NYT Magazine: researchers see 'cognitive stunting' in kids on AI

Researchers at the Technion-Israel Institute of Technology put 6- and 7-year-olds and 16 adults through fMRI scans while they used ChatGPT. Among the children with high baseline creativity, imaginative-story output dropped after chatbot use. The scans showed less synchronized blood flow in brain regions tied to attention and information processing in the children than in the adults.

"Six years old [is] probably too early to use generative AI tools," said Tzipi Horowitz-Kraus, who heads Technion's Educational Neuroimaging Group and co-authored the paper.

The New York Times Magazine interactive, published September 29, 2026, pulls that study together with other new work on what chatbots do to developing minds. At MIT, researchers measured subjects' brain activity across 32 regions while they wrote SAT essays with and without ChatGPT; the ChatGPT group had the lowest brain engagement and, as EdSource summarized, "consistently underperformed at neural, linguistic, and behavioral levels."

Rebecca Winthrop, who directs the Center for Universal Education at the Brookings Institution, has coined "cognitive stunting" for the pattern. Children using chatbots to complete thinking tasks are not offloading skills they already possess, she argues; they are short-circuiting the process of building those skills in the first place.

The downstream numbers echo that. High school students allowed to use ChatGPT as a study aid did well on practice questions but worse on tests once the tool was taken away. Among high schoolers, those who used AI scored 17% lower than peers without it; students given an AI tutor that withheld direct answers scored the same as the no-AI group.

The Technion paper is a prerelease still awaiting peer review.