Agentic Curation – What's a beast without an algorithm? (www.mindsinmachines.ai)

Andrej Karpathy recently launched HackerTube, an innovative platform that curates YouTube videos based on recommendations from Hacker News comments. This concept not only showcases how easily new AI-driven content curation can be implemented but also highlights a widespread demand for better content discovery mechanisms amidst the overwhelming amount of available information. Unlike existing algorithms optimized for user engagement, HackerTube relies on a straightforward rule: if a video is recommended on Hacker News, it gets included. Such a method appeals to users seeking quality over quantity in their viewing experiences. The significance of HackerTube lies in its potential to inspire a new era of content filtering through community engagement, leading to the birth of additional projects like HackerBooks and Wikihack. Moreover, the essay explores the concept of "Agentic Curation," where personal AI agents could serve as customized recommendation systems tailored to individual preferences. By experimenting with multiple AI tools, Karpathy discovered that diverse agents arrived at similar high-impact video suggestions, indicating the promise of personal recommendation engines that prioritize meaningful content discovery, ultimately giving users more control over their media consumption.