Children readily engaged with an AI-powered toy robot and often accepted its answers as true in a small study that raises questions about how to teach AI literacy from an early age.
Researchers at the University of Cincinnati invited 41 children aged five to ten to play three interactive games with Miko, a commercially available robot that uses generative AI. They then spoke to the children about their experience.
The researchers found that the children were enthusiastic about the robot, including those who had not seen one before. They also tended to believe what it told them. The findings were presented at the Comm Horizons conference at the University of California, Davis.
“The biggest takeaway for me is these children are extremely excited about robots and AI,” said Kelly Merrill Jr, an assistant professor at the University of Cincinnati. “AI can be a scary thing for parents.”
The study offers a close look at how a small group of children responded to one toy. It does not establish how children generally understand AI or whether using an AI toy changes their behaviour over time. For the researchers, the children’s apparent trust highlights the need to help them question what these tools say.
Nancy Jennings, a professor at the university who worked on the study, said children need opportunities to learn both what AI can do and where it falls short. She encouraged parents to talk through children’s experiences with the technology and help them think critically about its responses.
The researchers also asked whether children saw Miko as a toy, a friend or something else. Jennings said some children who reported difficulty making friends at school particularly valued talking to it. They described the robot as a friend that would not judge or bully them.
That response may help explain the appeal of conversational toys, but it also points to an important distinction for adults supporting children’s use of them. Merrill said the tools should complement human relationships rather than replace them.
During one game, the researchers observed the robot redirect children when a story became too frightening or involved guns. The observation shows how that toy responded in those interactions; it does not settle broader questions about the safeguards used by AI toys.
Those questions include the suitability of the content toys generate and how they handle children’s information. The researchers noted criticism of some AI toys over age-appropriate safeguards, as well as privacy concerns raised by consumer advocates.
For many of the children in the study, speaking to a device was already familiar. Merrill said about half regularly used AI agents at home, often to request music through voice assistants such as Siri or Alexa.
The findings give educators, designers and families a reason to examine how children decide whether to trust a machine’s answer. As conversational tools become part of children’s play, the researchers argue that AI literacy should begin early enough to help them recognise that a confident-sounding response can still need checking.