Watson

Watson

Inspiration

The level of propaganda, false and misleading information on the internet increases with the advances in new technologies. While many people rightly fear the dangers of a powerful AI in the wrong hands, we believe the same technology can be used to protect regular people from it. In this project we show how LLMs coupled with web search can be effective in identifying false information on the web.

What it does

Watson is a browser plugin that works like Twitter's Community Notes, but for media articles. We augment LLMs with relevant sources from Brave Search to identify, explain, and provide a source of false and misleading statements. First, Claude flags potential issues in the article. Then we perform relevant web search for sources from data analytics platforms, wikipedia and news. Finally, we have Claude assess whether the statements are false based on quotes from a source and explain its reasoning.