π About the Project
As someone who has always loved debates but also felt nervous about speaking or unsure if my points were strong enough β I wanted to build something that could actually help people improve in real time. That's how AI-Debate Arena was born during this hackathon.
Itβs a platform where you just enter (or even speak!) your argument, and the system evaluates it live using AI β giving feedback on how logical, emotional, and convincing your words are. No judges. No bias. Just insights.
π‘ What inspired me?
The idea came from a debate session where I realized we often rely on subjective feedback. What if we could make that more analytical and data-driven β but still easy to use and fun?
Plus, I wanted to make a tool especially for students like me who participate in debates or MUNs and want to practice smarter.
π οΈ What I Learned
- Prompt engineering is powerful but tricky β I had to iterate to get consistent output from the AI.
- Real-time feedback requires performance tuning, especially when using APIs and rendering charts.
- Design matters: I learned how to balance UX with functionality.
π» How I Built It
- I created the frontend using HTML, TailwindCSS, and plain JavaScript to keep it simple and fast.
- The backend is a Flask API that connects to OpenAI and returns scores like sentiment, logic, and persuasion.
- I added Chart.js for visual feedback and used the Web Speech API to let users speak their arguments.
- Supabase Auth helps with basic login/signup (planned for the extended version).
π Challenges I Faced
- Getting the AI to respond in JSON format and parsing it correctly.
- Cross-origin issues between frontend and backend while testing locally.
- Making the chart update dynamically after each input (took me a while!).
π Accomplishments I'm Proud Of
- I built something that I would genuinely use to prepare for debates.
- Real-time analysis that feels natural and helpful β not robotic.
- Clean and simple UI with meaningful results, not just random scores.
π± What's Next?
- Adding debate practice rooms where people can compete and get AI + peer scoring.
- Voice analysis to give tips on tone, pauses, and emphasis.
- Saving performance history for long-term improvement.
π©βπ» Made with love, caffeine, and late-night Google searches π
Built With
- api
- chart.js
- flask
- html
- javascript
- openai-api
- speech
- supabase
- tailwindcss
- web
Log in or sign up for Devpost to join the conversation.