If you believe your child’s ability to think well will shape the rest of their life, this matters

In a 2025 MIT study, researchers wired up the brains of some willing students, over a number of months, while they wrote essays.

One group was given large language models (LLMs) to use, like ChatGPT. One group was not.

The group using LLMs showed significantly less brain activity — and they struggled to recall what they’d written. But the really worrying part came next: when the LLM group was asked to write another essay, this time without AI, their brains didn’t fully switch back on.

Writing is a full-brain workout. LLMs offload much of that work. And like a muscle, if you stop using your brain, it weakens.

But here’s the crucial nuance the headlines miss: At the end of the study, the “brain-only” group were allowed to use an LLM. And their brain activity increased. They used AI to stretch their thinking, not replace it.

This implies that if children meet AI before they’ve built their thinking muscle, they may never build it. But used later and in the right way, it becomes a force-multiplier for a well-trained mind.

This isn’t an argument to ban AI. It’s an argument to teach kids how to think first — so that when they use AI, they’re curious, critical and engaged, not passive and dependent. That’s the mission behind The Park.