 Our fifth presenter is Jeremiah Milbauer, whose presentation is titled Knowledge Augmentation with Artificial Intelligence. From writing to libraries to telecommunications and the web, humanity has always invented new tools to organize and share knowledge. In my thesis, I argue that artificial intelligence could be the next great knowledge tool if we use it to augment our intelligence rather than replacing it. So my work focuses on AI systems that help us explore the connections between ideas and across contexts. So imagine you're reading a document and every time you encounter a new idea, it's highlighted and annotated. Imagine automatic fact verification embedded in every news article, explanations in every lease agreement, or background info summaries when you read a scientific paper from a new field. No more helpless googling, no more endless Wikipedia tabs. Artificial intelligence has the potential to embed an expert-level knowledge discovery process directly into our lives. So I began my research by looking at high-level connections across communities. I used natural language processing to train conceptual maps of ideologically distinct communities and then mathematically aligned those maps to find where the communities agreed and disagreed. This technique was able to automatically discover instances of coded in-group language as well as conceptual analogies across community divides. Essentially instances where people use different terminology to talk about the same shared underlying ideas. Now these techniques could help scientists connect new ideas to their existing expertise or help all of us explore new information when we don't really know how to search for it. My work has also looked at sort of specific connections between ideas in knowledge artifacts like documents. So normally comparing documents at this very, very fine detailed level would be too computationally expensive for artificial intelligence. So I developed an algorithm which accelerates this by storing and reusing the artificial intelligence's own internal representation of each sentence. And this year as a night fellow at CMU's Center for Informed Democracy I applied those ideas to build a tool which turns news articles into documents like you see here with annotated links that connect disputed and supported information across a whole collection of related news articles. Going forward I plan to collaborate with law librarians to apply these same techniques to trace the lineage of ideas through the history of Supreme Court cases. So from law to the news to science and beyond if we use artificial intelligence to augment our intelligence we can accelerate discovery and democratize access to information for everyone. Thank you for listening.