• 

  start screen

• 

  loading screen

• 

  first round of a debate

• 

  second round of debate

• 

  game over screen

DebateBros: AI-Powered Debate Judge

Inspiration

The inspiration for our project was to create a game that leverages AI to enhance debating skills. Debating is a critical skill that improves logical reasoning, communication, and confidence. However, fair and unbiased judging can be challenging, especially in informal settings.

What it does

DebateBros is a game where two players engage in a debate, one arguing for (Pro) and the other against (Con) a given topic. The AI acts as a neutral judge, scoring the arguments on:

• Logical coherence
  
• Evidence quality
  
• Persuasiveness
  
• Relevance to the topic
  

How we built it

We built DebateBros using:

• Python – Core logic of the game
  
• Pygame – For rendering the user interface and visuals
  
• Ollama (LLaMA 3) – Local large language model for evaluating debates and generating topics
  
• Requests Library – For handling AI-generated responses
  
• JSON Processing – Formatting AI responses into structured outputs
  

Challenges we ran into

1. Finding the Right AI Model

   • Initially, we explored OpenAI’s API, but it had rate limits and required constant internet access.
     
   • We then switched to Ollama (LLaMA 3) to run AI locally, but it required significant storage space and proper setup.
     

2. Game UI and Integration

   • Combining Pygame with AI-generated text required careful design to ensure a smooth experience.
     
   • Scaling background images and rendering text properly within the debate interface took multiple iterations.
     

Accomplishments that we're proud of

• Successfully integrating AI-powered debate judging with structured feedback.
  
• Creating an interactive debate experience with topic generation and evaluation.
  
• Running the AI model locally without relying on external APIs.
  
• Implementing a visually appealing game interface with Pygame.
  

What we learned

• Prompt engineering: How to design prompts for structured AI responses.
  
• Large language models (LLMs): Setting up and using Ollama locally.
  
• Game development: Using Pygame to create an engaging experience.
  

What's next for DebateBros

• Multiplayer Mode: Allowing online debates between players.
  
• Smaller AI Model Support: Exploring lightweight AI models to reduce storage needs.
  
• Voice Input & Speech-to-Text: Let users debate using voice instead of typing.
  

Saba Tchumburidze, Saba Danelia, Nani Chkhenkeli

Built With

• llama3
• ollama
• pygame
• python