Beyond Bias: Assessing News Through Constructive Dialogue

As a group of three from The Browning School, we were inspired by our dialogue class and the Perspectives program from the Constructive Dialogue Institute, along with the speakers from this morning who emphasised the importance of starting with real problems and designing for real-world impact. We have noticed that almost all bias detectors focus on labeling content as "left" or "right," which, in most cases, reinforces polarization by highlighting extreme differences. Our goal was to take a constructive dialogue approach by recognizing that people across the political spectrum often share values like freedom, rights, and safety, but prioritize them differently, and build a tool that encourages understanding rather than judgment.

This website allows users to enter a news article or topic, gathers a series of varied sources and assesses bias on a continuous left-right spectrum with more than 20 possible positions. Instead of classifying articles from a binary standpoint, it provides side-by-side summaries from both left and right perspectives, explains why someone might believe each side of the argument, and includes source links as evidence. We built our project using Python along with NiceGUI for the interface, SerpAPi and BeautifulSoup for source collection, and an LLM tool for analysis. A main challenge that we faced was avoiding oversimplification while keeping our interface both clear and intuitive. Through this process, we learned how design choices can either deepen division or support constructive dialogue, and how thoughtful UI and framing can help bridge that gap.