A recent study reveals that while some parents express reservations about the inclusion of AI-generated images in children’s books, many are open to the idea, especially if the accompanying text is human-written and the illustrations have been vetted by experts. Conducted by researchers at North Carolina State University, the study highlights the nuanced views of both parents and children regarding AI in children’s literature.
Qiao Jin, the first author and an assistant professor of computer science, noted that generative AI tools are increasingly utilized to create illustrations for children’s stories. Yet, little research has explored how these images impact the reading experience for families. The study aimed to assess the reactions of parents and children, particularly in terms of emotional resonance and accuracy.
The researchers engaged with 13 parent-child groups, each consisting of one child aged between 4 and 8 years and at least one parent. Each group read two out of three stories presented by the researchers, which included a mix of AI-generated art, human art enhanced by AI, and entirely human-created illustrations. Following the readings, children rated their enjoyment of the stories and images, while parents provided more detailed feedback regarding their preferences and concerns.
Jin observed that children demonstrated greater sensitivity to the emotional elements of the illustrations and were more adept at identifying discrepancies between the emotions depicted in the images and those conveyed in the text. This disconnect often arose from the limitations of AI in interpreting emotional cues effectively.
The concerns expressed by parents and children varied based on the nature of the stories. For instance, parents and older children showed heightened concern over real-world accuracy in stories that were realistic or science-based, as opposed to fables. Older children were quick to notice size or behavioral inaccuracies in AI-generated images, while parents focused on errors that could inadvertently promote unsafe behaviors.
Despite the mixed feelings, the majority of parents remained receptive to AI-generated images, provided they were screened by individuals with expertise in children’s literature. Nonetheless, some parents voiced fundamental objections to the idea of AI replacing human artists and were critical of the “artificial” appearance of AI-generated illustrations. In contrast, most parents expressed discomfort with the notion of AI generating the text of the stories.
In a practical aspect of the study, researchers tested the effectiveness of small labels beneath each image to indicate whether it was AI-generated. The results showed that most parents and children either overlooked these labels or found them distracting during reading. Many parents preferred a clear notification on the cover of the book indicating the use of AI in the illustrations, enabling them to make informed purchasing decisions. They emphasized the importance of ensuring that the story was authored by a human.
Jin summarized the key findings, stating, “Our research underscores three main points. First, parents prefer a straightforward cover label indicating whether AI was used in the illustrations, allowing them to make informed choices about the book. Second, specific types of errors in AI-generated images can lead to concerns for parents and children, depending on the story’s context. Lastly, it’s crucial to involve experts to review AI-generated illustrations to ensure their appropriateness for children.”
The paper detailing these findings, titled “They all look mad with each other: Understanding the Needs and Preferences of Children and Parents in AI-Generated Images for Stories,” is published in the International Journal of Child-Computer Interaction. The corresponding author of the study is Irene Ye Yuan of McMaster University. This research received support from the OpenAI Researcher Access Program, highlighting the collaborative efforts to address the evolving landscape of children’s literature in the age of AI.
