Yann LeCun, a leading figure in artificial intelligence research often dubbed the “godfather of AI,” has announced his departure from Meta at the end of the year to launch a new machine learning start-up. The move signals a strategic shift toward a different approach to AI development, one that diverges from the current focus on large language models (LLMs).
Dissatisfaction with Current AI Trends
LeCun, Meta’s chief AI scientist for the past seven years, has become increasingly critical of LLMs like ChatGPT and Meta’s own Llama. While acknowledging their utility, he believes they represent a dead end in the pursuit of true artificial intelligence. His concern centers on the fact that LLMs excel at predicting language but lack the fundamental understanding of the physical world necessary for genuine intelligence.
“LLMs are great, they’re useful…They are not a path to human-level intelligence. They’re just not.”
This stance has implications for Meta’s massive investments in generative AI, suggesting a potential re-evaluation of resources and priorities within the company. LeCun argues that the overwhelming focus on LLMs is diverting attention and funding from more promising avenues of research.
The Rise of “World Models”
LeCun’s new venture will prioritize “world models,” AI systems that learn by observing and interacting with the physical world, similar to how humans develop intuition. Unlike LLMs that predict the next word, world models predict events, understand cause-and-effect, and build a deeper, more robust form of intelligence.
The concept is rooted in how humans learn: not by memorizing sequences but by forming internal representations of reality that allow for flexible reasoning and problem-solving. The new start-up, dubbed AMI (French for “friend”), will continue research previously conducted at Meta’s FAIR lab and New York University.
A Strategic Independence
LeCun emphasized the need for an independent entity to maximize AMI’s impact. While he acknowledged a partnership with Meta, the specifics remain undisclosed. The company’s goals extend beyond Meta’s immediate commercial interests, suggesting a broader ambition to develop AI that can revolutionize various sectors of the economy.
This move underscores a growing debate within the AI community: whether scaling existing LLMs is the most effective path forward, or if a fundamental shift in approach is required. LeCun’s departure and the launch of AMI signal a clear bet on the latter.
In essence, LeCun is moving beyond the hype surrounding LLMs to pursue AI that truly understands and interacts with the world—a direction that could redefine the future of intelligence.
