About Debatron High
Inspiration
In high school, I noticed classmates shying away from debates and public speaking—even when they had great ideas. At the same time, AI debate highlights and “argument battles” were going viral on social media. I wanted to combine the excitement of AI-driven conversation with a safe practice ground for students to sharpen their critical-thinking and speaking skills. Thus, Debatron High was born: an edutainment app that turns debate prep into an engaging, gamified experience.
What I Learned
- Prompt Engineering & LLM Tuning: Crafting prompts so the AI could play a believable “opponent” taught me how subtle wording changes impact response quality.
- Real-Time Feedback Loops: Implementing criteria (clarity, logic, tone) and mapping AI “grades” to constructive suggestions deepened my understanding of evaluation metrics.
- Gamification Psychology: Designing a leveling and badge system revealed how small rewards (e.g., “Mic Drop” badge) can keep learners motivated.
- UI/UX for Teens: Balancing a fun “Gen Z” aesthetic with clear navigation reinforced the importance of audience-focused design.
How I Built It
- Core AI Engine:
- Used AWS PartyRock to spin up a Chat-based AI that could adopt “For” and “Against” personas.
- Developed prompt templates that dynamically inject debate topics and sides.
- Used AWS PartyRock to spin up a Chat-based AI that could adopt “For” and “Against” personas.
- Feedback & Scoring Module:
- Defined rubrics for clarity, logical structure, and persuasive language.
- Employed PartyRock’s function-calling abilities to have the AI return JSON-formatted scores and suggestions.
- Defined rubrics for clarity, logical structure, and persuasive language.
- Frontend Interface:
- Designed in Figma with neon-accented cards, simple icons, and emoji-infused button labels.
- Built a lightweight React app (hosted on Netlify) that communicates via REST to the PartyRock backend.
- Designed in Figma with neon-accented cards, simple icons, and emoji-infused button labels.
- Gamification Layer:
- Created a leveling system (Rookie → Legend) and 12 custom badges.
- Stored user progress in a simple Supabase database.
- Created a leveling system (Rookie → Legend) and 12 custom badges.
Challenges Faced
- Maintaining Debate Flow: Early iterations produced AI responses that were either too generic or too verbose. I refined prompts and added length constraints to keep exchanges snappy.
- Balancing Fun & Academic Rigor: It was tricky to weave in Gen Z slang without undermining the educational tone. I settled on “subtle emojis” and “playful micro-copy.”
- Real-Time Performance: Latency spiked when the AI generated feedback on lengthy arguments. I introduced caching for common topic snippets and limited free-mode turns to keep things smooth.
- UI Responsiveness: Designing for both mobile and desktop required rethinking card layouts several times—especially the “live scoreboard” during a debate round.
Debatron High represents the intersection of AI innovation and modern edutainment. I hope it helps students everywhere gain confidence and sharpen their voices—one AI-powered battle of wits at a time!
Built With
- amazon
- amazon-web-services
- bedrock
- gpt
- netlify
- openai
- partyrock
- partyrock)
- tailwind
Log in or sign up for Devpost to join the conversation.