Yann LeCun's new venture is a contrarian bet against large language models (www.technologyreview.com)

Yann LeCun, a prominent AI researcher and Turing Award winner, has launched a new company, Advanced Machine Intelligence (AMI), positioning it as a counter to the prevailing focus on large language models (LLMs). Instead, LeCun advocates for the development of "world models," which aim to create AI systems that better understand and predict real-world dynamics. He believes that the current trend toward proprietary models, particularly in the US, is a strategic misstep and emphasizes the need for open-source AI. By establishing AMI in Paris, LeCun aims to cultivate European talent and address the sovereignty concerns many countries have regarding AI. The significance of LeCun's venture lies in its potential to shift the narrative away from LLMs toward AI that can reason, predict, and learn about the world through diverse data modalities, including video and sensor inputs. With a focus on the JEPA architecture, which learns abstract representations of the environment rather than just generating text, AMI seeks to create intelligent systems capable of real-world applications, such as predicting behaviors in complex industrial settings or enhancing domestic robotics. LeCun's initiative reflects a growing call within the AI community for foundational research that transcends the limitations of current LLM technologies, emphasizing that true AI advancement will require a broader understanding of physical phenomena.