Children's susceptibility to content generated by artificial intelligence

Advancements in artificial intelligence (AI) tools can make it difficult to discern what is “real”. Though children are interacting with AI-generated materials through entertainment and learning applications, no published work directly considers children's ability to distinguish human-created from AI-generated information. In the current study, 37 children (6-10 years old) heard a narrative about a “SmartBot” (representing AI) that could write text, invent photos, and create art. They were then asked to discern which in a series of stimuli including short texts and images of objects, art, and faces, had come from the SmartBot vs. a human teacher and answer questions regarding their beliefs about AI. Children also completed measures of abstract reasoning and receptive vocabulary, and parents completed a home technology use survey. A separate sample of 49 adults (18–34 years) served as a comparison group for accuracy scores on the set of stimuli. On average, children performed at or below chance in discriminating human-created from AI-generated content across modalities, and their overall accuracy was significantly below that of adults. Children's individual discernment varied and was significantly and negatively associated with parent reports of children's at-home technology use. Children had varied beliefs about AI agency and experience, but broadly understood that an AI agent (the SmartBot) is not alive (95%). Our findings suggest that spending more time on digital devices is associated with susceptibility to artificial content and risk for misattributing AI content as human, highlighting the importance of early AI literacy to support children's engagement with AI technologies.