Almost one in three children using AI trust it completely, researcher says

Nearly one in three children who use artificial intelligence tools trust the information they receive “very much,” according to a new survey of 9- to 14-year-olds in Denmark, DR reports.

The study, conducted by Norstat for DR’s ULTRA:BYTES project, found that 31% of children who use AI—primarily tools like ChatGPT, SkoleGPT, and MyAI—place high trust in its responses. Many rely on AI for homework help, explanations, and fact-checking, with some using it daily.

“Fundamentally, chatbots make statistical predictions, so you can never fully trust them,” said Karl-Emil Kjær, a researcher at Aarhus University and advisor to the project. He noted that AI systems are designed to sound confident and human-like, which can make their answers seem more credible than they are.

In interviews with fifth-grade students at Dyveke School in Copenhagen, several children admitted they rarely question AI responses. “I just believe what it tells me,” said 12-year-old Angelina Mannio. Another student, 11-year-old Alma Højgaard Olsen, added, “When it sounds true, it just sounds right.”

Kjær emphasized that parents play a key role in helping children develop critical thinking. “Parents can discuss how AI works—comparing it to a parrot that repeats words without understanding them—and explore its limitations together,” he suggested. While he doesn’t advocate banning AI use, he urged caution: “We need to be careful how we rely on it.”

Some students, like 11-year-old Albert Augo Birkelund Jørgensen, said they now plan to verify AI answers more thoroughly. Others, like Højgaard Olsen, intend to reduce their usage.

ULTRA:BYTES is a DR initiative aimed at helping children, teachers, and parents critically engage with technology. Schools can participate via dr.dk/ultrabytes.