Common Ground

Inspiration

We began by researching the most pressing social issues dividing America today, using tools like Gemini, ChatGPT, and Claude to understand the landscape of polarization. We realized that in heated debates, finding a middle path is incredibly difficult. We were inspired to create a tool that acts as an entirely unbiased mediator, one that isn't swayed by human emotion, possesses a vast knowledge base, and can objectively guide differing viewpoints toward mutual understanding.

What it does

Common Ground is an AI-powered mediation app designed for anyone trying to bridge the gap between differing perspectives. It objectively listens to each side of an argument and actively "steel-mans" each user's case, presenting their strongest possible argument so the opposing side can genuinely understand their perspective. Because it acts as a truly unbiased mediator with access to comprehensive knowledge bases, the app also provides real-time fact-checking to keep the conversation grounded in reality while fostering empathy and collaboration.

How we built it

We started our development process with extensive brainstorming and research using multiple LLMs. Once we had our core concept, we drafted a highly detailed list of project requirements. We then took this comprehensive document and uploaded it to Claude AI, which we used to generate and build the code for our entire application from the ground up.

Challenges we ran into

Because we relied on Claude AI to generate the bulk of our project code, our biggest hurdle was dealing with AI hallucinations. Initially, the AI didn't have the full, overarching context of the application at its disposal, leading to disjointed code and unexpected errors. We had to quickly shift gears and adapt our workflow to solve this. We addressed the issue by creating a meticulous, detailed requirements document that Claude could constantly reference, effectively grounding the AI and keeping the build on track.

Accomplishments that we're proud of

We're incredibly proud of successfully leveraging AI not just as a research assistant, but as our primary development tool to build a fully functional application. Conceptualizing a tool that directly tackles a complex human issue like social polarization, and effectively prompting an AI to build the architecture for it, was a massive win for our team.

What we learned

We learned a crucial lesson about AI-assisted development: context is everything. You cannot simply ask an AI to build a complex app without giving it strict parameters. We learned how to properly engineer our prompts and structure our development cycle by prioritizing detailed, clear requirement documentation before writing a single line of code.

What's next for Common Ground

If we had more time, our immediate next step would be deployment. We plan to take the app live, publish it to a public-facing website, and get it into the hands of real users so they can start having more productive, empathetic, and fact-based conversations.