Meta introduced the latest iteration of its custom-made chips designed for artificial intelligence workloads.
The new version marks a performance improvement over the previous generation of this AI tool and helps power Meta’s ranking and recommendation ads models on Facebook and Instagram, the tech giant said in a Wednesday (April 10) company blog post.
“These chips are part of our growing investment in our AI infrastructure and will enable us to deliver new and better experiences across our apps and technologies,” the company said in the post.
Last year, the company launched its Meta Training and Inference Accelerator (MTIA), its first-generation AI inference accelerator, designed in-house with Meta’s AI workloads in mind.
“The next generation of MTIA is part of our broader full-stack development program for custom, domain-specific silicon that addresses our unique workloads and systems,” the release said.
The new version, Meta added, “more than doubles the compute and memory bandwidth of our previous solution while maintaining our close tie-in to our workloads. It is designed to efficiently serve the ranking and recommendation models that provide high-quality recommendations to users.”
In an interview last month with PYMNTS, John Licato, an assistant professor of computer science and engineering at the University of South Florida, noted that Meta is among the contenders poised to continually challenge OpenAI’s position in the AI space.
“At this point, perhaps the most significant factor is access to a tremendous amount of computing power,” he said. “Companies like Google and OpenAI have millions (perhaps billions) of dollars of GPU processors, as well as more advanced computing technologies like [tensor processing units (TPUs)].”
In other related AI news, Meta last week announced it modified its approach to handling media that has been manipulated with AI and other means.
Three of the company’s platforms — Facebook, Instagram and Threads — will now label a broader range of content as “Made with AI” when they detect industry standard AI image indicators or when the people uploading content disclosed that it was generated with AI, Monika Bickert, vice president of content policy at Meta, wrote in a company blog post.
Meta’s previous policies had only covered videos that used AI to make it appear that a person said something they didn’t say.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.