There seem to be two vocal schools of thought right now: AI is either “the best thing in the world” or “we have to avoid AI at all costs.” But you say that AI is good for some things and bad for others, and that it should not be used by kids without oversight from their caregivers. How did you arrive at that conclusion?
Through the developmental science. I’m a researcher, I’m a pediatric physician, and my work has always been around how do we ensure healthy development of our children, and specifically how human connection, and language and interaction, supports healthy development.
So I was really thinking through how we navigate this unprecedented moment, when we have technology that can mimic the interaction that wires a child’s brain but also can support humans in being more present.
You talk a lot about the biology of human connection and what that does to our brains: We learn to become people by being around other people. What is your biggest concern about technology, specifically AI, and its effect on those human relationships?
Relationships and human connection are the core of not just learning but learning to be human and to connect with others. And suddenly we have these technologies that humans, because we are such social beings, want to attach to deeply. The younger you are, the more likely you are to attach and connect.
We learn how to have relationships with these friction-filled, good-enough parents and good-enough aunts and good-enough teachers. That’s how we learn that other people have thoughts and feelings, and things cannot go perfectly but can still be okay.
When you have these technologies that can mimic empathy but do it in a frictionless, sycophantic way, we risk teaching children that relationships can be that way. And we risk raising children who may reach every developmental milestone but in the end not be able to connect to other human beings.
Is there something about AI and its effect on children that feels different to you than technologies like smartphones and social media?
All the technologies we’ve built — and technology has always rewired our brains in some ways — has been passive. We’ve engaged with it, and Big Tech has built it in a way so we really engage with it, but it’s always been a one-way street. This new technology doesn’t just position itself as an engagement mechanism but an intimacy mechanism, a way that you build relationships.
Your advice to parents raising children in the AI era feels evergreen: Spend quality time with your child and trust your instincts. Why do you think parents are struggling with that? Why does it feel like there must be some different approach that’s specific to AI?
All those skills that we’ve been trying to optimize our kids for — knowledge accumulation, being best at math and science, and getting into the best preschools and colleges — now all of a sudden, the rules of the game have changed, because AI can do all of those skills with superhuman abilities.
Children still need to learn to read, write, and do all of these things, but that’s not going to be the big differentiator. The big differentiators are those distinctly human skills, right? Curiosity, creativity, critical thinking, resilience, agility — AI erodes those skills.
Some parents fear that not exposing their kids to technology in general, and AI specifically, will hold them back. What is your advice for how to approach that?
AI literacy — understanding what AI is, how to use it, how to think about it — is very different from using it.
To those parents, I would say that you can still learn about AI, or even use it sporadically, without relying on it. Learning the skill of how to use an AI chatbot is easy. But developing human skills takes work.
It’s like going to a gym. If you went to a gym and had somebody else lifting those weights, you’re not going to develop those muscles. And those muscles for human skill formation are only made by not using AI, but that doesn’t mean that you don’t understand it. AI literacy is super important, and understanding how to use it in safe ways is super important.
You developed a framework called DETECT, which is an acronym for things to consider when evaluating AI products or services designed for children. How do parents use it?
Every few weeks there’s a new AI toy, tutor, companion promising to make parenting easier or make kids smarter. And my advice to parents is not to evaluate each one from scratch in a panic every time something new hits the shelf.
What is this thing actually designed to do? Was it ethically trained on data that reflects real diversity? Are there signs of trouble or accounts of harm to other kids? Is there evidence that it delivers on what it promises? Is your kid’s data being handled with confidentiality? And what is it actually teaching your child, not just academically but about relationships and values?
You don’t have to be a technologist to run through these questions. You need about 10 minutes and maybe a healthy amount of skepticism towards marketing claims. I’ll add one more thing: Trust your gut. I call it parental GPS. It is actually a biological phenomenon. If something about a product feels off, that feeling is data, too.
Applying that framework, you reach the conclusion that AI chatbots are not suitable for kids to use. Why?
We don’t understand what it means to be connected and attached to these AI companions. Relationships aren’t formed by an AI chatbot saying, “You’re right, don’t leave, let’s talk about your thoughts and feelings.”
Science is pretty clear: Relationships and friendships aren’t actually built on alignment. It’s on missteps, ruptures, repairs. Those are the foundations for building really strong relationships. It’s sort of like eating a fruit snack every day. A fruit snack was made in the image of real fruit, but if [young children] eat fruit snacks every day, they will come to prefer that super-sweet, hyper-palatable, easy-to-go-down [substitute]. All of a sudden, when you start eating that real apple, it doesn’t always taste perfect, and it feels like a lot of work.
You focus on raising young children in the age of AI, but how should parents navigate the use of AI by older children?
How you approach it looks very different. It’s much easier in the early ages to say, “You know what, Grandma, that [AI-powered] stuffed animal that looks really cute, that will talk forever to your grandbaby? That’s not a good thing.” You have more control.
[With older children,] have a strong relationship, where they feel like they can come to you and talk to you. Help them understand how these technologies work. There’s nothing better to get children to question things than when you share how [companies are] preying on their own vulnerabilities. I have young adults who say, “This is not helping. I feel like my writing is getting worse, Mom.” They’re feeling it.
But it’s an unfair fight. The fact that we are having to navigate this without guardrails is sort of like telling parents, “Here, we’ve got all these car seats. We don’t know if they’re safe — go out and let us know.”
We’ve heard from parents who have made decisions about the use of technology, specifically AI, in their homes with their children, but they don’t have any control over what happens in school. How can parents get involved in decisions regarding the use of technology in the classroom?
They have a lot more agency than they realize. Without being “that parent,” go in and say to your teacher or administrator: “Does our program have an AI policy for the year? Which AI tools are approved and which are prohibited? How are you getting supported?” Because many of these teachers deserve support and learning, just like we do.
I was struck by an anecdote toward the end of the book in which an acquaintance tells you that she wondered if her children would someday need to live quarantined from those raised with unfettered access to AI. What do you make of that potential future?
My biggest concern is how it played out with ultra-processed food.
Food companies found out they could create this new food in the lab at a much cheaper rate, and it led to food deserts in under-resourced communities, where they had access to this as a “better than nothing” framing, leading to health consequences for individuals who weren’t fortunate enough to receive organic or farm-fresh food.
In that same way, AI can be built to replicate human interaction or substitute for humans. You’re already hearing about people bringing humanoid robots into under-resourced schools, with the framing of, “Well, this is better than nothing in these overcrowded schools” or for families working three jobs. Why not build technology to allow parents or teachers to be more present instead? This is a revolutionary technology. Let’s use it to actually lead to human flourishing, so all children have an opportunity.
This interview has been condensed and edited for clarity.
This article was edited by Jason Chen.
