Kids Can Feel AI Hurting Them, They Have to Use It Anyway

I recently read a newly released Gallup poll about Gen Z. Two numbers stood out as particularly interesting to me, especially with regard to how AI is already affecting childhood development. The first was that 80 percent of Gen Z say AI is likely to make future learning harder. The second was that those who believe they will be prepared to use AI after high school rose 12 points in a year to over 50 percent. Together, these numbers describe something oddly in conflict.

For the past year, I’ve written about the structural problem of AI in education. The architecture of generative AI systems creates reasonable conditions of foreclosure during different cognitive development windows when those capacities are supposed to form. This new Gallup poll showed me a different layer. Children inside these systems are feeling worried, yet confidence in their future proficiency is rising. They have been absorbing the contradiction that the adult world has yet to resolve for itself.

This is not teenage optimism but cognitive dissonance.

The Environment Creates the Contradiction

Dissonance has a behavioral shape. Gen Z teens who use AI every day reported an 18-point drop in excitement about AI over the year. These weren’t the skeptical users. They were the heaviest. Every prior consumer technology I have watched move through a population has the opposite shape. The people on the tool every day are its advocates. They selected into the habit. They like it. Some of them like it too much. They do not, as a rule, sour on it faster than the people who barely use it.

Across the full Gen Z sample, excitement is down 14 points. Hope is down 9. Anger is up 9. These young adults are: “Not convinced that AI enhances creativity or critical thinking, and the majority believe it may come at a cost, particularly to learning.” The teenagers are saying it.

The environment the teenagers are living in is actually not produced by the technology itself. It is produced by specific decisions. School boards adopt policies. Administrators authorize platforms. Vendors market AI fluency as preparation. Employers signal it. Universities revise the curriculum. Policymakers write AI literacy into standards without specifying what literacy means when the tool is undermining the cognitive capacities literacy depends on. Each of these actors makes different decisions, but collectively, they build this room. Our children are forced to grow up in it.

Inside the Windows

The brain builds different cognitive abilities in different stages of time, and the windows during which those capabilities form are biological. Take, for instance, the 7 to 12 window where foundational skills like reading fluency, math automaticity, and logical reasoning are supposed to become automatic. Then, there's the window from middle school through high school where identity, abstract thinking, and metacognition are refined. The 18 to 25 window is when the prefrontal cortex completes the scaffolding for independent self-regulation, decision making, and judgment. Capacities that get practiced during these windows find a permanent home in the mind. Capacities that are not practiced never really get comfortable.

Stanovich (1986) gave us the language for what happens when foundational skills fail to automate within this schedule. He called it the Matthew Effect. Later learning is not impossible, but it is permanently more effortful. Initial deficits compound on themselves as the child grows.

Children need to feel that their own judgment shapes the world they live in. That feeling is what fuels motivation and what teaches them to trust their own observations as evidence worth acting on (Ryan and Deci, 2000).

Children can hold mixed messages from adults for a while. What they cannot hold for long is the sense that their own evidence, their own lived experiences, their own attention, their own learning, and their own bodies don’t count and don’t make a difference.

The teenager who says AI is likely to make her learning harder and then reports rising confidence, she will be prepared to use it, is neither lying nor confused. She is adapting. She has watched the adults around her say one thing and build another. She is learning what her signals are worth.

Remaining Question

The teenagers are reporting, in more than one register, that the tool the adults around them require them to use during the developmental windows when their cognitive capacities are forming is hurting that formation. The research supports them. The developmental biology supports them. Children are not reasoning independently from nothing. They are listeners first. What they tell us is filtered through the conversations they have been hearing. That filter is why their voices should be part of this conversation. They show us what is already being absorbed.

The teenagers in this poll are absorbing two positions at once. AI is hurting their learning. AI is what they need for their future. Which position wins in their minds depends on who they listen to, and for how long, during the years that matter most for forming judgment. Right now, the adults talking the loudest are the ones building the room. The children inside it are quieter, and they are reporting what the room feels like from the inside.

Does it matter if they are correct? We can either treat what children are feeling as evidence that they understand their own development, or we can keep pushing them, more efficiently each year, to live inside a contradiction that we will not resolve.

Right now, what I see is the second.