Anthropomorphism in Children's Interactions with LLM Chatbots (arxiv.org)

A recent systematic review has explored how children interact with large language model (LLM) chatbots, focusing on the phenomenon of anthropomorphism—the tendency to attribute human traits to non-human entities. By analyzing 35 empirical studies published between 2022 and 2025, researchers identified key drivers of this behavior, such as the construction of human-like personas, adaptive scaffolding, and non-human embodied design. These factors contribute to children's attachment and social responses to chatbots, leading to complex outcomes like the formation of social ties and moral reasoning towards these AI entities. Significantly, the findings highlight both the benefits and risks associated with these interactions, revealing that children can experience paradoxical social responses and develop dual consciousness about the chatbots. This research plays a crucial role in informing the future design of LLM chatbots targeted at young users, suggesting that developers should consider children's developmental needs to promote safe and beneficial interactions. By understanding how children relate to chatbots, the AI/ML community can enhance the usability and ethical implications of AI technologies in educational and social contexts.