I’ve given some version of the same parent education talk for 10 years, and every few years I have to throw out the middle part to update it with whatever tech-related topic is most top of mind.
I host a parent education night every fall––an evening when families crowd into the multipurpose room ready for me to tell them what to worry about. In the early years, it was screen time: how many hours are acceptable and when to take the phone away at night. Then came social media and cyberbullying: a harder topic, because now the harm had faces and names. Next was digital citizenship: footprints, reputations, and being kind in the comments section. Every fall a fresh anxiety would crest, so I’d rebuild the talk around it, and parents would head home clutching a slightly different list of apps to fear.
This year, of course, it’s AI. And in rebuilding the talk yet again, I finally saw what I’d been missing.
Underneath every one of those nights, the parents were really asking a single question: How much should I let them? And every year technology professionals like me give them a number, a limit, a filter, a contract. A dial to set correctly. But the dial never measured what these parents were actually afraid of—not the amount, but the direction.
It finally clicked for me in our faculty room last spring, listening to teachers wrestle with AI the way we’d once wrestled with phones. It was never how much. It was: Is the child still the one in charge? What we’re all afraid of is children losing their agency: their ability to think for themselves, make their own choices, and take charge rather than be led by technology.
Once I could see that, I did what technology directors do when something won’t leave them alone. I built a model for it.
The AI Triangle
I call my model the AI Triangle, which puts AI at the center of three questions:
1. Literacy: Can the child understand and check what the AI produces, or do they just believe it?
2. Citizenship: Are they acting fairly and safely with it, toward other people and toward themselves?
3. Wellness: Have they noticed what it’s doing to their attention, their thinking, their inner life?

The AI sits at the center on purpose. Each earlier parent worry pressed on a single corner: screen time is mostly wellness, cyberbullying is mostly citizenship. AI is the first thing I’ve had to explain that presses on all three at once—one tool that finds information, generates content, moves through a child’s social world, and holds a conversation. That’s why the old single-issue talks always felt incomplete; each described one corner of a shape nobody had drawn yet. And because every corner asks about the child’s relationship to the technology, the framework outlasts any single tool—today’s AI included.
The triangle isn’t a score to tally or a limit to set; it’s three questions to walk through whenever a child picks up an AI tool. Can they check what it told them, or are they just believing it? Are they using it fairly—toward classmates, toward the truth, toward themselves? And have they noticed what it’s doing to their attention and mood? The value of the shape is that it answers which corner a given moment is really about, so a shapeless “How much AI is too much?” becomes a question someone can actually act on.
What the Research Tells Us
I’m a technology director, not a researcher, so I went looking to find out whether the research backed up my hunch that, with any technology, what children do with it matters more than how much time they spend on it. It does. Two of thelargest reviews of screens and adolescent well-being I read—a 2019 article in Nature Human Behaviour’s 2019 and a 2020 article in the Journal of Child Psychology and Psychiatry—put the effect of screen time at real but small, small enough that what a child does with a tool matters far more than how long they spend on it.
Two children can spend the identical hour on the identical app and walk away in completely different places—one more scattered and anxious, the other more curious and connected. The clock reads the same; the activity is everything. That gap between what and how long is exactly what the triangle is built to make visible. It turns a vague worry about screens into three specific questions to ask.
With AI, the same finding only gets louder. One student uses a chatbot to outsource the thinking inherent in an assignment; another uses the same tool to sharpen it. There’s a term for outsourcing thought: “metacognitive laziness,” coined in a 2025 British Journal of Educational Technology article about students who leaned on ChatGPT. When this occurs, a child hands over not just the task but the act of noticing whether they’ve understood anything at all. Yet a 2025 article in the Forum for Linguistic Studies found that a chatbot used with structure can produce measurable gains in the exact critical thinking we’re afraid of losing.
One tool, opposite outcomes. The variable was never the technology; it was whether a human was refining their thoughts, not outsourcing them. That’s the most hopeful thing I get to tell a room full of anxious parents—the intervention that matters most isn’t something they buy. It’s teaching their children to run the triangle themselves—to ask, of any AI tool, whether they can check it, whether they’re using it fairly, and whether they’ve noticed what it’s doing to them.
What Trusted Adults Can Do
Across every study I’ve read, the strongest protective factor is a child’s willingness to bring a problem to a trusted adult while it’s still small—an answer they suspect is wrong but are tempted to hand in anyway, a late-night AI “friend” that’s starting to crowd out real ones, a faked image moving through the group chat. So keep the channel open, and let the triangle give the conversation its shape.
For literacy, when they show something AI made, don’t grade it—get curious: “How could we tell if that’s actually true?” and then look it up together, once, so checking becomes a habit instead of a lecture. For citizenship, set the rules by task rather than by the hour: its fine to brainstorm with a chatbot, not fine to pass its work off as your own or to make an image of a real person without asking. For wellness, narrate your own use out loud—“I’ve been on this thing too long, I’m putting it in the other room”—because kids calibrate their own grip on these tools by watching how the adults around them hold theirs. Through all of it: ask, don’t interrogate, and meet what they show with curiosity rather than alarm, because alarm slams the door.
The worry puts on a new costume every fall, but underneath it’s always the same one: Are our children remaining the authors of their own attention, judgment, and relationships? AI didn’t invent that question. It’s simply the version of the technology that finally forces every one of us to see what mattered all along.
Not how much will we let them use it, but how do we make sure they are the ones in charge?