Meta is again restructuring its artificial intelligence division, breaking the unit into four distinct teams. The reorganisation, announced internally this week, is intended to increase focus on the battle for cutting-edge technology. However, the restructure has also raised concerns about layoffs. This is a dramatic reversal after the frenzied hiring binge of CEO Mark Zuckerberg over the last year.
These shifts were detailed in a memo from newly hired Meta chief AI officer and former Scale AI CEO Alexandr Wang. Wang explained that the company required ‘sharper focus’ to get ready for the future era of superintelligence. The newly titled Meta Superintelligence Labs (MSL) will now be structured through four units:
One for large language models like Llama.
FAIR, Meta’s veteran AI research division.
A consumer goods unit, headed by former GitHub CEO Nat Friedman.
An infrastructure unit headed by Aparna Ramani, which manages computing hardware and data centres.
The restructuring follows months of high levels of spending and recruitment. Meta has been poaching star scientists from competitors such as OpenAI and Google, with packages worth hundreds of millions. Zuckerberg has vowed to spend ‘hundreds of billions’ on talent and infrastructure to seek ‘superintelligence’, AI that can beat humans at complex tasks.
Even so, The New York Times noted that Meta is contemplating laying off or reassigning staff in its AI division, which is now several thousand strong. Top leaders could also leave. Although nothing has been decided, the potential for layoffs has undermined morale, particularly as artificial intelligence was touted as the firm’s most significant growth driver.
Also Read: Meta & Character.AI Accused of Misleading Kids With Deceptive AI Marketing
Meta disbanded its AGI Foundations team earlier this year, where generative AI work had been conducted. Leaders were being reshuffled across teams, Connor Hayes transferred to Threads, and Loredana Crisan departed for Figma. The repeated reshuffles indicate that an organisation has not found firm ground in its efforts.
The recurring churn is a sign of urgency and turbulence. While their competitors OpenAI and Google have already developed consumer-facing models, Meta is still perfecting its systems.
The New York Times reports that the company is even considering using third-party platforms, open-source, and licensed models rather than sticking to in-house technology alone.
For the present, the formal position is that dividing MSL into four groups will add clarity. However, with possible layoffs, leadership turnover, and strategy changes, Meta’s journey to superintelligence is far from clear-cut.