Four New Groups to Streamline AI Development
Meta Platforms has launched a major restructuring of its artificial intelligence operations by splitting its “Meta Superintelligence Labs” into four distinct divisions. The change comes as the company seeks to streamline operations and speed up the development of competitive AI products, amid growing friction within its AI ranks.
Strategic Shifts and Abandoned Models
The reorganization will see teams dedicated separately to research, infrastructure, superintelligence, and product development. Meta is also reconsidering its commitment to open-source AI, with internal debates over shifting to closed-source models. Sources report that the company has abandoned its earlier frontier model “Behemoth” due to disappointing performance, opting to build a new model from scratch.
Leadership Shakeups and Talent Departure
Mark Zuckerberg’s appointment of Alexandr Wang to lead the superintelligence team has caused rifts with veteran AI staff. The hiring of Shengjia Zhao, co-creator of ChatGPT, also stirred friction, as he began re-evaluating existing teams. Meanwhile, key personnel have exited: Joelle Pineau joined Cohere, Angela Fan moved to OpenAI, and Loredana Crisan is reportedly leaving soon.
Collaborations and Outside Models
Meta is exploring partnerships beyond its own AI models, considering integration of open-source and licensed closed-source systems. Nat Friedman and Daniel Gross are now leading the applied research division. The company aims to leverage external technologies while consolidating leadership to regain momentum in a rapidly evolving AI landscape.
