EmoTune
EmoTune is an interactive web-based learning tool designed to teach children about AI bias. The project aimed to simplify complex AI concepts through playful, intuitive activities that encouraged young learners to explore how to interact with AI classifiers.
Case Study Context
Role: UX Designer & Developer
Industry: AI Education
Timeline: 4 weeks (Nov. 2023 - Dec. 2023)
Tools: Figma, Google Suite, VSCode (JavaScript)
Project Focus
"Design and develop a playful, engaging tool that teaches children about AI bias"
Simplifying Complex AI Concepts for Children
Children are increasingly exposed to Artificial Intelligence (AI) through various technologies, yet most have little understanding of the biases embedded within AI integrations. For our Directed Research Group, our goal was to design and develop a learning tool that was both engaging and educational, allowing children to grasp how AI is not always correct and can be biased — especially when it comes to emotional interpretations.
Problem Statement:
“How might we create an interactive tool that teaches children about AI bias in a way that is fun, accessible, and educational?”
Why it Matters: Raising Awareness of AI Bias Early:
AI is becoming an integral part of daily life, yet many adults and children alike are unaware of the biases that can arise within these systems. By simplifying complex concepts, we’re equipping the next generation with the tools they need to question technology’s role in shaping their world.
Source: Power of Zero & Boston Children's Wellness Lab (Pulse Survey)
Making AI Bias Fun & Understandable — in 4 Weeks
Creating an engaging tool that simplified AI bias for children posed several key challenges:
Simplifying Complex Concepts: Breaking down AI bias into terms and visuals that children could easily understand without overwhelming them was a critical challenge.
Designing for Playfulness and Education: Striking the right balance between delivering educational content and maintaining a sense of play was key to keeping children engaged.
Ensuring Accessibility: The interface had to be intuitive enough for young children to navigate independently.
Limited Timeframe: Developing the Minimum Viable Product (MVP) within 8 weeks added an additional layer of pressure.
From Insights to Interactive Learning
We collaborated with the UW Learning, Epistemology, and Design (LED) Lab, which had ongoing research focused on helping children (ages 6-12) understand AI bias. Our first step was to glean insights from their findings and explore AI tools that could enhance our educational web application.
These insights guided our decision to select a speech emotion API (Application Programming Interface) that analyzes voice inputs and generates text outputs of detected emotions, aligning perfectly with our goal of giving children hands-on exposure to AI bias in a tangible and engaging way.
Expanding the Approach:
Initially, our sitemap had only one branch dedicated to speech emotion detection. However, after consulting with our mentor and reviewing research insights, we made a pivotal decision: we broadened our approach by implementing three distinct APIs, each representing a different type of voice detection AI.
Challenges and Opportunities:
This decision introduced new challenges, particularly given the limited timeline and the added complexity. Incorporating two additional APIs required us to develop more:
Character designs
Flavor text
Narration
Despite the extra workload, we recognized this as an opportunity to enrich the experience. By allowing children to compare the outputs of different AIs, we exposed them to varying levels of AI bias.
Empowering Critical Thinking:
By encouraging children to compare & contrast how different AIs interpret the same voice input, we gave them the tools to:
Make informed decisions
Engage in meaningful discussions about bias in technology
This approach broadened the EmoTune experience, ensuring children could explore AI bias from multiple perspectives. Ultimately, it aligned with our goal of nurturing a more inclusive and informed generation of technology users.
Crafting a Gamified, Interactive Experience
Building on our research findings, we moved into the design phase, focusing on creating an engaging, intuitive platform that catered to young children.
Initial Wireframes (Week 2):
We developed basic wireframes that featured a simple platform for children to record their voices and receive AI-generated emotion outputs. Our initial focus was placed on simplicity, as our target audience had limited familiarity with technology.
Visual Enhancements:
We replaced circular indicators with horizontal bars to provide clearer visual feedback on the AI’s analyses. This change made it easier for children to follow and compare the AI processes.
Design Pivot — Gamification:
After discussions with our mentor, we introduced a gamified approach to boost engagement. This involved creating three distinct characters:
AIden (speech-emotion AI)
Woofster (language-accent AI)
Monstro (English-accent AI)
These characters guided children through prompts, offered feedback, and tracked progress, turning the learning experience into a fun, interactive journey.
The Final Solution
EmoTune’s final design was an engaging, interactive platform that simplified AI bias for children through playful characters and voice-input interaction. Through the step-by-step narrative, the characters demonstrate how AI could misinterpret emotions and speech, helping children see the potential for bias in real-world applications.
Key Features:
Interactive Character Engagement: Children interact with characters to explore AI’s interpretation of their voices.
Simple, Intuitive Design: Navigation and activities are designed with young users in mind, ensuring accessibility and ease of use.
Try it Yourself!
Validation:
By week 4, we were ready to take our demo and conduct a usability test with UW KidsTeam. We had the opportunity to test our launched website with a diverse group of children, including a 6-year-old girl, an 8-year-old boy, and a 10-year-old boy. Their feedback provided invaluable insights that both validated our efforts and highlighted areas for improvement.
Next Steps & Considerations
As we look ahead, several areas offer exciting opportunities for expanding EmoTune’s reach and effectiveness:
Future Improvements:
What I Learned from EmoTune
The development of EmoTune reinforced several important lessons about user-centered design, particularly for young learners:
The EmoTune project not only allowed us to build an engaging educational tool but also highlighted the power of design in shaping the future of learning. By merging technology and creativity, we opened new avenues for children to explore complex subjects like AI bias in a way that resonates with them. This project reinforced our belief in the potential of user-centered design to make learning accessible, interactive, and fun.