Understanding how children conceive of AI is essential as AI-enabled technologies become part of their everyday lives. Yet most research relies on verbal or survey-based methods that capture only a limited view of children’s perspectives. We address this gap through a multi-modal, participatory study with 50 primary-school children aged 6–11. We combine group discussions with picture and sticker tasks, craft-based prototyping, and embodied theater activities. Across all modes, children described AI as conversational and often humanoid, but in the design and performing mode they created non-anthropomorphic, relational devices that revealed power dynamics and boundaries rarely visible in recall modes alone. The findings show that children’s understandings are mode-dependent and multi-layered, suggesting that researchers and designers can access children’s lived experiences and imaginative futures with AI by using a mix of multi-modal methods.
No takes yet. Share an insight, caveat, or question.
Honauer et al. (2026) studied this question.