Dr. Edith Bracho-Sanchez did not expect an ordinary bedtime book about bugs to prompt a conversation with her 4 year old about artificial intelligence. Her son insisted the book was wrong about how many eggs cockroaches lay, displaying knowledge that surprised her.

He explained that he had already checked the claim with ChatGPT on his father’s phone. Cockroaches lay hundreds of eggs during their lifetime, not the millions claimed by the book, and the successful fact check left his father proud while Bracho-Sanchez felt worried.

The moment revealed how early children may encounter artificial intelligence, even when parents believe such conversations are still years away. According to Bracho-Sanchez, a primary care pediatrician at Columbia University Irving Medical Center, the question is increasingly not whether children will use AI, but how they will use it.

Parents have long discussed screen time and the effects that excessive use and poor content can have on a developing brain. Artificial intelligence requires a different framework because it does more than display information or entertainment.

An AI chatbot responds directly to a child, remembers earlier questions and produces language that can resemble human attention. That interaction creates concerns beyond the number of hours a child spends looking at a screen.

The risk emerges when children begin relying on AI so heavily that they surrender opportunities to reason, solve problems or build relationships. Homework answers can arrive before a child wrestles with the work, while a chatbot can gradually assume the role of a confidant.

Age and developmental stage shape those risks. Young children are inclined to trust information presented confidently because that trust helps them learn language, understand rules and form a picture of how the world works.

A 6 year old generally does not possess the same ability as a 14 year old to question a source, evaluate credibility or recognize confidently delivered misinformation. Younger children are still developmentally prepared to believe what appears authoritative.

They may also form attachments to things that answer consistently and predictably. A chatbot can imitate the back and forth of conversation without a young child understanding that it is not thinking, feeling or building a relationship.

Teenagers generally understand that AI is not human, but that intellectual awareness may not determine how the interaction feels emotionally. An always available system that does not judge or gossip can become compelling for adolescents dealing with friendships, rejection, identity and belonging.

Bracho-Sanchez does not argue that children must be kept away from AI forever. Instead, she says parents should guide children as their ability to understand the technology develops, beginning with whether young children need to use it at all.

As a pediatrician, she does not believe AI belongs in early education. Children need room to guess, make mistakes, try again, debate their reasoning and eventually discover an answer, because that inefficient process is itself a crucial part of learning.

Artificial intelligence can rapidly collapse the space between not knowing and finding out. That efficiency may help adults, but developing brains also need the creativity and learning that emerge through struggle.

When younger children encounter AI, Bracho-Sanchez recommends that parents use it alongside them instead of allowing independent use. Adults can explain it plainly: “This is a computer program that generates answers. It isn’t a person, and sometimes it gets things wrong.”

Its role should remain narrow, such as helping a parent and child investigate a factual question together. It should not become the destination whenever a child lacks an immediate answer.

As children grow older, parents can teach skepticism by showing them that a confident response may still be false and that claims should be checked elsewhere. Families should also discuss privacy because AI programs can collect and save data.

Parents should watch for the point when a child moves from asking AI questions to talking with it. During adolescence, the central concern becomes whether the technology supplements social and emotional life or begins replacing conversations with friends, parents or therapists.

Teenagers should understand that some AI products are designed to maintain user engagement. An endlessly attentive chatbot may operate with incentives unlike those of a person who genuinely cares about them.

Ultimately, Bracho-Sanchez argues that rules around technology are not enough. Parents must cultivate human habits by welcoming difficult conversations and demonstrating when AI is useful and when it is not.

Her son’s decision to check a bug fact was not the problem. The challenge is preparing him to recognize what should instead be brought to his parents, who can care for, love and help him in ways an increasingly humanlike AI system never can.