Meta FAIR Lab

Fundamental AI Research laboratory at meta, led by yann-lecun. The lab focuses on developing artificial-general-intelligence (AGI) through world-models and Self-Supervised Learning.

Core Research Paradigms

Recent Developments

  • vl-jepa (2026-04-14): A new vision-based approach to AGI that departs from the current llm-centric paradigm.
    • Core Thesis: “Language is not Intelligence.”
    • Represents a strategic shift away from generative-ai-driven approaches.
    • Emphasizes vision-based predictive modeling as a foundation for true intelligence, rather than relying on text-only generative processes.
  • 2026 04 14 New paper for a vision approach to AGI not LLM

Source Notes