What Happens to Your Child’s Brain When AI Does the Hard Parts
A September 2026 report from the American Psychological Association synthesizes research from MIT, UNC Chapel Hill, and others on what regular AI use does to the developing adolescent brain. MIT researchers measuring electrical brain activity found that students using a large language model for essay writing showed weaker neural connectivity than those who wrote without it. A separate study found that frequent AI outsourcing raised anxiety and lowered distress tolerance over time. This is not about grades. It is about what happens to the brain during the one window when what kids practice most shapes what they can do for the rest of their lives.
A September 2026 American Psychological Association report synthesizes research finding that adolescents who regularly use AI for schoolwork may show weaker neural connectivity, reduced social-emotional development, and higher anxiety over time. MIT researchers used brain scans to measure the difference: students who used a large language model while writing essays showed less active communication between brain regions compared to those who wrote without AI. A separate University of North Carolina study found that frequent AI outsourcing raised anxiety rather than reducing it. Adolescence is a sensitive developmental window, and the habits kids form with AI right now carry more neurological weight than at any other point in life.
What this issue covers: Why psychologists are raising questions about cognitive offloading during the adolescent brain’s sensitive developmental window. Why the neural connectivity finding matters more than grades alone can show. Cognitive Resilience as the specific capacity worth protecting now. Two signs your kid is already building it. Three moves for this week. Plus The Cold Start, a 50-minute family game that puts everyone inside the experience of productive struggle intentionally.
More than half of American teenagers now use AI chatbots to help with schoolwork. That number stopped surprising me a while ago. What stopped me cold was a new report from the American Psychological Association asking a question most of us hadn’t thought to ask: what is all this doing to the developing adolescent brain? Not the grades. Not the effort. The brain itself. Psychologists have spent the past year running studies that point toward an answer. It’s not what I was expecting, and I think it matters.[1]
Psychologists now have evidence that regularly outsourcing cognitive work to AI may weaken the brain connections adolescents are supposed to be building during these years.
A new report from the APA’s Monitor on Psychology, published September 2026, brings together research from psychologists and neuroscientists studying AI’s effects on the developing adolescent brain.[1] The findings span three areas: how adolescents learn and retain information, how they develop social and emotional skills, and how they handle difficulty and anxiety. More than half of U.S. teenagers now use AI chatbots for schoolwork. The report isn’t about whether that number is surprising. It’s about what that number means for the brain doing the outsourcing.
Researchers at MIT used electroencephalography, technology that measures electrical brain activity, to monitor what happens inside participants’ heads while they write essays with and without AI assistance. Students who used a large language model during the task showed measurably weaker neural connectivity compared to those who wrote without it.[2] Their brain regions were communicating with each other less actively when AI was handling part of the cognitive load. The scanning picked up a difference that grades alone would never reveal.
The working concept is “cognitive offloading”: handing a mental task to an external tool before the brain has had to do much work itself. Done occasionally, that’s just using a tool wisely. Done repeatedly, across hundreds of homework sessions and study periods, researchers believe it may weaken the cognitive circuits that would otherwise develop through that effort. Like a muscle that rarely gets a full workout, the concern is whether those circuits end up less capable when the tool isn’t available.
Adolescence is a once-in-a-lifetime developmental window, and the brain is unusually moldable during these years, which means the habits kids form with AI right now carry more weight than they would at any other point in life.
Developmental psychologists use the phrase “sensitive period” to describe stages when the brain is especially shaped by experience. Adolescence is one of them. During these years, the brain actively prunes and strengthens connections based on what it’s repeatedly asked to do. That same mechanism lets a teenager absorb a new language, master a sport, or develop expertise in something they love. It also means the brain is shaped by what it doesn’t do, by the work it never had to do alone.
The social and emotional findings in the APA report are where I sat up the straightest. Researchers are finding that AI companions, tools that respond instantly, never push back, and never create awkward silences, may limit adolescents’ development of empathy, conflict resolution, and perspective-taking. Not because teenagers are choosing AI over people. But because a portion of the friction and discomfort that makes human interaction hard is also what teaches young people how to navigate it. The training ground for those skills isn’t comfortable conversations. It’s the uncomfortable ones.
Outsourcing hard things didn’t make life feel easier. It made sitting with difficulty harder.
Perhaps the most counterintuitive finding is about anxiety. Researchers at the University of North Carolina found that young adults who frequently outsourced tasks to AI didn’t report lower stress. They reported higher anxiety over time and showed signs of declining distress tolerance.[1] The less often their brains had to work through discomfort, the less capable they became at tolerating it. Outsourcing hard things didn’t make life feel easier. It made sitting with difficulty harder.
The skill is Cognitive Resilience: the capacity to stay in difficulty long enough for the brain to do its own work, rather than outsourcing every moment of struggle to a tool.
Cognitive resilience isn’t willpower in the old-fashioned sense. It’s a developed capacity, built through repeated experiences of starting a hard thing, staying in it, tolerating the friction, and getting through it. Every time a child works through a problem without immediately reaching for AI, that circuit gets exercised. The research suggests this is more important than it sounds, because the alternative may not just be short-term dependence but a measurable reduction in what the brain can do on its own.
This is what makes the teenage years feel particularly high-stakes right now. The adolescent brain isn’t just learning content. It’s building infrastructure. My 13-year-old working through a frustrating essay, my 9-year-old stuck on a word problem, my 7-year-old untangling a conflict with a friend: each of those moments is doing something to the brain’s architecture that goes beyond the specific problem being solved. The research is beginning to show that AI, when used as an escape hatch rather than a scaffold, may shortcut that infrastructure-building.
Common Sense Media found that 72 percent of children have now used social AI, and roughly one in three of those children have discussed a serious personal problem with an AI instead of a human.[3] That number is worth sitting with. Not because AI responses are harmful, but because talking to a human about something hard, tolerating the awkwardness, reading the room, navigating disagreement, is practice. Adolescence is the developmental window when that practice is most formative. And it’s also the window when AI is most convenient as an alternative.
A child building cognitive resilience tends to reach for help only after they’ve genuinely tried first, and can tolerate frustration long enough to stay in a hard thing before it gets easier.
The most visible sign is what they do in the first few minutes of a hard task. A child building this skill doesn’t immediately open a chatbot or ask someone else to solve it. They sit with it. They might complain about it. They might stare at it. But they stay. You might notice they talk through the problem out loud, try a wrong approach first, or write something imperfect before asking for guidance. Last week my youngest spent four minutes staring at a math problem she clearly hated before she figured it out. I watched her and said nothing. She was genuinely proud of herself in a way she wouldn’t have been if I’d stepped in after thirty seconds.
Socially, look for how your kid handles small friction with friends or siblings. Children building this skill don’t shut down or immediately avoid when things get uncomfortable. They feel the discomfort and push through a conversation anyway. They can tolerate being wrong in front of others. They can stay in an argument and come out the other side without needing it resolved on their terms. That capacity for social distress tolerance is trained exactly the same way cognitive resilience is: by not running from the hard thing.
The most effective thing you can do is create moments where your kid has to work through something difficult without AI as a first resort, then stay nearby without solving it for them.
Try a first-attempt rule for homework this week. Before your kid reaches for AI, they have to write one paragraph, attempt five problems, or sketch out a rough idea on their own first. This isn’t about banning a tool. It’s about making sure the brain gets its reps in before the shortcut kicks in. Two minutes of genuine effort changes what the brain does with the answer it eventually gets.
When you see your kid stuck and frustrated, resist the urge to fix it immediately. Try saying “let’s give it two more minutes before we look anything up” and then wait it out with them. That two-minute window, as uncomfortable as it can feel for both of you, is where the brain work actually happens. Sitting beside them without solving it is harder than it sounds.
Have one honest conversation this week about what your child uses AI for when things get hard emotionally, not just academically. Not as an interrogation, just as a check-in. If they’re turning to AI to process something that feels too heavy to bring to a person, you want to know that, and you want to be available as the alternative.
The research isn’t telling us to take away the tools. It’s telling us to be intentional about the moments we leave space for difficulty. Those moments are doing something the tools can’t replicate.
I am not trying to raise kids who avoid technology. I want my daughters to be genuinely fluent with AI, the same way I want them to be comfortable with any powerful tool. But fluency means knowing when to use something and when not to, and that second part only develops through practice. The practice of doing hard things without help. Staying in discomfort. Figuring out the next step without a shortcut available. That’s not an old-fashioned idea. According to the research, it’s what the developing adolescent brain actually needs.
The APA didn’t release this report to alarm parents. They released it because the data is accumulating, and adolescence doesn’t wait for the research to be settled. My kids’ teenage years are happening right now. I’d rather be paying attention and adjusting as I learn than waiting for the consensus to arrive. I suspect you feel the same way.
The Cold Start
Why This Activity Works
New research shows that the adolescent brain builds measurable cognitive capacity during moments of genuine difficulty, but only when it stays with the struggle long enough for something to shift. AI shortcuts the discomfort but also shortcuts the growth. The Cold Start puts your whole family inside that experience intentionally. It isn’t about producing perfect work. It is about recognizing what productive struggle feels like from the inside, so that choosing to stay in it next time is a recognizable choice rather than an invisible one. The 4-minute check-in matters too: pausing to name what you’re feeling without fixing it is distress tolerance practice in real time. That’s the same capacity the research says AI outsourcing is quietly eroding.
Ask This at Dinner
Listen for whether they can describe a shift in their thinking rather than just the outcome. A kid who says “I stopped trying to figure out the whole thing and just focused on one small part” has discovered something important about how their brain works under pressure. A kid who says “nothing clicked, I was just stuck” is also telling you something worth following: ask what stuck felt like at minute 2 versus minute 7. Both answers open the conversation. The goal isn’t a correct reflection. It’s the habit of paying attention to what happens inside when hard things stay hard.
Three free reads to go further
This kind of thinking,
delivered weekly.
Raised Nimble translates AI and learning research into practical guidance for parents. Free, every Friday. No fluff.
No spam. Unsubscribe anytime.