Why Your Child’s Conversation Skills Matter More in the Age of AI, Not Less

Something happened in the same week in late May 2026 that I don’t think got enough attention. The US government issued a formal health advisory warning that children’s screen time had become a public health crisis. One week later, the country’s second-largest teachers’ union called for banning AI chatbots from elementary schools entirely. Two completely separate institutions, looking at the same evidence, arriving at the same conclusion. Something is being crowded out. And I’ve been thinking about what that something is, because I think it’s the skill I’m most worried about for my daughters.

The government now officially classifies children’s screen time as a public health crisis, and AI chatbots are part of what triggered it.

Adolescents today average more than four hours of screen time daily, and nearly half of them lose track of how much time they have spent on devices.^[1] More than four hours. That is more time than most kids spend on homework, eating meals, or playing outside combined. The advisory targets the entire digital ecosystem, including what it specifically calls “chatbots and other AI interfaces,” as part of the problem.

What changed isn’t just the amount of time kids spend on screens. It’s the nature of what is filling those hours. AI companions are fundamentally different from video games or social media. They respond. They listen. Character.AI alone has 20 million monthly users, and more than half of them are under the age of 24.^[3] The number of AI companion apps available surged 700% between 2022 and mid-2025.^[3]

This is not a future concern. By the time the Surgeon General issued this warning, 70% of American teens had already used generative AI.^[4] The American Federation of Teachers responded by proposing a total ban on AI in elementary schools and a prohibition on social companion chatbots for anyone under 16.^[2] These are not fringe positions. These are mainstream institutions responding to something they are watching happen in real time.

The issue is not that AI is bad for learning. It’s that AI makes real conversation feel unnecessary at the exact ages when kids most need to practice it.

Here is what I noticed at home. My youngest, who is seven, is completely comfortable asking our home assistant anything. She asks what the fastest animal is, whether butterflies sleep, what would happen if you drank seawater. And she gets answers instantly. Clear, confident, immediate answers. No hesitation, no “I don’t know,” no impatient look.

Then she tries to explain something to her sister, who is nine. And suddenly everything is different. Her sister talks back. Gets confused. Misunderstands something. Asks her to repeat herself. And my youngest, who has endless patience with the AI, has very little patience for this. She gets frustrated almost right away. I don’t think this is a character flaw. I think it’s a training effect. When most of your questions get answered by a tool that never pushes back, you don’t get practice at what real human conversation actually requires: the misunderstandings, the clarifications, the constant small adjustments. Those moments of friction are exactly where the skill gets built.

Researchers have a term for what happens when a tool does something so reliably that you stop building the underlying capacity yourself: deskilling.^[3] For adults, this might mean losing the ability to navigate without GPS when it is always available. For children, who are still forming these abilities for the first time, the concern runs deeper. The pathways that support reading social cues, managing conversational repair, and interpreting tone in real time develop through practice. If that practice is quietly replaced by AI interaction, those pathways may not get the exercise they need during the years they are most ready to form.

The skill is in-person communication: the ability to hold a genuine conversation with a real human, read their reactions, adjust in real time, and recover when something goes wrong.

In-person communication sounds obvious. Everyone knows how to talk. But there is a real difference between talking and communicating. Communicating means reading the signals coming back at you. It means noticing when someone looks confused before they say anything. It means adjusting your words and pace in real time. It means tolerating the discomfort of a conversation that is not going the way you planned, and keeping going anyway.

AI is specifically designed to eliminate that discomfort. It is patient. It rephrases when you are confused. It doesn’t get frustrated. It doesn’t have bad days. The APA has noted that AI chatbots’ “overly supportive and overly agreeable nature may lead children to be stunted in development of empathy skills.”^[3] For information retrieval, frictionlessness is exactly what you want. For building communication skills, that same frictionlessness is the problem.

In-person communication is a skill that scales. A kid who can hold a genuine conversation, read a room, handle talking to someone who disagrees, and recover from an awkward moment has an advantage in nearly every area of life: friendships, school presentations, future job interviews, and close relationships. AI cannot coach this from the outside. You can only build it by doing it, repeatedly, with real people who have their own feelings, reactions, and bad days.

Your kid is building in-person communication skills when they can navigate a conversation that doesn’t go the way they expected, without shutting down or reaching for a device.

One clear sign is comfort with silence. Kids who have had plenty of practice with real conversation learn that silence in a discussion is not failure. It is thinking. They wait. They hold the space. When my middle daughter is upset and I ask her what’s wrong, she sometimes takes a long time to answer. She’s not ignoring me. She’s finding the words. That is a skill she built through actual conversations, not through talking to a chatbot that fills every pause with an instant response.

Another sign is repair. Every real conversation breaks down at some point. Someone gets confused, or hurt, or talks past each other. Kids who have practiced genuine conversation develop the instinct to notice when something went wrong and try again. This looks like “I think I said that wrong” or “Wait, let me try that differently.” That instinct comes from repetition with real humans, not from practice with AI tools that always understand you perfectly the first time and never require you to adjust.

The most effective thing you can do is create more time for genuine back-and-forth conversation, especially conversation that has some friction in it.

Create a no-device conversation window every day. This doesn’t have to be long. Even 20 minutes at dinner, in the car, or before bed where everyone is present and talking creates real practice. The key is that it’s genuinely back-and-forth, not one person talking while others listen. Ask your kid a question you actually don’t know the answer to. Let the conversation go somewhere unexpected.

Let conversations stay awkward when they are. When your kid is stuck, resist the urge to fill the silence or smooth things over. Silence is part of real conversation. Let them search for the words. When a miscommunication happens, don’t immediately fix it. Let your kid notice it and try to repair it themselves. That moment of discomfort is where the skill gets built.

Have your child explain something to a family member instead of looking it up. When your child learns something new at school, have them explain it to you or to a sibling over dinner. Not the AI. Not a search engine. Them, in their own words. Watch what happens when the listener gets confused. Can your kid adjust? Rephrase? Try again with different words? That loop is the real practice, and it only works with a real human on the other end.

The Surgeon General’s advisory and the AFT’s proposal both focus on reducing screen time. But the solution isn’t just less AI. It’s more of something specific: more unscripted, in-person, back-and-forth conversation where your kid has to do the real-time work themselves. That can happen around your dinner table tonight.

I am not anti-AI. I use it every day, and my daughters do too. But I have started thinking more carefully about which parts of their development need some protection from the friction-removal that AI does so well. In-person communication is one of those things. The friction, the awkwardness, the moments when something doesn’t land right, those are the reps that build the muscle.

My youngest asked our home assistant something the other day and got a perfect answer instantly. Then I asked her to explain the same thing to me. She stumbled. She repeated herself. I played dumb at one point and she had to try again with different words. By the end, she had figured out how to explain a concept to a real human who didn’t already know it. That is harder than talking to AI. That is also the skill that actually matters.