This as-told-to essay is based on a conversation with Lena Hall, a senior director of developer and AI strategy at Akamai Technologies and cofounder of an AI startup. It’s been edited for length and clarity.
My son woke up one morning determined to build a game. It wasn’t a passing idea; he talked about it all day in the way curious kids sometimes do.
I handed him paper and asked him to sketch it out. He sketched the levels, characters, and what he imagined it would feel like to play.
That’s when I thought we could try something. I remembered OpenAI’s Codex app had a dictation mode.
I wanted to see whether a voice-based coding tool could turn a young child’s idea into something playable, even without technical vocabulary.
My son was really excited about vibe coding a game
My son lit up the moment he realized we were going to try to make a real game.
I started by showing him the tool. I opened the app and explained it simply: It could take what he said and turn it into something on the computer. I showed him the dictation button, and he was immediately captivated by the bar that displayed sound waves.
He said he wanted a game where typing a word would turn the user into a character. He pressed the dictation button himself and sent the prompt.
He watched Codex work and asked why it was taking its time. At one point, he asked with complete sincerity, “Humans think faster, right mama?”
I laughed, as it was apparent to me at that moment that, in contrast to engineers appreciating how much AI accelerates work, a child doesn’t yet have the context of how much sophisticated reasoning or complex computations are being performed.
A few minutes later, it was done. Codex had generated a browser game, but not quite the way he’d imagined. I told him we could go back and ask Codex to fix it.
His next prompt was more specific: “I want names of specific characters. And I want the characters to jump up and collect coins as it goes.”
My role throughout was pretty simple. I explained the tool, asked questions that helped him say out loud what he already knew he wanted, and stayed present.
My child’s learning came from play
I didn’t sit my child down and say, “We’re going to learn about AI.”
He came to me with an idea, and I found a way to take that idea seriously. The learning was a consequence of the play, not the other way around.
When children lead with genuine interest, they absorb skills and concepts deeply because they’re motivated from within.
I saw articulation, persistence, and the ability to refine an idea after feedback — a set of skills that are hard to teach directly.
He also naturally broke the game into parts — characters, movement, and collecting coins — which is an early form of structured thinking.
That kind of sustained creative commitment in a young child is something I want to nurture, and this experience rewarded it.
How I made sure using AI wasn’t harmful at his age
The most important principle for us is that AI should never replace thinking. It should extend it.
The moment a child offloads their curiosity, decision-making, or creative struggle to an AI, the most valuable part of the experience disappears.
I try to make sure he understands, in age-appropriate ways, that AI isn’t a person.
It’s a tool made by people, not a friend or an authority. This matters because children are naturally wired to form emotional bonds and attribute human qualities to things that respond to them. Building that understanding early is foundational. It shapes how they’ll relate to these systems for the rest of their lives.
Before we started, we reviewed a simple rule. We don’t share personal information with apps — real names, where we live, school, or family details.
I’m also deliberate about screen time. We keep it limited and focused on creating rather than consuming. The whole session was under twenty minutes. He spent more time drawing the game on paper that day than he did building it.
Do you have a story to share about vibe coding? Contact this reporter at cmlee@businessinsider.com.
Read the full article here
