Meta's AI Reorganization: The Race to Superintelligence

Meta is reorganizing its AI teams to focus on key areas: research, product, training, and infrastructure.

Alexandr Wang, the leader of Meta Superintelligence Labs (MSL), emphasized the importance of taking 'superintelligence' seriously.

The reorganization involves establishing four distinct teams: TBD Lab (training and scaling models), FAIR (fundamental research), Products & Applied Research (product-focused models), and MSL Infra (infrastructure).

The AGI Foundations team is being dissolved, with members distributed across the new teams.

This reorganization is Meta's latest effort to gain a competitive edge in the AI arms race, amidst internal tensions and talent acquisition from rival AI labs.

TBD Lab:: Focused on training and scaling large AI models to achieve superintelligence, exploring new directions such as an 'omni' model.

FAIR (Facebook AI Research):: Serving as an 'innovation engine' for MSL, integrating research ideas into larger model runs conducted by TBD Lab.

Products & Applied Research:: Bringing product-focused research efforts closer to product development, including teams working on Assistant, Voice, Media, Trust, Embodiment, and Developer pillars.

MSL Infra:: Unifying infrastructure teams to accelerate AI research and production through advanced infrastructure, optimized GPU clusters, and comprehensive developer tools.

Q: What is Meta Superintelligence Labs (MSL)?

Q: Why is Meta reorganizing its AI teams?

Q: What is the 'omni' model?

Meta is making significant investments and changes in its AI division to pursue artificial superintelligence.

The reorganization reflects the growing importance of AI and its potential impact on technology and society.

Meta's efforts to streamline research, product development, and infrastructure aim to accelerate its progress in the AI arms race.

The changes may affect how Meta develops and integrates AI into its products and services.