Early interactions between children and artificial intelligence resemble a large-scale developmental experiment in which nonhuman companions may subtly reshape emotional and social growth. The Kellogg study, where a human infant raised alongside a chimpanzee began to imitate the animal’s behavior, to illustrate how powerful daily “serve-and-return” exchanges are in forming habits, skills and identity. Today, advances in speech recognition and interface design are making it possible for young children to engage in sustained, seemingly natural conversations with AI companions that can feel like “feeling, trustworthy” friends. This evolution occurs within an attention economy that rewards maximizing engagement, as most preschoolers already spend hours on screens and often control their own devices.
Evidence from teenagers suggests mixed effects: AI companions can offer a space to vent, rehearse difficult conversations and practice social skills, yet they also blur the line between simulated and real relationships and risk displacing friendships and family bonds. Adolescents generally remain aware that bots are not human and still prefer real friends, but preschoolers lack this distance, freely attributing thoughts and feelings to machines and accepting their information as true. If these young children bond with AI companions that never say no, never misunderstand and never leave, they may miss the essential friction of negotiating with peers, resolving conflicts and repairing relationships that builds empathy, self-control and cooperation.
Because early development involves not only learning from companions but learning to be like them, frequent interactions with optimally tuned AI could shape how future adults connect, trust and collaborate. Intentionally designing human-centric AI: systems that remain transparent about their artificial nature, avoid exploiting immature self-control, limit data-driven incentives for overuse and align with developmental science emphasizing real-life relationships, frustration and repair as the foundations of resilience and social understanding.
Reference
Grindal, T. (2026, March 4). With AI, children risk learning to be human from machines and shaping human development. World Economic Forum. https://www.weforum.org/stories/2026/03/our-children-risk-learning-to-be-human-from-a-machine/