Child-Centered AI · UX Case Study

Designing Playful AI for Emotional Expression in Children

A participatory UX and interaction design project with 27 children (ages 6–12). We explored how AI and physical play can help children express difficult emotions, feel understood by peers, and create “emotional artifacts” they can keep, trade, and talk about together.

My Role

UX Research Lead · Product Designer · Educator

Methods

Participatory Design (PD1–PD4), Child Co-Design, Prototyping, Iterative Testing

Timeline

2024–2025 (school and after-school settings)

Fig 1. Co-design and prototype testing sessions. Children (a) created stories, (b) designed “empathy games,” and (c)generated stickers to support each other.

Problem Statement

Context & Challenge

Children are under emotional stress (anxiety, isolation, peer conflict) but have fewer safe, kid-controlled spaces to process those feelings together. Research shows play (storytelling, pretending, co-inventing rules) helps kids rehearse and regulate emotions socially.

Gap in Existing Approaches

Most “emotional learning” tools are adult-led and individual (“tell me how you feel”). But children actually process emotion in groups: by joking about it, acting it out, and discovering “you felt that too.” Current tech and AI rarely support that peer-to-peer process.

Our Response & Research Aim

We built and tested playful tools (with physical and AI-powered) that let kids turn feelings into shareable stickers and artifacts they can discuss together. Our guiding question: How can we design AI-mediated technologies that foster spaces where children collaboratively share, and reflect on their emotional experiences through play?

Research Goal

Understand how children naturally express difficult feelings with peers.
Translate those authentic behaviors into design requirements.
Prototype an AI-assisted tool that supports empathy between children — not just self-calming.

Early research board showing quotes, emotion cards, and story prompts

Fig 2. Early framing of the emotional landscape children described.

Design Process

We followed a child-centered design thinking flow: Understand → Observe → Define → Ideate → Prototype → Test. Each phase shows what we did with children, what we learned, and how that moved the design.

Fig 3. Mapping four participatory design sessions (PD1–PD4) into a staged design process.

1 · Understand Mapping children's emotional reality

I started by reviewing HCI / learning sciences work and interviewing educators to map where kids feel emotional strain: embarrassment in front of peers, conflict, fear of “getting in trouble” for reacting. I also examined existing SEL-style tools.

We saw a pattern: most products ask children to report and regulate individually. But children tend to regulate socially through role-play, teasing, co-inventing rules, and rewriting what happened together.

Design direction: Don’t just design for emotional control. Design for emotional connection.

Fig 4. kids need social, low-stakes emotional spaces.

Understand · Takeaways

Kids don’t just want to “calm down.” They want to feel seen and not alone.
Emotional work often happens in groups, not in isolation.
Any tool we build should support shared storytelling, not only self-reflection.

2 · Observe Participatory Design with Children (PD1 & PD2)

Goal: Watch how children already express, negotiate, and support emotions with each other. We ran two co-design sessions.

PD1 · Storytelling Dice

We made oversized “emotion dice” (anger, fear, disgust, sadness, embarrassment). Kids rolled a feeling, then told a real story from their life, acting it out and drawing it.

They loved “negative” emotions. One child said disgust was their favorite because it was “dramatic.” This showed us that kids are comfortable exploring hard feelings when it’s playful and performative.

PD2 · Empathy Game (“You Are Not Alone”)

In small groups, kids designed empathy games. A player shared a stressful moment (“I was nervous before presenting”), and others signaled if they felt the same. Empathy sounded like “same, you’re not weird,”.

Insight: Children kept creating tangible emotional tokens such as drawings, cards, stickers. They treated these like badges that say “this is how I felt.”

That behavior is what led us toward the concept of emotional stickers.

Fig 5. Kids openly shared fear, anger, and embarrassment while creating storytelling dice game.

Fig 6. Children naturally externalized emotions into collectible visual objects.

Observe · Takeaways

Kids want to show emotion, not just name it.
They want to compare experiences and hear “me too.”
They love keeping physical reminders of being supported.

3 · Define Translating insight into design requirements

After PD1 and PD2, we defined a core requirement:

Design requirement:
Give kids a playful, ownable emotional artifact — something they can edit, laugh about, gift to a peer, or keep for themselves. Not a score. Not a “diagnosis.” A thing they control.

We then synthesized our findings into three design pillars:

Curiosity – “What happens if I roll this?” “Can I regenerate it?”
Ownership – “That’s not how I felt. I’ll change it.”
Enjoyment – Humor and cuteness lower defensiveness and make honesty feel safe.

These three became our rubric. Every UI choice had to support curiosity, ownership, and enjoyment.

Fig 7. Synthesis wall: emotional safety emerges when kids feel agency and delight.

Design Principle

The tool should not tell kids how to feel.
It should give them something to talk about together.

4 · Ideate Concept: The AI Emotional Sticker Tool

I designed an interaction model where a child:

Tells a short story about when they felt something intense (angry, scared, embarrassed).
Gets a “feel-better” sticker: an image + message generated by AI.
Edits that sticker or creates one for a friend.

This supports what we call the Campfire Model: the sticker is the shared object in the middle. Kids gather around it, react to it, correct it, and laugh about it — instead of putting one child “on the spot.”

Design shift: AI is not “the authority.” AI is a creative partner that sparks conversation.

Fig 8. “Campfire” flow: the artifact sits in the center, not the child.

Why stickers?

Kids in PD sessions already treated hand-drawn stickers like emotional badges. We amplified that behavior instead of forcing a new behavior.

5 · Prototype Figma flows & AI interaction design

I built mid- and hi-fidelity prototypes in Figma. The design goal was to make the flow feel intuitive, story-driven, and emotionally safe for children. The interface encourages playful exploration while giving children control over how emotions are represented and shared.

Story Input: Children describe what happened and select how they felt (sad, angry, scared, or worried).
AI Output: The system generates a “feel-better” sticker — an image and short supportive message in a kid-friendly tone.
Customization: Kids can edit, regenerate, or restyle the AI output to better match their emotions and stories.

Wireframe · Story Input & Emotional Flow

Low-fidelity wireframe of child storytelling interface

Fig 9. Early wireframe sketch showing how children share short stories and select emotions through icons. The goal was to make the process conversational, visual, and low-pressure.

Process Flow · Feel-Better Sticker Generation

Process flow showing AI sticker generation and feedback loop

Fig 10. Process flow for the Feel-Better Sticker tool, mapping how children’s stories and emotions are transformed into AI-generated stickers that invite reflection, sharing, and discussion.

C.3 Technical Overview of the AI Tool

The AI Emotional Sticker tool was developed as a web-based prototype using React/Node.js on Replit. It connects children’s self-expression with generative AI models through a simple, safe, and creative workflow:

Input stage: Children upload their activity sheet, select an emotion, and write a short description of their story.
Generation stage: OpenAI’s ChatGPT-4o model produces a message and image prompt describing the child’s emotion, which is then sent to DALL·E 3 to generate a visual sticker. All outputs are safety-checked using OpenAI’s moderation tools.
Display stage: The tool presents the generated sticker, message, and emotion summary on a results page that encourages reflection and sharing.
Regeneration stage (optional): Children can refine or recreate stickers and messages using ChatGPT-4o or Google Gemini 2.0 Flash, allowing them to experiment and personalize emotional expression.

This iterative loop between storytelling, AI generation, and playful regeneration reflects the project’s central idea: AI as a creative companion that helps children see and reshape their emotions — not a judge that defines them.

Prototype link: Try the AI Emotional Sticker Prototype

Design Insight: Simplicity Enables Emotional Flow

During prototyping, we learned that simplifying interaction steps helped children stay emotionally engaged. Rather than complex menus or choices, the interface guided them gently from storytelling to reflection. Minimal friction meant more focus on what they felt—not how to navigate the tool.

The prototype emphasized customization as agency. Each interaction invited children to shape the sticker’s message, image, and tone—making the design feel personally theirs. This flexibility turned the AI into a creative partner, not an authority. By giving children the ability to regenerate, rewrite, or restyle outputs, the tool fostered ownership and emotional authorship—helping them see their feelings as something they could represent, change, and proudly share.

6 · Test Iterating with children

We tested two modes of interaction: self-sticker (“for me”) and peer-sticker (“I make one for you”). Children compared how they portrayed their own emotions versus how peers represented them. This contrast surfaced moments of reflection: “I wasn’t mad — I was scared.” The dialogue was playful, yet deeply introspective.

By externalizing emotion into a sticker, critique shifted from the person to the artifact. Children could safely negotiate meaning without fear of being wrong. One child laughed and said, “Your sticker made me braver.” The design thus enabled social reflection—a form of empathy that emerged through shared interpretation, not instruction.

Interestingly, children also critiqued the AI itself: “The AI’s message is weird.” “It sounds like a teacher.” “I want it to be slow.” These comments reframed AI not as a storyteller, but as a co-learner—a participant open to feedback, adaptation, and humor.

The tangible stickers amplified engagement. Children shared them, traded them, and displayed them proudly—transforming digital emotions into collective emotional artifacts. Physicality became the bridge between personal reflection and social connection. As one said, “It makes me feel happy because I made it myself.”

Children explore the AI Emotional Sticker tool together — Fig 11. Kids used AI output as a safe third party they could critique together.

Printed AI stickers — Fig 12. Stickers became emotional keepsakes and gifts.

Impact in testing

Kids openly discussed fear, anger, and embarrassment — without shutting down.
They negotiated “this is how I see your feeling” vs “this is how I actually felt.”
The AI became a social tool, not a judge. They laughed at it, fixed it, improved it together.