ex-Meta Chielf AI scientist Yann LeCun just raised $1bn to build Large World Models

Yann LeCun, one of the founding fathers of modern AI and former Chief AI Scientist at Meta, has launched a monumental new venture. The project has reportedly secured a staggering $1 billion in initial funding from a consortium of investors, marking one of the largest single raises in AI history. LeCun's goal is to pioneer "Large World Models" (LWMs), a new class of AI fundamentally different from the Large Language Models (LLMs) that power ChatGPT and Claude.

While LLMs are trained primarily on text and code, LWMs are designed to learn from video and multimodal sensory data to build an internal model of how the physical world works. This approach, championed by LeCun for years, aims to overcome the lack of common-sense reasoning and true understanding in current AI. The project represents a direct challenge to the scaling-centric approach of OpenAI and others, betting that a new architecture is needed for artificial general intelligence (AGI).

The substantial funding will be used to assemble a top-tier research team and acquire the immense computational resources required for training world models on unprecedented scales of video data. If successful, LWMs could power a new generation of robots, autonomous systems, and AI assistants that intuitively understand physics, cause-and-effect, and the nuances of real-world environments.