I Let My 8-Year-Old Teach an AI for 10 Minutes. Here's What Happened.

I almost didn't do it.

It was a Tuesday evening, dinner dishes still on the counter, my 8-year-old Maya bouncing off the walls with leftover energy. I'd been meaning to try this AI game a friend had mentioned — something called Teach the Robot — but I kept putting it off. Too tired. Too complicated. Maybe on the weekend.

That Tuesday, I ran out of excuses.

I pulled it up on the iPad, handed it to her, and said: "See if you can figure this out."

What happened in the next 10 minutes surprised me more than anything she'd done in school all year.

Minute 1–2: Confusion (Expected)

Maya stared at the screen. A little robot was asking her to teach it how to sort objects — showing her examples of what belonged in each category, waiting for her to drag items into the right buckets.

"What am I supposed to do?" she asked.

"Teach it," I said.

"Teach it what?"

"How to sort. Show it examples."

She dragged one item. Then another. The robot made a guess. Got it wrong.

"It's not very smart," she said, not unkindly. The way you'd say it about a puppy.

Minute 3–5: The Lightbulb

She started dragging faster. Showing it more examples. The robot's guesses improved. She noticed.

"Wait — it's getting better."

"Why do you think that is?"

She paused. Really thought about it. "Because I showed it more?"

"Exactly."

The look on her face at that moment — I wish I'd taken a photo. It wasn't the look of a child who'd been told something. It was the look of a child who'd figured something out.

Minute 6–8: The Hard Question

Then the robot made a mistake. A clear one. And Maya got frustrated.

"Why did it do that? I showed it the right answer!"

"What do you think went wrong?"

She chewed on this. "Maybe... I didn't show it enough times?"

"Maybe. Or maybe the examples you showed it weren't different enough from each other."

She looked at me. "So it only knows what I teach it?"

"Exactly."

Another pause. Longer this time.

"So if you teach it something wrong... it thinks wrong?"

I felt a chill. Not the bad kind. The kind you get when a conversation goes somewhere you didn't expect.

"Yes," I said. "That's exactly right."

Minute 9–10: The Question That Stayed With Me

She played for another minute, then put the iPad down and looked at me with that particular 8-year-old seriousness that means something important is coming.

"Who teaches the real AIs?" she asked. "The ones on our phone?"

I didn't have a simple answer. I told her it was lots of people — engineers, researchers, sometimes regular people who don't even know they're doing it. I told her those AIs make mistakes too, for the same reasons her robot did. I told her that's why it matters to understand how they work.

She nodded slowly. Then: "Can we do more tomorrow?"

What Those 10 Minutes Actually Taught

I went to bed that night thinking about what had just happened — and what it meant.

My daughter had, in 10 minutes, grasped concepts that universities teach in semester-long courses:

• Supervised learning — machines learn from labeled examples
• Training data quality — garbage in, garbage out
• Algorithmic bias — if the training data is flawed, the model is flawed
• The limits of AI — it only knows what it's been taught

She didn't know those terms. She didn't need to. She understood the ideas — which is exactly the point of AI literacy.

Why This Matters More Than Any App on Her iPad

My daughter uses AI every day. YouTube's recommendation algorithm. Autocorrect. The filter on her school's search engine. Siri mishearing her and doing something completely different.

Before that Tuesday, she experienced all of it as magic — mysterious, unknowable, just the way things work.

After 10 minutes of teaching a robot, she understood something fundamental: AI is not magic. It's math plus decisions made by people. And if the people make bad decisions, the AI does too.

That's not a lesson I could have taught her with a lecture. It took a game. It took her hands on it, her frustration, her lightbulb moment.

It took 10 minutes.

Your Turn

Teach the Robot is free. No account needed. Works on any tablet, phone, or laptop. It takes exactly as long as you have — 5 minutes or 50.

If your child is between 5 and 12, try it tonight. Don't explain too much. Just hand it over and say: "See if you can figure this out."

Then watch what happens.

Play Teach the Robot — free, no account needed →

And if you want to go deeper — our grade-band packs take children from basic sorting all the way through AI ethics, bias detection, and policy-making. One-time purchase. Permanent access. No subscriptions.

Because the question Maya asked me — "Who teaches the real AIs?" — deserves a real answer. And the earlier children start asking it, the better.

HiKIDAI is an AI literacy curriculum for kids aged 5–18. Browser-based games and printable activities aligned with AI4K12, ISTE, and UNESCO standards.I almost didn't do it.

It was a Tuesday evening, dinner dishes still on the counter, my 8-year-old Maya bouncing off the walls with leftover energy. I'd been meaning to try this AI game a friend had mentioned — something called Teach the Robot — but I kept putting it off. Too tired. Too complicated. Maybe on the weekend.

That Tuesday, I ran out of excuses.

I pulled it up on the iPad, handed it to her, and said: "See if you can figure this out."

What happened in the next 10 minutes surprised me more than anything she'd done in school all year.

Minute 1–2: Confusion (Expected)

Maya stared at the screen. A little robot was asking her to teach it how to sort objects — showing her examples of what belonged in each category, waiting for her to drag items into the right buckets.

"What am I supposed to do?" she asked.

"Teach it," I said.

"Teach it what?"

"How to sort. Show it examples."

She dragged one item. Then another. The robot made a guess. Got it wrong.

"It's not very smart," she said, not unkindly. The way you'd say it about a puppy.

Minute 3–5: The Lightbulb

She started dragging faster. Showing it more examples. The robot's guesses improved. She noticed.

"Wait — it's getting better."

"Why do you think that is?"

She paused. Really thought about it. "Because I showed it more?"

"Exactly."

The look on her face at that moment — I wish I'd taken a photo. It wasn't the look of a child who'd been told something. It was the look of a child who'd figured something out.

Minute 6–8: The Hard Question

Then the robot made a mistake. A clear one. And Maya got frustrated.

"Why did it do that? I showed it the right answer!"

"What do you think went wrong?"

She chewed on this. "Maybe... I didn't show it enough times?"

"Maybe. Or maybe the examples you showed it weren't different enough from each other."

She looked at me. "So it only knows what I teach it?"

"Exactly."

Another pause. Longer this time.

"So if you teach it something wrong... it thinks wrong?"

I felt a chill. Not the bad kind. The kind you get when a conversation goes somewhere you didn't expect.

"Yes," I said. "That's exactly right."

Minute 9–10: The Question That Stayed With Me

She played for another minute, then put the iPad down and looked at me with that particular 8-year-old seriousness that means something important is coming.

"Who teaches the real AIs?" she asked. "The ones on our phone?"

I didn't have a simple answer. I told her it was lots of people — engineers, researchers, sometimes regular people who don't even know they're doing it. I told her those AIs make mistakes too, for the same reasons her robot did. I told her that's why it matters to understand how they work.

She nodded slowly. Then: "Can we do more tomorrow?"

What Those 10 Minutes Actually Taught

I went to bed that night thinking about what had just happened — and what it meant.

My daughter had, in 10 minutes, grasped concepts that universities teach in semester-long courses:

• Supervised learning — machines learn from labeled examples
• Training data quality — garbage in, garbage out
• Algorithmic bias — if the training data is flawed, the model is flawed
• The limits of AI — it only knows what it's been taught

She didn't know those terms. She didn't need to. She understood the ideas — which is exactly the point of AI literacy.

Why This Matters More Than Any App on Her iPad

My daughter uses AI every day. YouTube's recommendation algorithm. Autocorrect. The filter on her school's search engine. Siri mishearing her and doing something completely different.

Before that Tuesday, she experienced all of it as magic — mysterious, unknowable, just the way things work.

After 10 minutes of teaching a robot, she understood something fundamental: AI is not magic. It's math plus decisions made by people. And if the people make bad decisions, the AI does too.

That's not a lesson I could have taught her with a lecture. It took a game. It took her hands on it, her frustration, her lightbulb moment.

It took 10 minutes.

Your Turn

Teach the Robot is free. No account needed. Works on any tablet, phone, or laptop. It takes exactly as long as you have — 5 minutes or 50.

If your child is between 5 and 12, try it tonight. Don't explain too much. Just hand it over and say: "See if you can figure this out."

Then watch what happens. Play Teach the Robot — free, no account needed →

And if you want to go deeper — our grade-band packs take children from basic sorting all the way through AI ethics, bias detection, and policy-making. One-time purchase. Permanent access. No subscriptions.

Because the question Maya asked me — "Who teaches the real AIs?" — deserves a real answer. And the earlier children start asking it, the better.

HiKIDAI is an AI literacy curriculum for kids aged 5–18. Browser-based games and printable activities aligned with AI4K12, ISTE, and UNESCO standards.