DAVOS (January 21): Metaplatforms’ new artificial intelligence lab delivered its first high-profile AI model internally this month, the company’s chief technology officer announced on Wednesday.
At a press conference on the sidelines of the World Economic Forum’s annual meeting in Davos, CTO Andrew Bosworth said the model built by the Meta Superintelligence Lab team, formed last year, shows great promise.
“It’s basically six months of work, and it’s not completely even,” Bosworth said, adding that the team’s AI model was “very good.”
Media outlets reported in December that Meta was working on a text AI model code-named “Avocado” scheduled to be released in the first quarter, along with an image- and video-focused model code-named “Mango.” Bosworth, known as “Boz,” declined to say which models were delivered in-house.
Mehta’s efforts are being closely watched after CEO Mark Zuckerberg’s big move to shake up AI leadership, founding new labs and poaching talent with sky-high offers in hopes of helping the company win on the competitive technology frontier.
The company faced criticism over the performance of its Rama 4 model as rivals such as Alphabet Inc.’s Google gain traction in the game of innovative and lucrative AI.
Bosworth said the technology is not yet mature.
Speaking about the development cycle in general, Bosworth said, “There’s a tremendous amount of work that needs to be done after the AI is trained to actually deliver the model in a way that can be used internally or by consumers.”
But he said the meta’s big bets from 2025 are starting to show positive results.
Uploaded by Magessan Varatharaja
