Decoding X’s algorithm: Here’s how Grok open-sources and curates the “For You” feed

Elon Musk's platform, X, has taken a significant step towards transparency by revealing the inner workings of its 'For You' feed. The company has made its algorithm public on GitHub, providing insight into how content is curated for users around the globe. This announcement marks a pivotal shift from established manual processes to a more sophisticated AI-driven approach. The newly open-sourced algorithm operates through a structured 10-step process, as outlined by X. Initially, the system collects personal data from users, including the accounts they follow and their interactions with various posts through likes, shares, and comments. This data forms a digital profile that aids in tailoring the content shown to each user. Following this, the algorithm utilizes a high-speed system known as 'Thunder' to gather posts from users’ networks, alongside identifying relevant content from outside their immediate connections. This involves leveraging artificial intelligence to pinpoint posts from unfamiliar accounts that resonate with the user's interests. Moreover, the algorithm enriches the selected posts with additional context, such as the user's subscription status, whether a post is behind a paywall, and metadata regarding video quality. It also filters out duplicates, outdated content, and any posts that the user has muted, ensuring the feed remains fresh and relevant. The Grok AI model plays a crucial role in ranking these posts by calculating engagement likelihood scores. This advanced model assesses each post's potential for interaction, enhancing the visibility of high-quality content while minimizing the reach of less relevant material. Ultimately, the 'For You' feed is generated from the top-ranked posts, with a final safety check in place to eliminate any deleted, spam, or inappropriate content from appearing in users' feeds.