Editor's Note: Dr. Edith Bracho-Sanchez, the author of this article, is a primary care pediatrician at Columbia University Irving Medical Center in New York. She hosts the American Academy of Pediatrics' "Healthy Children" podcast for parents.
(CNN) -- My 4-year-old son insisted that the information provided in the book we were reading before bed about insects was wrong. “Cockroaches lay hundreds of eggs, not millions,” he said, displaying knowledge beyond his years.
I was surprised and asked him if he asked the question to Siri because it was the only technology he knew how to use, and he answered confidently: “Siri is only good for weather, mom. If you want real information, you should ask GPT chat.”
He had verified the information in the insect book using the artificial intelligence application on his father's phone a week ago. It turns out that cockroaches lay hundreds of eggs during their lives, not millions as stated in the book. His father was proud of him, but I was worried.
I wasn't ready to talk to my preschooler about artificial intelligence. I thought this conversation would come later. But when it comes to technology, even young children may know a lot more than we realize. With AI, the question is not whether they should use it, but how, because many of them are already using it.
As a pediatrician, I've spent years discussing screen time with parents, but we don't talk about AI as much. When it comes to screen time, the risks are tangible and familiar: too much screen time can harm the developing brain, and the quality of the content matters, too.
But thinking about the risks that artificial intelligence poses to children requires a new way of understanding the issue.
Artificial intelligence is not satisfied with being a tool in front of the child, but it interacts with him in ways that televisions, small tablets, or even interactive video games cannot provide. It remembers what the child asked the day before, and responds immediately and personalized, in a way that closely resembles human attention.
The danger does not lie in the number of hours children spend using artificial intelligence, but rather in their excessive reliance on it and entrusting aspects of their lives to it. They may be less likely to think when they get the answers to their homework before trying to do it themselves, or to make friends when a chatbot slips into the role of a person they trust, without anyone deciding to do so. As with screen time, the child's age and developmental stage play a key role.
Young children are naturally inclined to trust information given to them by adults, which is how they learn language and grammar and understand the world around them.
A 6-year-old does not yet have the ability to constantly question the source of information or evaluate its reliability, and may not realize that what appears reliable may be false. As for a 14-year-old teenager, he is more able to question information.
This difference extends to relationships as well. Young children easily form emotional bonds with objects that interact with them constantly and in a predictable manner. The chatbot may respond to them every time, and mimic the real dialogue pattern, without the child realizing that the other party is not actually thinking, feeling, or building a relationship with him.
Teens can also form emotional connections with AI, but the motivation may be different. For a teenager facing the challenges of friendship, rejection, identity, and belonging, having something that is non-judgmental, non-confidential, and always available can be very attractive.
As for younger children, they may not even understand what they are talking to. He's not a toy, and he's not a person, but he talks like a person and often sounds like a knowledgeable adult.
In general, teens understand that an AI is not a person, but knowing this intellectually does not always mean treating it emotionally as such, especially late at night, when the teen is alone or feeling anxious or sad, and does not want to seek support from a parent.
It's not about turning children away from AI altogether, because that may not be possible, but about guiding them as their understanding of it develops.
But with young children, a key question arises: Do they even need to use AI? As a pediatrician, I don't think they should use it in early education. The child needs an opportunity to think about the question and try to arrive at the answer before he has it ready.
Artificial intelligence programs interfere with a fundamental aspect of learning during childhood. The child guesses, makes mistakes, tries again, and discusses the reasons until he arrives at the answer. This process is not wasted time, but rather part of learning itself. A lot of learning and creativity happens between “I don't know” and “I've got the answer.”
Here lies one of the challenges. Artificial intelligence is able to shorten this distance quickly. This speed may be useful for adults, but it is not always what a child's developing brain needs. Sometimes, the best response to “I don't know” is simply: “What do you think?”, even if it takes longer to arrive at the answer.
When young children use AI, it is best to be accompanied by an adult, not alone. It can be explained to them simply: “This is a computer program that generates answers. It is not a person, and sometimes it makes mistakes.”
It is also better to keep its use limited, such as using it together to answer a specific question, rather than it becoming the child's first solution whenever he does not know the answer immediately.
As children grow older, it becomes important to teach them to question information, verify it from other sources, and know exactly what to keep from their data.
In adolescence, the most important question becomes the role that artificial intelligence plays in their lives: Does it add to their social and emotional world, or does it gradually replace their real relationships, such as talking with a friend, a parent, or a therapist?
Teens should also know that some AI programs are designed to keep them engaged, and that an always-available chatbot may not have the goals of someone who actually cares about them.
Therefore, it is not just about setting limits on the use of artificial intelligence, but about teaching children human skills that technology cannot replace, the most important of which is turning to others when they face difficult feelings. This begins by opening the door to dialogue with them, and by setting an example for them in the way we use artificial intelligence, explaining to them what it can benefit them in, and what it cannot replace.
The problem was not that my son used an artificial intelligence program to verify information about insects. The question is whether over time he will learn to distinguish between where he should turn to me and his father instead. We have to teach him that AI, no matter how confident and human-like it may seem, will not care for it, love it, or be willing to help it the way we do.
