Home Page
Comments Tab
Search Results
Comparative study
Cited Sources
System Architecture

Unbiased

All sides. Your judgment. Three angles. One clear view.

Inspiration

Every day, millions read news that shapes their worldviews, often without realizing how much bias shapes the message. Unbiased was born from a simple question: “What if readers could see the same story from every angle before forming an opinion?” Our goal is to rebuild trust in information by empowering judgment, not replacing it.

What It Does

Unbiased is a topic-driven “news shorts” web app designed to reduce bias in online information.

For every trending topic, the app displays three concise takes:

For – Arguments supporting the topic
Against – Counterarguments opposing it
Unbiased/Neutral – Factual background and middle-ground insights

Each summary is linked to its source for verification.
Integrated with ElevenLabs API, which provides natural-sounding text-to-speech.
Color psychology: For in Purple, Against in Orange → prevents subconscious bias.
This helps reduce polarization and misinformation while encouraging independent thinking

How We Built It

We built Unbiased as a blend of AI, psychology, and accessibility to promote transparent information. The frontend, made with Next.js and Tailwind CSS, ensures a clean and responsive design. On the backend, Next.js API routes connect with the Gemini API for real-time topic discovery, while Supabase caches trending topics for up to eight hours to improve speed and efficiency.
To make news more inclusive, we used the ElevenLabs API for natural text-to-speech summaries so users can listen as well as read.
Core features include a trending dashboard for top stories, a smart search to explore any topic, and AI-generated summaries that show For, Against, and Neutral perspectives. Users can also share their thoughts through the community comment system, creating open, respectful discussions around the news.

Challenges We Ran Into

One of the main challenges we faced was maintaining a neutral tone in AI-generated summaries. Getting the model to consistently separate opinions from facts required several rounds of prompt-tuning and evaluation. We also struggled with hyperlink formatting, as ensuring that each article source displayed cleanly and correctly linked back to its origin required both prompt adjustments and custom parsing logic.
On the development side, we encountered Git merge conflicts while managing parallel feature updates from different teammates, which briefly disrupted our workflow. Coordinating changes across the backend (API routes) and frontend (UI updates) taught us the importance of version control discipline and communication during rapid development. Despite these hurdles, resolving these issues strengthened our collaboration and made our final product more stable and polished.

Accomplishments We’re Proud Of

We built a functional and visually balanced news summarizer that seamlessly connects every stage of the user experience. The system integrates Gemini for real-time news discovery, Supabase for efficient data storage and caching, and ElevenLabs for natural text-to-speech conversion. Additionally, we implemented a community-driven comment system that encourages civil, open discussions. Altogether, the project delivers a smooth, accessible experience that blends AI, UX psychology, and journalism into one cohesive pipeline.