Friday, July 19, 2024
HomeAIMeta is building a giant AI model to power its ‘entire video...

Meta is building a giant AI model to power its ‘entire video ecosystem,’ exec says

A significant investment in artificial intelligence includes the construction of an AI system to run Facebook’s entire video recommendation engine across all of its platforms, a company executive said Wednesday.  

Facebook CEO Tom Alison stated that Meta’s “technology roadmap that goes to 2026” includes establishing an AI recommendation algorithm that can fuel both the company’s TikTok-like Reels short video service and more traditional, lengthier films.  

Alison stated onstage at Morgan Stanley’s innovation conference in San Francisco that Meta has traditionally employed a different model for each of its products, including Reels, Groups, and the primary Facebook Feed. 

As part of Meta’s ambitious AI push, the corporation has spent billions of dollars on Nvidia graphics processing units, or GPUs. They’ve become the primary chips used by AI researchers to train the massive language models that power OpenAI’s well-known ChatGPT chatbot and other generative AI models.  

Alison stated that “phase 1” of Meta’s tech strategy entailed moving the company’s current recommendation algorithms from more traditional computer processors to GPUs, which would assist to improve product performance overall. 

As interest in LLMs grew last year, Meta executives were impressed by how these large AI models could “handle lots of data and all kinds of very general-purpose types of activities like chatting,” Alison explained. Meta saw the potential for a massive recommendation model that could be applied across products and, by last year, had constructed “this kind of new model architecture,” Alison said, adding that the business tested it on Reels.  

This new “model architecture” assisted Facebook in obtaining “an 8% to 10%  gain in Reels watch time” on the primary Facebook app, which Alison claimed helped indicate that the model was “learning from the data much more efficiently than the previous generation.” 

“We’ve really focused on kind of investing more in making sure that we can scale these models up with the right kind of hardware,” he went on to say.  

Meta is presently in “phase 3” of its system re-architecture, which includes attempting to test the technology and implement it across many 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,” she said. “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.” 

Alison explained how it will function if it is effective, saying, “If you see something you like in Reels and then return to the Feed, we can show you more content like that.” 

Some generative AI projects Meta is considering include incorporating more sophisticated chatting tools into its core Feed, so a person who sees a “recommended post about Taylor Swift,” could perhaps “easily just click a button and say, ‘Hey Meta AI, tell me more about what I’m seeing with Taylor Swift right now.'”  

Meta is also experimenting with integrating its AI talking tool into Groups, which may allow a member of a Facebook baking community to ask a dessert-related question and receive an answer from a digital assistant.  

“I think we have the opportunity to put generative AI in kind of a multiplayer kind of consumer environment,” she remarked. 



Please enter your comment!
Please enter your name here

Most Popular

Recent Comments