Former Meta chief AI scientist Yann LeCun has launched AMI Labs to develop advanced artificial intelligence capable of understanding physical reality, moving beyond the limitations of current language models.

Key Points

• Yann LeCun founded Paris-based AMI Labs after leaving Meta in 2025 to pursue AI that can navigate real-world physical environments.
• AMI Labs secured over $1 billion in seed funding from investors including Nvidia and Jeff Bezos’s private wealth fund.
• The company is developing Joint Embedding Predictive Architecture (JEPA), a system designed to create world abstractions rather than relying on statistical text prediction.
• LeCun argues that Large Language Models (LLMs) like ChatGPT lack true reasoning and are unsuitable for complex robotics tasks like household chores.
• Researchers are increasingly focusing on "World Models," such as Google’s Dreamer, which allow AI to learn through mental simulations of future scenarios.

Why it Matters

Current AI models are limited by their reliance on statistical patterns, which prevents them from performing reliable tasks in unpredictable physical environments. Developing systems that can reason about the real world is essential for the future of robotics and could shift the industry away from text-based generation toward autonomous, functional intelligence.