Make-Believe or AI Besties: Explorations of Children’s Interac- tions with AI-Agents Across Cultures

Are AI agents the new but not harmless imaginary friends? Children’s imaginary friends are a natural and beneficial part of early development, helping them navigate emotions, cope with loneliness, build social skills, and explore creativity. As children increasingly interact with different types of AI agents in their everyday life – from Large Language Models, such as ChatGPT for learning and creating content to conversational agents like Alexa and Google, and with AI agents poised to be our personal assistants, therapists (like WoeBot), and even friends (like AI pro- files on social media); childhood social and emotional experiences are forever transformed. Further, there is little parental or school guidance regarding these interactions, raising concerns about the impact on social-emotional development of children and the need for more research on child- AI interactions, especially in family contexts.

In the Besties project, we will explore children’s (ages 10-13) interactions with diverse AI agents exploring how such interactions influence their social and emotional development and perceptions of friendship. As imagination, make-believe, concepts of time and future are culturally and contextually situated, Besties will engage participants in at least four locations out of Finland, India, Japan, USA, South Africa, and Vietnam, which is made possible by a rich network of collaborators, bringing diverse perspectives towards AI agents and their impact, and diversifying design futuring.

Research question: Besties asks – how do children interact with AI agents and what is the social, emotional impact of these interactions on children? And how do adults, teachers and parents, influence these interac- tions? How do the interactions and impacts vary across cultures?

Expected Outcomes: The project explores how children interact with AI agents in their everyday lives, and if and how those interactions differ across cultures, and how adult stakeholders impact those interactions. Expected results include a diverse data set of children’s designs of their future friends (published online as a zine), and possibly a framework for ethical child-AI interactions focusing on children’s social and emotional wellbeing, creativity, and relationships with technology seeking to impact policies on children and AI. The cross-cultural nature of the work promotes diversity and inclusion in scientific work, which is under threat, and promotes scientific renewal in CCI and other fields, such as, teaching, education, psychology, and future studies.

Project PIs: Sumita Sharma (INTERACT), Henna Tiensuu (BISC), Megumi Iwata (LLP), and Yucong Lao (HCCS).

Project Collaborators: Project pilots are run by international partners, including English teacher Junko Tanaka (Japan), Asst. Prof. Jillet Sam, Indian Institute of Technology, Kanpur, (India), Asst. Prof. Noura Howell, GeorgiaTech (USA), Prof. Nobert Jere, University of Fort Hare (South Africa), and M.Ed. Tam Le Hong (Vietnam).

Project Activities (Mar-Dec 2025):

Besties recruited two summer workers for the AI Youth Ambassador program (June 2nd - 13th) who explored the use, design, and development of LLMs together with other ambassadors in week one. In week two, the Besties AI Youth Ambassadors created materials for introducing LLMs to young children and conducted a market analysis of LLMs for children.
Besties pilot - Oulu. Besties conducts its first (90-minute) pilot workshop in Oulu on June 17th with PIs Sharma, Tiensuu, and Iwata, and collaborator Howell with ten participants (ages 9-12 years) at the University of Oulu.