Scientists Are Studying What AI Does To Young Brains. The Biggest Questions Are Still Unanswered.

Artificial intelligence has become part of everyday life for millions of children and teenagers before researchers have had enough time to determine what frequent chatbot use may mean for developing brains.

A growing body of research is now trying to answer that question, with scientists examining everything from attention and creativity to emotional development, critical thinking and the effects of outsourcing difficult mental tasks to AI.

A new peer-reviewed review published in AI & Society found that research into adolescents and AI remains fragmented and is developing far more slowly than young people are adopting the technology. The authors said scientists still know "very little" about how AI chatbots may affect adolescent development and called for a coordinated research agenda covering different types of AI use.

The study, led by researchers at the University of North Carolina at Chapel Hill and Hopelab, gathered input from developmental scientists and other professionals to identify the AI uses they believe should be studied most urgently. It also sought expert views on research methods and possible protections for young users while longer-term evidence is being collected.

That gap matters because chatbot use is already widespread among teenagers.

About 64% of U.S. teens ages 13 to 17 said they had used an AI chatbot, according to Pew Research Center, while 54% had used one for help with schoolwork. One in 10 said chatbots helped them with all or most of their schoolwork, while 12% reported using them for emotional support or advice.

One of the earliest attempts to look directly at children's brain activity while they use generative AI comes from researchers at Technion-Israel Institute of Technology.

The small study compared 15 children ages 6 and 7 with 16 adults while both groups held a five-minute creative conversation with ChatGPT inside an fMRI scanner. Researchers measured connectivity across brain networks involved in cognitive control, attention and memory.

Children showed lower connectivity than adults within several control and attention networks and less integration between some of those networks while using ChatGPT, according to the study. Adults also showed a relationship between stronger cognitive-control connectivity and creativity that was not observed among the children.

But the findings suggest caution is needed. The research involved only 31 participants and remains a preprint rather than a peer-reviewed study. It also compared children with adults during one short activity, meaning it cannot establish that using ChatGPT causes developmental harm or produces lasting brain changes.

The researchers themselves said the findings point to differences in how children and adults implement creative thinking during AI-assisted tasks rather than proving that children lose cognitive ability.

That distinction is important as concern grows around the idea of "cognitive offloading" — using tools to perform mental work that a person might otherwise complete themselves.

A July paper in Frontiers in Developmental Psychology argued that adolescence could be particularly important because executive functions such as planning, evaluating information and constructing arguments are still developing. The authors proposed what they called a "cognitive offloading paradox": AI is especially useful at carrying out some of the same effortful tasks young people need to practice while building those skills.

The authors were explicit that this remains a hypothesis rather than an established effect. They said there is currently no longitudinal or experimental evidence showing that a particular pattern of AI use causes long-term impairment in adolescent executive-function development.

Researchers are also increasingly distinguishing between AI that supports a child's thinking and AI that replaces it.

A separate developmental-neuroscience framework published this summer in Policy Insights from the Behavioral and Brain Sciences examined how interactive AI chatbots fit into a period of adolescence marked by heightened social sensitivity, developing regulatory systems and substantial neurobiological change.

The researchers argued that digital technologies can interact differently with young people depending on how they are designed and used. An AI system that helps a student work through a problem may create a different developmental experience from one that simply supplies the finished answer.

Existing educational research offers some evidence for that distinction.

In one experiment involving nearly 1,000 high school math students in Turkey, students using a general-purpose AI assistant performed better while they had access to the tool but later performed worse than students who had studied without it when the AI was removed. Students using a modified AI tutor designed to provide guidance rather than answers did not show the same decline in later performance.

Developmental researchers have therefore cautioned against treating "AI use" as a single behavior. Asking a chatbot to explain a difficult concept, generating an entire school assignment and talking to an AI companion about personal problems involve very different cognitive and social processes.

The questions extend beyond learning. A study published in JAMA Pediatrics found that nearly one in five U.S. adolescents and young adults ages 12 to 21 reported using AI chatbots for mental-health advice in 2025. Most users had not told another person that they were using a chatbot for that purpose.

Researchers studying adolescent development have raised questions about whether frequent social interaction with highly agreeable AI systems could affect skills that normally develop through human relationships, including handling disagreement, interpreting social signals and responding to the needs of another person.

An August paper in Perspectives on Psychological Science described adolescence as a sensitive period for developing social belonging, emotional regulation, autonomy and identity. The authors said generative AI could offer benefits, including opportunities to practice interactions or explore ideas, while also potentially displacing experiences that help those abilities develop.

Again, the authors stressed that many of those possible benefits and harms have not yet been established through long-term studies.

The speed of adoption is one reason researchers say those questions have become urgent. A separate Common Sense Media survey of 1,204 children ages 9 to 17 found that 86% used or interacted with AI and nearly a quarter did so every day. More than four in 10 said a parent or guardian had never spoken with them about AI safety.