<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/f27093cb0ae74422b088fbefa632fe2a&quot; frameborder=&quot;0&quot; width=&quot;1920&quot; height=&quot;1440&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1440</height><width>1920</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1440</thumbnail_height><thumbnail_width>1920</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/f27093cb0ae74422b088fbefa632fe2a-cc9b3ccc3684c5ec.gif</thumbnail_url><duration>175.772</duration><title>Hindsight: Tracking Video Narratives and Contradictions</title><description>This Loom introduces Hindsight, a video understanding tool focused on issue tracking and narrative analysis. The founder explains that it uses a large dataset from numerous YouTube channels, with transcripts for more than 20,000 videos, and organizes content into clusters by topics like government reform and social justice. For specific issues such as AI regulation, users can compare multiple moments in different videos by jumping to exact frames and reviewing the associated transcript. It also supports tracking whether key figures contradicted earlier statements, using an accountability feature to compare what someone said before versus after, and allows filtering and searching across videos over time.</description></oembed>