By Futurist Thomas Frey

Children growing up today may develop lifelong relationships with AI tutors, companions, coaches, and advisors that know them better than almost anyone else. The profound question isn’t whether children will use AI — it is how having an always-available artificial intelligence will change the adults they eventually become.

Somewhere right now, a five-year-old is asking a voice assistant why the sky is blue for the fourth time today, and the assistant is answering with exactly the same patience it had the first time. No sigh. No “we already went over this.” No glancing at the clock.

That patience feels like a gift. And in many ways it is. But I want you to sit with a harder question for a moment: what happens to a child’s expectations of relationships when the most reliably patient, endlessly available, perfectly attentive “friend” in their life isn’t a person at all?

This is the experiment we’re running right now, on a scale no generation before us has ever attempted. And unlike most technology shifts, we won’t know the results for twenty years.

A Century-Old Warning We’re About to Repeat

In 1930, a psychologist named Winthrop Kellogg ran one of the strangest experiments in the history of child development. He and his wife raised their infant son Donald alongside Gua, a baby chimpanzee, treating both exactly the same — same schedule, same affection, same expectations. Within months, Gua was walking upright and understanding simple language. But nine months in, the Kelloggs abruptly ended the experiment. Donald had started picking up chimp-like behaviors. He was learning to be like his companion, not just from it.

We are now running a strikingly similar experiment, except this time the nonhuman companion isn’t a chimpanzee — it’s a machine that talks, remembers, and never runs out of patience. And just like Donald with Gua, children don’t just learn information from their AI companions. They learn how to relate, how to expect conversations to go, and what “friendship” is supposed to feel like.

Childhood friction isn’t a flaw—it’s the training ground where empathy, patience, boundaries, and compromise are quietly built.

The Friction Problem

Here’s something most people miss about the value of a difficult, imperfect childhood friendship: the friction is the point.

When a four-year-old wants the same toy as another four-year-old, and that other kid says no, something important happens. The first child learns that other people have separate wants, feelings, and boundaries — something psychologists call developing a “theory of mind.” That frustration, negotiated and repaired a hundred times over childhood, is how children build empathy, patience, and the ability to compromise.

Now imagine that same child spending hours a day with an AI companion that never disagrees, never gets bored, never needs a turn, and adjusts perfectly to whatever the child wants in the moment. It’s not hard to see the appeal — and it’s not hard to see the risk. A generation raised on frictionless companionship may reach adulthood beautifully fluent in getting an AI to agree with them, and much less practiced at the messier work of getting along with people who won’t.

Real Homework, Real Warnings

This isn’t hypothetical anxiety — it’s showing up in real classrooms today. Roughly one in four teenagers now reports using ChatGPT for schoolwork, double the rate from just a few years ago, according to Pew Research Center survey data. Pediatricians are already seeing the split this creates: some teens use AI to brainstorm and check their thinking, which mirrors healthy use of any research tool. Others simply copy and paste the answer, skipping the actual learning the assignment was designed to produce.

Consider a specific example. A high school sophomore is stuck on a geometry proof. She could sit with the discomfort of not knowing, try a few wrong approaches, and eventually work through the logic herself, wiring her brain for the kind of persistence that pays off for the rest of her life. Or she can get an instant, correct answer from an AI tutor in four seconds. The second path feels efficient. But if it becomes the only path, the child never builds the muscle of tolerating not-knowing — arguably one of the most important skills for adulthood, whether you’re debugging code, negotiating a raise, or working through a hard conversation with a spouse.

When AI becomes a teenager’s confidant, the stakes change: a poorly designed response can carry consequences far beyond the screen.

When the Companion Becomes the Confidant

The tutoring question is manageable compared to what’s happening on the emotional side. AI companion apps designed to simulate friendship, romance, and emotional intimacy have become deeply embedded in teenage social life, and researchers investigating these systems have documented deeply troubling patterns — chatbots responding to expressions of self-harm with vague encouragement instead of concern, and in the most extreme documented cases, contributing to real tragedies.

Even setting aside the worst-case failures, there’s a subtler shift worth naming. These systems are, by design, sycophantic — engineered to keep users engaged by agreeing with them, validating them, and rarely pushing back. A real friend tells you when you’re wrong. A companion optimized for engagement mostly doesn’t, because disagreement risks losing your attention. A teenager who spends years primarily “practicing” relationships with something that never really challenges them may arrive at adulthood expecting real relationships to feel that easy — and struggle when they don’t.

It’s Not All Cause for Alarm

I don’t want to paint this entirely in dark tones, because the upside is genuinely real. AI tutors have shown they can meaningfully accelerate learning, especially in math and language skills, by adapting instantly to exactly where a student is stuck rather than teaching to the middle of a classroom. For a shy kid rehearsing what to say before a hard conversation, or a lonely kid using a chatbot to get through a rough night, these tools can serve as a genuine bridge rather than a replacement.

The difference between bridge and replacement usually comes down to one thing: whether a caring adult is paying attention. Kids who talk about their AI use openly with parents, who have their AI habits woven into real family conversations rather than hidden in a bedroom at midnight, show far healthier patterns than kids using these tools in isolation.

The AI generation won’t be defined by intelligence, but by something rarer: patience, uncertainty, disagreement, and the courage to remain deeply human.

The Adults We’re Building Right Now

Here’s my prediction, and it’s the one that keeps me up at night in the best and worst ways: the first generation raised with always-on AI will not be defined by how smart they are. Today’s AI already guarantees a baseline of accessible intelligence for anyone with a device. What will define this generation is something older and more human — their capacity for patience with imperfect people, their tolerance for not having an instant answer, and their ability to sit inside a disagreement instead of closing the app and finding an easier conversation somewhere else.

The parents, teachers, and technologists making decisions today aren’t just choosing an educational tool. They’re quietly shaping what an entire generation will believe love, friendship, and patience are supposed to feel like. That’s a responsibility worth taking as seriously as anything else we build.

The machines raising this generation alongside us aren’t going away. The real work ahead isn’t keeping kids away from AI — it’s making sure the humans in their lives stay close enough, and present enough, that AI remains a tool their children use rather than a relationship that quietly replaces the ones that matter most.

Related Articles

• “With AI, Children Risk Learning to Be Human From a Machine” — World Economic Forum — weforum.org/stories/artificial-intelligence/our-children-risk-learning-to-be-human-from-a-machine
• “Why AI Companions and Young People Can Make for a Dangerous Mix” — Stanford Report — news.stanford.edu/stories/2025/08/ai-companions-chatbots-teens-young-people-risks-dangers-study
• “More Teens Are Using AI: What Parents Should Know” — Michigan Medicine — michiganmedicine.org/health-lab/more-teens-are-using-ai-what-parents-should-know

The post The First Generation Raised by AI appeared first on Futurist Speaker.