The internet has been bad for modern discourse. Opinions and spin -- perspectives -- are being presented as facts. There is a complete lack of transparency in the media. The line between someone's perspective and an actual fact has been blurred. 

If you look in the news about AI, the consensus is pretty gloomy. This project is an attempt to combine data from the internet, machine learning, and a bit of Chat GPT to create a product that works for good. 

I believe that the solution lies in computation. Computational journalism. A problem caused by social media algorithms can only be solved by a counter-algorithm. That is what we have made with Perspectives. 

Basically, this app used all the terrible articles on the internet as data to train a model that could spit out words that are typically used in biased contexts. These words, in conjunction with the WordMoversDistance algorithm, can find the "distance" between biased words and an article. The smaller the distance, the more biased the article. 

This gives you, the user, much more transparency in your news diet. You can see the facts, and then you can see different perspectives. 

For much more unhinged version of this description, you can visit the blog on the TopNews page :)

Credit to Los Gatos High School junior David Roh for designing the TopNews about page, and Los Gatos High School junior Edwin Hou for various ideas