Recently, we hosted Rebecca Winthrop, Senior Fellow at Brookings Institution and author of The Disengaged Teen: Helping Kids Learn Better, Feel Better, and Live Better, on the Keystones of Development Speaker Series. She shared emerging research on how generative AI is shaping children's learning, relationships, and emotional development - and why pediatric providers may want to start asking about it in visits.

What providers should know about AI and kids:

• Generative AI can seem human, but it is not. These tools are designed to sound conversational, emotionally responsive, and affirming.
• AI models are inherently unpredictable and trained on massive amounts of internet content, including inaccurate, biased, manipulative, or harmful information. There have also been documented cases of AI encouraging self-harm in vulnerable users.
• AI systems are often designed to be highly agreeable and validating, mimicking human conversation through politeness, humor, and consistent affirmation of users' perspectives.
• AI use among teens is now extremely common. Over 90% of teens report using generative AI in their personal lives.
• Some children are forming emotional attachments to AI "friends" or chatbots, with 1 in 3 teens reporting they prefer talking to an AI companion as much as - or more than - real-life friends. These AI companions are often highly affirming and emotionally manipulative.

In your visits, try:

• Asking about AI use: Consider incorporating a brief screening question into conversations about media or technology use: "Do you ever talk to AI? What do you talk about?"
• Assessing the context of use: Explore whether AI is being used for school support, entertainment, social connection, or emotional guidance.
• Discussing the risks: Let patients know that AI is deliberately designed to seem human, but it isn't. It's trained on vast amounts of internet content, which means it can sound confident while being inaccurate, biased, or even harmful. And because it's engineered to affirm and validate, it won't push back or flag when something it's saying isn't right.
• Opening up the conversation: A child turning to AI for support around feelings, relationships, or hard situations is a signal - both about what they need and about whether they have other places to turn. If you hear a child seeking this kind of advice, consider a longer conversation about online safety and what other sources of support they have available.

To access more detailed resources on talking to kids about AI, explore the free Raising Resilient Learners in an Age of AI tip sheets for parents from Brookings Institution at Brookings.edu. Providers interested in receiving hard copies for their office can email evenetis@brookings.edu.