/** * wael added 2-28 */ /** END */

Meta Ventures into AI: Building a Unified Model to Revolutionize Video Recommendations

Meta, the parent company of Facebook, is embarking on a groundbreaking endeavor to revolutionize its video recommendation engine across all platforms. Revealed by a senior executive, Meta is developing an artificial intelligence (AI) system poised to power its entire video ecosystem, encompassing both short-form content akin to TikTok and longer traditional videos.

Tom Alison, Head of Facebook, unveiled Meta’s ambitious project during Morgan Stanley’s tech conference in San Francisco. Historically, Meta utilized distinct models for each of its products. However, recognizing the need to evolve, Meta is consolidating its approach with a unified AI model, as reported by CNBC.

The move mirrors an endeavor to catch up with TikTok, renowned for its formidable recommendation algorithm that leverages diverse machine-learning techniques to deliver personalized content to users. TikTok’s popularity has posed a significant challenge to Meta’s platforms, prompting the strategic expansion of Reels and, now, the development of a comprehensive AI model.

To power this initiative, Meta has invested billions in Nvidia graphics processing units (GPUs), crucial for training large language models utilized in generative AI chatbots like ChatGPT. By transitioning to GPUs, Meta aims to enhance the performance of its recommendation systems significantly.

Initial tests conducted on Reels showcased promising results, with Meta observing an 8-10% increase in Reels watch time. Encouraged by these findings, Meta aims to extend the application of its new AI model to various products within its video ecosystem.

Tom Alison underscores the potential of the unified model to deliver more engaging and relevant recommendations while enhancing responsiveness. Furthermore, Meta seeks to integrate its AI advancements into its chatbot assistant, enabling users to access relevant information seamlessly. For instance, users presented with content about Taylor Swift can effortlessly inquire about her next tour date through the Meta Assistant, eliminating the need for external searches or waiting for responses from other users.

In group settings, Meta’s AI-driven assistant stands ready to expedite queries, exemplified by promptly addressing baking-related questions. By leveraging AI, Meta envisions a more dynamic and interactive user experience, positioning itself at the forefront of innovation in the realm of social media and video content delivery.

#MetaAI #Meta #AI #VideoRecommendations #UnifiedModel #ArtificialIntelligence #MetaAssistant #TikTok #Reels #MachineLearning #Nvidia #GPUs #SocialMedia #TechInnovation

For more insights and updates, visit our KI Design blog here.
Stay connected with us on Twitter for the latest news and discussion