young person needs to use this tool well, and they are the skills almost no one is deliberately building for it. A kid who has been raised to verify before they trust, to protect what is private, and to own the call they made does not become a different person when they open an AI. That is the parenting that transfers. Still In Charge is just a vocabulary for it.

To students. Making a clip or a filter with AI is fun, and I am not going to pretend it is not. You are allowed to mess around. Everybody needs the frivolous stuff; that is not the problem.

Here is what no one is telling you plainly. The same tool you use to make a filter can teach you almost anything you have ever been curious about, in your own words, at your own pace, until it actually makes sense. The thing you were told you were bad at. The thing your school does not offer. The thing you are genuinely interested in but figured was for other people. You can chase your own curiosity instead of whatever an influencer is being paid to sell you that day. Almost no one shows you that side, so here it is.

And this is the part that matters most: your time online is your choice. Not your parents', not the app's, not the algorithm's. The app is built to keep you scrolling because that is how it makes money, and there is nothing wrong with you for enjoying it. But you get to decide how much of your attention it gets, and what you do with the rest. That is real power, and almost no one tells you that you have it. You do.

The students I worked with were never short on capacity. The kids who are shown what this tool is really for will gain something they never got handed. The one who is not will look, from the outside, exactly like they did: like someone who just could not keep up. I have seen that verdict be wrong too many times to watch it get written again by accident.

That is the whole purpose. Everything else is detail.