Generative AI in Children's Creative Collaboration: Impact, Perception, and Design Guidelines

Paper Title

Generative AI in Children's Creative Collaboration: Impact, Perception, and Design Guidelines

Publication Info

• Topic area: Human-computer interaction and generative AI in children's collaborative creativity.
• Keywords: Generative AI, children's collaboration, creative co-design, Joint Media Engagement, AI-mediated interaction, child-centered design, peer collaboration, AI literacy, multimodal AI tools, social dynamics.

Background and Problem

• Problem / challenge: Limited understanding of how generative AI (GenAI) mediates children's creative collaboration, particularly its impact on social dynamics, children's agency, and collaborative processes.
• Significance: Collaboration is essential for children's cognitive and social development, yet GenAI's role in fostering or hindering these interactions remains underexplored. Addressing this gap is critical for designing AI systems that support healthy peer interactions.
• Motivation and related work: Previous studies have focused on GenAI's ability to enhance creativity and learning outcomes but have not examined its influence on the collaborative process itself. Existing research on Joint Media Engagement (JME) has primarily focused on passive media, leaving a gap in understanding how responsive technologies like GenAI shape social and relational dynamics.

Solution

• Proposed approach: Investigating GenAI's role in children's creative collaboration through participatory design (PD) sessions and focus group interviews, using multimodal GenAI tools (ChatGPT, Stable Diffusion, MidJourney, InVideo AI).
• Novelty:
  1. Empirical evidence on how children interact with diverse GenAI tools and perceive its impact on collaboration.
  2. Insights into GenAI's dual role as a social bonding agent and a social barrier in peer collaboration.
  3. Design implications for creating socially attuned GenAI systems that respect children's agency and enhance collaboration.
• Procedure and key techniques:
  • Conducted five co-design sessions with 28 children (ages 5–11) and focus group interviews with 17 parents.
  • Observed children's interactions with GenAI tools across text, image, voice, and video modalities.
  • Analyzed data using reflexive thematic analysis to identify patterns in social dynamics, agency, and perceptions of GenAI.

Results

• Concrete findings:
  • GenAI fostered positive social dynamics by enabling "Human vs. AI" teaming and shared ownership in co-creation.
  • GenAI disrupted collaboration when roles among children were unclear, AI ignored group dialogue, or AI dominated children's agency.
  • Children and parents envisioned GenAI as a socially attuned "older sibling" that scaffolds creativity while allowing playful disagreement.
• Advantage over baselines:
  • GenAI facilitated shifts from "my creation" to "our creation," particularly among children who were not close friends.
  • It introduced new forms of convergence and divergence in collaboration, enhancing shared engagement and critique.
• Experiments / evaluation:
  • Design sessions included activities such as story creation, science poster development, and artifact co-design using GenAI tools.
  • Focus group interviews captured parental perspectives on GenAI's role in collaboration.
  • Data collection included video recordings, design artifacts, and analytical memos.
• Limitations and future work:
  • Study conducted in a controlled university setting, which may not fully reflect everyday environments.
  • Limited number of GenAI tools used; future work could explore systems explicitly designed for multi-user collaboration.
  • Longitudinal studies needed to understand how children's perceptions and AI literacy evolve over time.

Summary

This study explored how generative AI influences children's creative collaboration, revealing both supportive and disruptive effects on social dynamics and agency. GenAI fostered shared ownership and playful engagement but sometimes undermined peer interaction by dominating roles or ignoring group dialogue. Children and parents envisioned a socially attuned AI that acts as a responsive partner rather than a directive authority. The findings contribute empirical evidence and design guidelines for creating GenAI systems that enhance collaboration while respecting children's agency, offering insights applicable to broader contexts of human–AI partnerships.