Meta Is Developing a Giant AI Model to Power Its 'Video Ecosystem' Across Its Platforms, Including Facebook

By Jace Dela Cruz

Mar 07, 2024 01:56 AM EST

Meta is heavily investing in artificial intelligence (AI), including developing an AI system to power its video recommendation engine across all its platforms, including Facebook. 

This illustration photograph taken on October 30, 2023, shows the Meta (former Facebook) logo on a smartphone in Mulhouse, eastern France.
(Photo : SEBASTIEN BOZON/AFP via Getty Images)

Meta Banks on Large AI Model

According to CNBC, Tom Alison, the head of Facebook, revealed at Morgan Stanley's tech conference in San Francisco on Wednesday that Meta's technology roadmap through 2026 includes creating an AI recommendation model capable of powering both its Reels short video service and longer videos.

As part of this AI initiative, Meta has invested billions of dollars in Nvidia graphics processing units (GPUs), which AI researchers widely use to train large language models (LLMs) like ChatGPT.

Alison outlined the company's phased approach to this initiative, with "phase 1" involving transitioning to GPUs to enhance product performance. Meta then developed a new model architecture tested on Reels and observed an 8% to 10% increase in Reels' watch time on the Facebook app.

Currently, Meta is in "phase 3" of re-architecting its system, aiming to validate and implement the technology across multiple products within its video ecosystem.

READ NEXT: Meta Platforms Experience Widespread Outage; Facebook, Instagram, and Threads Inaccessible For 2 Hours 

Powering the Video Ecosystem of Meta

The goal is to create a single AI model capable of powering the entire video ecosystem, offering users more engaging and relevant recommendations. 

Meta's investment in GPUs will also support broader generative AI efforts, including developing digital assistants integrated into various products like the core Feed and Groups. 

"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 noted, as CNBC reported.

"If we get this right, not only will the recommendations be kind of more engaging and more relevant, but we think the responsiveness of them can improve as well," he added.

READ MORE: Meta's Oculus Faces Repeated Hacks, Prompting Safety Reevaluation By Users 

© 2024 VCPOST, All rights reserved. Do not reproduce without permission.

Join the Conversation

Real Time Analytics