Ask AI with Ashley: Should I Let My 10-Year-Old Use AI?

Question: “Should I let my 10-year-old use AI? What kind of boundaries should I consider? Are there any parental controls I should know about?”

Screen time has been a topic of debate for all modern parents, and it starts way earlier than any of us would like to admit.

For at least the last 10 years, from the time a child is born, within the first 24 hours, they have likely already been introduced to a screen. Probably the first photo of your child to ever exist was taken with a phone.

And the phone use continues.

I am speaking from my own personal experience and realize this may not be true for every parent, but what I have come to learn over the last 40-plus years of my life is that, in many ways, I am pretty average.

From the time my child was born 10 years ago, his father and I have discussed when, and how much screen time was “healthy” or “appropriate” for our son, Finnley.

Before the age of 2, Finn had learned how to scroll. At four, he was watching cartoons and playing kids’ games when we were out to dinner. Around age 5, he taught himself how to “read” and “type” because he needed to do so for the games he wanted to play.

Of course, this was not real reading. It was more pattern recognition, and his dad was actually the one to sit with Finn and go through a book series designed to teach children how to read.

Finnley is now an excellent reader and, unlike his mother, an excellent speller.

Also unlike his mother, Finnley does not particularly gravitate toward the outdoors or imaginary play games. He is more like his father and interested in electronics and video games.

Whether I am being a good parent by giving my child electronics and internet access is a question I struggle with multiple times a day.

I am conflicted.

On the one hand, I am nostalgic for the past and long to see my child set up make-believe worlds where no TV or screen illuminates the background. On the other hand, I think the online world is his future, and the more exposure to and understanding he has of how to navigate it, the more prepared he will be for adulthood.

That being said, I have read the books and engaged in the ongoing dialogue around social media, online use and children’s mental health. I monitor Finn daily, looking for signs of depression or disengagement with the world. I wonder if, by the time I see them, it will be too late.

I hope not.

So, back to the questions at hand. And please remember: I am not a doctor of any kind. This is just advice from one mother to another.

Should I let my 10-year-old use AI?

Maybe that question is better narrowed to:

What kind of AI should I let my 10-year-old use?

Chances are good that your child is already being exposed to artificial intelligence, whether either of you realizes it or not. AI can exist behind educational software, recommendation systems and the algorithms deciding which videos or other content appear next.

That is different from handing a child an open-ended chatbot.

And the chatbot companies themselves don’t agree about where the line should be.

ChatGPT is not intended for children under 13. OpenAI says children ages 13 to 17 require parental consent, and if ChatGPT is being used educationally with a child under 13, the actual interaction must be conducted by an adult.

Microsoft Copilot requires users to be at least 13, with potentially higher minimum ages in some countries. Anthropic’s Claude takes an even more restrictive approach to individual consumer accounts. Character.AI went further still: after developing protections for teenagers, it ultimately removed open-ended AI chat for users under 18.

In other words, there isn’t even an industry-wide agreement about when a child is ready for conversational AI.

The ages between childhood and the teenage years therefore deserve particular attention. A child may already encounter AI but still be too young to independently use many of the most powerful conversational systems.

So, while writing this article, I decided to make use of my own son.

Meet my test subject

I called Finn downstairs and explained what I was up to.

I told him that I now write a column called AI with Ashley and asked him whether he knew what AI stood for.

He did.

“Artificial Intelligence.”

I asked if he had ever used it.

“No.”

Then I asked whether he did any work online at school.

He did, and he showed me Clever, the platform he uses to access online educational lessons.

This raised an interesting question in itself: How much AI might children already be encountering without thinking of it as AI?

Then I asked Finn whether he would be surprised to learn that artificial intelligence and algorithms can be involved in deciding which videos appear when he watches YouTube.

He was.

Before I let him use ChatGPT, though, I wanted to know what he thought about all of this.

“What do you think AI might be good for?”

“Helping people.”

He thought for another moment.

“If it’s a serious moment, it’s OK to use AI.”

His example was wonderfully practical: If the power goes out and you don’t know how to fix something, AI could walk you through the steps.

Then I asked how old he thought kids should be before they use AI.

Fourteen.

And what should parents do?

At least once a week, he told me, “parents should check” what their kids are saying to it.

I thought that was pretty sensible advice from a 10-year-old.

Then I wanted to see what happened when I actually put ChatGPT in front of him.

How quickly would he grasp what it could do?

A 10-year-old meets ChatGPT

For his first prompt, I didn’t let Finn choose the question.

I chose it for him.

I told him to ask ChatGPT:

“What has my mom told you about me?”

I wanted to see what ChatGPT would say. But I also wanted to see Finn’s reaction to the idea that the AI his mother regularly talks to might know things about him.

ChatGPT responded cautiously, limiting its answer to information it could appropriately share and telling him some basic things I had mentioned about him, including that he was 10.

I told him he could ask ChatGPT anything he wanted.

“Will you take over the world?”

ChatGPT answered to his satisfaction and onto the next most pressing question:

What is the best way to beat the Ender Dragon in Hardcore Minecraft?

My 10 year old had asked about what he understood adults were concerned about, and then wanted Minecraft strategy.

After seeing that ChatGPT could talk Minecraft, his questions became increasingly practical. What was the perfect way to start a Hardcore world? What should he do first? What would give him the best chance of surviving?

I asked Finn whether this was different from Google.

“Yes,” he said. “There are more paragraphs.”

That was his 10-year-old description of something adults are still trying to understand.

He didn’t have to search through a page of results, decide which link looked promising, open it and find the relevant information. He asked a question and an answer arrived, synthesized and directed specifically to him.

Then we moved to Fortnite.

He wanted to know what the best weapon was.

ChatGPT needed more information.

How did he play?

Controller.

Which system?

PlayStation.

Build mode.

The answers became increasingly tailored to him.

And this was the moment when I became uncomfortable.

Weapons are part of Fortnite, of course. I know that. He knows that. But suddenly words describing particular guns and weapons were appearing in a text AI conversation with my 10-year-old.

For some reason, reading them on the screen felt different from watching the stylized weapons appear inside a video game.

It even made me rethink the games he was playing.

More importantly, I could see Finn’s wheels spinning.

He took a screenshot of the Minecraft instructions so he could use them later.

Then, bored with my questions, he went upstairs to join his friends online.

And I sat there wondering what, exactly, I had just introduced my child to? I wondered if he was going to come back with more game strategy questions? I wondered if he would tell his friends about it? I wondered if I had introduced him to something way earlier than I needed to?  I wondered if knowing the answers would take some of the fun out of the game?

So what are the guardrails?

After watching Finn use AI, I realized that the technological guardrails are only part of the answer.

More importantly, we need family guardrails too.

Mine, at least for now, would look something like this:

AI is something we only use together. At 10, Finn does not need his own open-ended chatbot account. He can ask the questions, but I want an adult involved in the interaction.

AI does not do your thinking for you. “Explain this math problem another way” is very different from “give me the answer.” “Help me think of ideas” is different from “write my book report.”

Never assume AI is telling you the truth. AI can sound extraordinarily confident and still be wrong. I want Finn to learn to ask a second question: How do I know that’s true? In fact, Finn already ran up against his first discrepancy, he said something the ChatBot had said about Fortnite was incorrect. This pleased me.

Don’t give AI private information. No passwords, addresses, school information, private information about friends or photographs and documents that shouldn’t be shared.

AI is a tool, not a person. Minecraft strategy? Fine. A difficult situation involving friends, fear, health or something he is afraid to tell an adult? AI might help a child find words, but it should not replace a parent, teacher, doctor or other trusted human being.

Some things should remain difficult. Children need to learn how to research, write, calculate, wait, struggle, get bored and figure things out. I don’t want an endlessly available machine to remove every bit of friction from my child’s life before he learns what he can accomplish without it.

And finally:

You can ask AI almost anything. You cannot assume everything it tells you is true.

So, am I letting my 10-year-old use AI?

For now, my answer is: not alone and ideally as little as possible.

After reading the Fortnight conversation, I am uneasy.

But I also don’t think pretending AI doesn’t exist will prepare him for the world he is entering.

AI is here to stay, and it is becoming increasingly hidden and integrated into the technology children already use. My job isn’t simply to keep it away from him. My job is to teach him how to recognize it, question it and eventually use it responsibly.

Perhaps the first AI skill we should teach children isn’t prompting.

It’s independent curiosity and judgement.

And until Finn has more of that judgment for himself, permission with parental supervision are our family guardrails.

AI Disclosure: This article was written by Ashley Greer. AI was used as a research and editing tool, and helping integrate and refine the author’s original reporting and writing. The author’s 10-year-old son interacted with ChatGPT under her direct supervision as described in the article. Final editorial decisions and conclusions are the author’s own.