Overview
PlayFutures is a participatory design project aimed at fostering critical AI literacy among children. In an era where AI reshapes communities without the input of younger generations, this project sought to give children a voice. By combining Generative AI (ChatGPT) with tangible puppet-making and performance, we created a 4-hour workshop model where children (aged 9-12) reimagined local civic spaces. The result was a validated method for "Design Futuring" that helps children transition from passive technology consumers to active, critical creators.
Problem
Current AI literacy efforts often focus on technical skills, neglecting "child-specific" critical thinking.
Children encounter AI daily but lack the agency or voice to understand or influence how it impacts their own communities. We ask:
How do children critically evaluate and interact with AI platforms (like LLMs or Image Generators) when using them as design tools for the future building?
In what ways can a combination of generative AI, hand on-making, and theatrical performance scaffold children's process of working with gen-AI?
Goal
To develop a workshop framework that uses Design Futuring to scaffold critical thinking about AI, while engaging children in civic discussions.
A flow diagram of how the workshop unfolded
Left to Right: Snippets from the final puppet performance of Piku Niku Council at the Park, The Cat Lovers council at the library and the Oogy Woogy council at the museum.
Observations
Ideation of Civic Futures
Children proposed optimistic, highly specific changes to their environment:
Group 1 (Library): More natural light, comfortable outdoor seating.
Group 2 (Museum): "More life," specifically requesting a "rainbow wooly mammoth."
Group 3 (Park): A cleaner, larger lake with teeming wildlife (environmental focus).
Prototyping acted as friction
The users treated AI not as an oracle, but as a "faulty process tool."
Group 3 attempted to fuse images to create a specific creature. The AI failed to produce the "rainbow wooly mammoth" on the second attempt.
Participants described the tool as "surprising, confusing, and mean," highlighting the gap between their imagination and the AI's capability.
Iterations to Reclaim
The most significant UX finding occurred during the script generation phase.
Children rejected the AI-generated scripts for lacking "soul."
One team improvised their entire performance because the generated text felt "really AI-y." Another team cut two full lines of dialogue because "It wasn't their personality. It sounded unnatural."
Insights
Through reflexive thematic analysis of 4 hours of video/audio and 7 activity booklets, three core themes emerged:
1. Recognising AI as an "Imperfect Tool" for Design
Children quickly demystified the technology. Instead of viewing AI outputs as final solutions, they treated them as rough drafts. The "glitches" in the AI (e.g., failing to generate specific animal features) became teaching moments about the limitations of technology.
2. Preference for Authentic Visions
There was a strong user preference for children's authentic voice over automated efficiency.
Only one team agreed with the AI-generated ending. The other two teams completely rewrote the conclusions to better fit their narrative goals.
Children distinguished between "AI-like" content and their own preference, and they actively curate the output based on their vision.
3. Tangible for Expressing Visions and Critique
The physical act of making puppets provided a necessary counterbalance to the screen-based AI work. It allowed the children to step back from the interface and critique the digital output against their physical creation.
Reflections: AI Literacy through Civic Visions
PlayFutures demonstrated that engaging children in AI literacy requires more than just coding tutorials. By integrating Making, and Performance and Futuring we successfully created an environment where children could critique AI all through putting children as city planners or city council members.
The study proves that when users (even young ones) are given the license to edit and override AI (through authority like civic planners) they prioritize their own agency and creative vision, treating AI as a supportive—but fallible—tool rather than an authority.
Funding
Publication