- On Wednesday, Tom Alison, the head of Facebook, announced that the company is developing an AI model to enhance its entire video platform.
- During a presentation at Morgan Stanley’s tech conference in San Francisco, Alison revealed that this AI initiative is a key component of Meta’s technology roadmap through 2026.
- Meta’s CEO, Mark Zuckerberg, recently disclosed that the company is heavily investing in Nvidia’s graphics processing units to fortify its AI infrastructure.
Meta’s substantial commitment to artificial intelligence encompasses the creation of an AI system tailored to drive Facebook’s video recommendation engine across all its services, as per a company representative’s statement on Wednesday.
Tom Alison, Facebook’s head, elaborated that as part of Meta’s technology roadmap extending to 2026, they are working on an AI recommendation model to power various offerings, including the Reels short video service akin to TikTok and longer, conventional videos.
Historically, Meta has employed distinct models for different products like Reels, Groups, and the primary Facebook Feed, as outlined by Alison at the tech conference hosted by Morgan Stanley in San Francisco.
As Meta delves deeper into AI innovation, the company has allocated billions towards Nvidia graphics processing units (GPUs), the predominant choice for AI researchers training large language models (LLMs) such as the ChatGPT chatbot developed by OpenAI and other generative AI models.
Alison highlighted that the initial phase of Meta’s tech roadmap involved transitioning from conventional computer chips to GPUs for the recommendation systems, thereby enhancing overall product performance.
The surge in interest surrounding LLMs prompted Meta to recognize their potential in handling vast amounts of data and diverse activities like conversational interfaces. This realization led to the development of a comprehensive recommendation model tested on Reels, resulting in an 8% to 10% increase in watch time within the core Facebook app.
The ongoing efforts in re-architecting the system mark “phase 3” for Meta, focusing on validating the technology and extending its application across multiple products.
“Instead of just powering Reels, we’re working on a project to power our entire video ecosystem with this single model, and then can we add our Feed recommendation product to also be served by this model?” Alison stated, emphasizing the potential for more engaging and relevant recommendations with improved responsiveness.
The strategic accumulation of GPUs by Meta will support broader generative AI endeavors, including the development of digital assistants.
Exploratory projects within the realm of generative AI involve integrating advanced conversational tools into the core Feed, enabling users to seek further information seamlessly. Additionally, Meta is exploring the integration of AI chatting tools within Groups to facilitate interactions within specific community settings.
Alison envisions a scenario where generative AI can thrive in a collaborative consumer environment, enhancing user experiences across various Meta platforms.
