Raising Kids in the Age of AI

Something feels off lately. I sit down to write and the first sentence won’t come. Instead of just starting, I open a prompt window.

There’s a concept in cognitive psychology called cognitive offloading - delegating the work of remembering and reasoning to external tools. Calendars, notebooks: fine. But a 2011 study published in Science (Sparrow et al.) showed that when people know information is stored somewhere retrievable, the brain stops bothering to remember the content itself.[1] It only remembers where to find it. Not the information. Just the path to it.

I’m now outsourcing the very start of my thinking to AI. It’s not just tool dependency. It’s a change in how I think. If there’s such a thing as mental muscle, I can feel it quietly atrophying.

And as a parent, I worry about my kid.

Psychologist Robert Bjork spent decades demonstrating what he called “desirable difficulties,” the idea that the right amount of struggle during learning is what builds lasting memory and genuine understanding.[2] What comes easy is forgotten easily. The process of getting stuck, getting it wrong, and trying again is what actually changes the brain.

My generation took that friction for granted. When we didn’t know something, we looked it up the slow way. We learned things the wrong way first. That wasn’t inefficiency. That was learning.

Now a generation is coming that won’t have to do any of that.

AI researcher Andrej Karpathy put it plainly: “The goal is that the students are proficient in the use of AI, but can also exist without it.”[3] We still teach mental arithmetic even though calculators exist, because you need to understand the underlying logic to catch when the tool is wrong. AI is no different.

So what’s the mental arithmetic of the AI era?

Developmental psychologist Adele Diamond’s research shows the same thing over and over. The foundation of a child’s learning and social ability is executive function: working memory, inhibitory control, cognitive flexibility.[4] And these capacities develop explosively between ages three and seven.

The best way to build them is surprisingly simple. Physical activity, progressive challenge. Getting dirty. Falling down. Games with rules. Things AI can’t replace.

The most honest parenting I can do for a child turning four next month might be exactly this. Before letters, before AI. The ability to endure, to focus, to try, to fail, and to try again.

“Let’s try again!” - the power to say that.

AI summarizes everything so well that there’s no real reason to read the source anymore. And when people do read, they don’t read deeply. Add doomscrolling on top of that. Our attention spans, even as adults, are already wrecked.

Researcher Ying Xu’s team at Harvard showed that children interacting with AI made significantly fewer attempts to steer or lead the conversation compared to when they talked with people.[5] A tool that responds too perfectly stunts initiative, focus, and literacy alike. Muscles you don’t use will waste away.

So I want my child to experience deep absorption - often. I don’t know exactly what to teach her yet. But this much is clear.

Blocks, dirt, bugs, pretend play - she needs the experience of falling into something and seeing it through (though I’ll admit, sometimes I wish she’d wrap it up a little sooner lol).

Deep absorption is one of those rare moments that builds both executive function and initiative at once. No matter what new tools keep coming, the resilience to stay steady is ultimately made from experiences like these.

References

1. Sparrow, B., Liu, J., & Wegner, D. M. (2011). Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips. Science, 333(6043), 776–778.
2. Bjork, E. L., & Bjork, R. A. (2011). Making Things Hard on Yourself, But in a Good Way: Creating Desirable Difficulties to Enhance Learning. In M. A. Gernsbacher et al. (Eds.), Psychology and the Real World. Worth Publishers.
3. Karpathy, A. (2025, May). Post on X.
4. Diamond, A. (2013). Executive Functions. Annual Review of Psychology, 64, 135–168.
5. Xu, Y., Hu, Z., Laughter, L., Park, S., Kreijkes, P., & Warschauer, M. (2022). Learning with AI characters: Does character persona matter?. Child Development, 93(5), 1397–1415.